JPH1066084A - Video data compressing device and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to compress video data continuously including plural scenes and to improve the quality of a decoded image by controlling the compression ratio of non-compressed video data delayed for a prescribed time based on the difficulty of the non-compressed video data calculated from the difficulty of data obtained by preparatively executing the compression encoding of video data. SOLUTION: A host computer 20 receives the data quantity of compressed video data generated by preparatively executing the compression encoding of non-compressed video data by an encoder 162 in a simple two-pass processing part and the value of a DC component and the power value of an AC component of video data obtained after discrete cosine transformation processing through a control signal C16 and calculates the diffiiculty of a compressed video data pattern based on these received values. Then the host compuer 20 allocates the objective data of compressed video data generated by an encoder 18 to each picture through a control signal C18 based on the calculated difficulty, sets up the allocated result in a quantization circuit built in the encoder 18 and adaptively controls the compression ratio of the encoder 18 in each picture.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video data compression apparatus for compressing and encoding non-compressed video data and a method thereof.

[0002]

2. Description of the Related Art Uncompressed digital video data is stored in a moving picture (MPEG) format.
I-picture (intra c
oded picture), B picture (bi-directionaly coded
picture) and P-picture (predictive coded pict
ure), compression encoded in GOP (group of pictures) units, and a magneto-optical disk (MO disk;
When recording on a recording medium such as an o-oprical disc, the data amount (bit amount) of the compressed video data after compression encoding is
It is necessary to keep the quality of the video after decompression decoding high, while keeping it below the recording capacity of the recording medium or below the transmission capacity of the communication line.

For this purpose, first, non-compressed video data is preliminarily compression-encoded and the data amount after compression-encoding is estimated (first pass). Next, the compression rate is adjusted based on the estimated data amount. Then, compression encoding is performed so that the data amount after the compression encoding becomes equal to or less than the recording capacity of the recording medium (second pass).
(Hereinafter, such a compression encoding method is also referred to as “two-pass encoding”).

However, if compression encoding is performed by two-pass encoding, it is necessary to perform the same compression encoding process twice on the same non-compressed video data, which takes time. In addition, since the final compressed video data cannot be generated by one compression encoding process, the captured video data cannot be directly compression-encoded and recorded in real time (real time).

[0005] In addition, a plurality of uncompressed video data (hereinafter, also referred to as scenes) that are not correlated in the time direction due to editing processing.
Are continuously connected to form one uncompressed video data (edited video data), and this edited video data is, for example, a picture type sequence
When compression encoding is performed using B, P, B, P, and B, the first picture after compression encoding may be a P picture.
In order to decompress and decode the first P picture, it is necessary to refer to the picture immediately before the compressed video data generated from another scene. However, when a picture generated from another scene having no correlation is used for the expansion decoding of the first P picture, a huge amount of data is required because a motion prediction error is significantly increased, and only a limited data amount can be used. In such a case, the video after decompression decoding is deteriorated.

In order to solve such a problem, for example,
JP-A-7-193818 discloses an image processing method and an image processing apparatus. JP-A-7-193818
The image processing method and the image processing apparatus disclosed in Japanese Patent Application Laid-Open No. H10-15095 convert uncompressed edited video data including, for example, two scenes (a first scene and a second scene) into, for example, the picture type sequences I, B, P, B, P, B,
When compression encoding is performed using P, B, P, B, P, and B, second compressed video data obtained by compressing and encoding the second scene (I ₂ , B ₂ , P in the picture type sequence shown below)
₂ ) the first P picture is an I picture which does not refer to the last picture of the first compressed video data (I ₁ , B ₁ , P ₁ in the picture type sequence shown below) obtained by compression encoding the first scene. In order to suppress an increase in the amount of generated data, compression encoding is performed by changing the last I picture of the first compressed video data to a P picture.

That is, specifically, Japanese Patent Laid-Open No. 7-19381
The image processing method and image processing apparatus disclosed in 8 JP compresses encoded without changing the picture type sequence, the first compressed image data and second compressed image data, picture type sequence B ₁ , I ₁ ,
B ₁ , P ₁ , B ₁ , P ₁ , B ₁ , P ₂ , B ₂ , P ₂ , B
₂ , P ₂ , B ₂ , the last I picture of the first compressed video data is changed to a P picture, and the first P picture of the second compressed video data is changed to an I picture. To compress and encode the picture type sequences B ₁ , P ₁ , B ₁ , P ₁ , B ₁ , P ₁ , B ₁ ,
The first and second compressed video data of I ₂ , B ₂ , P ₂ , B ₂ , P ₂ , and B ₂ are obtained.

The present invention has been made by improving the above-mentioned prior art, and compresses and encodes video data continuously including a plurality of scenes to a predetermined data amount or less without performing two-pass encoding. Video data can be generated, and the quality of a video obtained by decompressing and decoding compressed video data obtained by compressing and encoding boundaries (scene changes) in the time direction between a plurality of continuous scenes can be maintained. It is an object of the present invention to provide a video data compression apparatus and a method therefor.

[0009]

In order to achieve the above object, a video data compression apparatus according to the present invention is characterized in that the head of a plurality of non-compressed video data which are continuously input is an I picture , A P picture, and a B picture, after being compressed into a predetermined picture type sequence, the picture sequence is changed to an I picture or a P picture, and the non-compressed video is reordered. First compression means for compressing data by the predetermined compression method to generate first compressed video data, and delay means for delaying the non-compressed video data whose order has been rearranged by a predetermined delay time; , Based on the data amount of the first compressed video data generated from the uncompressed video data corresponding to the predetermined delay time. Prediction means for predicting the data amount of the predetermined amount of the uncompressed first compressed video data, the predicted data amount of the first compressed video data, and the actually generated first compressed video data Head detecting means for detecting the head of the non-compressed video data based on the data amount (actual data amount) of the non-compressed video data. Changing means for changing the predetermined picture type sequence so as not to have a relationship with a picture; the generated first compressed video data; and the predicted first compressed video data.
Target value generating means for generating a target value of the data amount of the uncompressed video data after compression based on the data amount of the compressed video data, so that the data amount after compression becomes the generated target value. And second compression means for compressing the delayed uncompressed video data into the changed predetermined picture type sequence by the predetermined compression method.

[0010] Preferably, the head detecting means is arranged so that the actual data amount of the I picture and the P picture is such that the predicted ratio of the first compressed video data to the I picture and the P picture is out of a predetermined range. , The head of the uncompressed video data is detected at a position corresponding to the P picture whose data amount has increased.

Preferably, when the actual data amount of the B picture is larger than the predicted data amount of the B picture of the first compressed video data by a predetermined ratio or more, The head of the uncompressed video data is detected at the position of the I picture immediately before the B picture whose data amount has increased.

Preferably, in the predetermined picture type sequence, when the head of the uncompressed video data is compressed into a P picture, the changing unit compresses the head of the uncompressed video data into an I picture. Thus, the predetermined picture type sequence is changed.

Preferably, the changing means changes the predetermined picture type sequence so that the head of the uncompressed video data is compressed into an I picture.
A picture of the uncompressed video data, which becomes an I picture after neighboring compression, is compressed into a P picture,
The predetermined picture type sequence is further changed.

In the video data compression apparatus according to the present invention,
For example, if uncompressed video data is
Kens I, B, B, P, B, B, ..., P, B, B (above
Uncompressed video data compressed to a picture type sequence
Each picture of the data₁, B_Two,
B_Three, P_Four, B_Five, B₆, ..., P₁₃, B₁₄, B_FifteenNotation
), The replacement means is input continuously.
I of multiple scenes (uncompressed video data)
₁, B_Two, B_Three, P_Four, B_Five, B₆, P₇, ..., P ₁₃,
B₁₄, B_FifteenIn the order suitable for compression encoding, picture I
₁, B_-2, B_-1, P _Four, B₁, B_Two, ..., P₁₃, B₁₁,
B₁₂Replace with In other words, uncompressed video data
For example, a set of I-picture and P-picture
B picture is replaced with the immediately following I picture or P picture
Move it behind.

The first compression means preliminarily compression-encodes a plurality of scenes in which the order of the pictures has been changed by the replacement means, and obtains difficulty data necessary for determining the amount of data to be allocated to each of the compressed pictures. Of the first compressed video data used for. More specifically, the first compression unit converts each scene into a picture type sequence I, B, B, P, B,
GOP (group of pi) composed of B, ..., P, B, B
compression encoding) to generate first compressed video data. Since the order of the pictures of the scene is changed as described above, the first picture of the scene immediately after the scene change (the boundary in the time direction between a plurality of scenes) is an I picture or a P picture.

The delay means is sufficient for calculating, for example, the time during which a predetermined number of pictures of each scene are input, that is, the amount of data allocated to each of the pictures of the compressed video data obtained by compressing each scene. Each scene to be input is delayed by a time sufficient to obtain the first compressed video data required to generate a large amount of difficulty data. The prediction means approximates, for example, a straight line to the data amount of the first compressed video data generated by the first compression means, and further converts a straight line obtained by the approximation into an ungenerated part of the first compressed video data. The extrapolated and ungenerated first compressed video data data amount for each picture is predicted for each picture type.

The head detecting means compares the data amount of the predicted picture with the data amount of the actually generated picture when the predicted ungenerated portion of the first compressed video data is actually generated later. Then, the head part (scene change part) of the scene is detected. Specifically, the head detecting means compares, for example, the data amount of the predicted I-picture and P-picture with the data amount of the actually generated I-picture and P-picture, and calculates the ratio of the actual data amount to the predicted value. Is out of the predetermined range, it is detected that a scene change has occurred in portions corresponding to these I-pictures and P-pictures. Also, specifically, the head detecting means uses the fact that the data amount of the B picture after the scene change increases as much as the P picture, and for example, the data amount of the predicted B picture and the actually generated B picture The data amount of the picture is compared with that of the picture, and if the actual data amount is larger than the predicted value by a predetermined ratio or more, a scene change has occurred in a portion corresponding to the I picture and the P picture immediately before the B picture. Detect that. As described above, by monitoring not only the data amount of the I picture and the P picture but also the data amount of the B picture, the head detecting means can surely perform a scene change portion.

[0018] The changing means is a P picture in which the first picture of the scene detected by the first detecting means has a relationship with the previous picture (the last picture of the previous scene) (refers to the data of the previous picture at the time of decompression). , The picture type sequence is changed so that the first picture of the scene is compressed into an I picture having no relation to other pictures. Further, the target value generating means may generate the compressed video finally generated based on the data amount of the generated first compressed video data and / or the predicted data amount of the first compressed video data. A target value of the data amount of the data (second compressed video data) is generated.

[0019] The second compressing means uses the same MPEG system as the first compressing means, for example, so that the data amount of each picture after compression becomes equal to the data amount indicated by the corresponding target value. Each of the scenes thus compressed is compressed into a picture type sequence changed by the changing means, and second compressed video data of each scene is generated.

Further, in the video data compression method according to the present invention, the head of a plurality of non-compressed video data which are continuously input is composed of a combination of I picture, P picture and B picture by a predetermined compression method. After being compressed to a predetermined picture type sequence, the data is compressed to become an I picture or a P picture to generate first compressed video data, and the first compressed video data of the first compressed video data corresponding to the predetermined delay time is generated. A data amount of a predetermined amount of the uncompressed first compressed video data is predicted based on the data amount, and a predicted data amount of the first compressed video data and the actually generated first compressed video data are calculated. The first picture of the uncompressed video data is detected based on the data amount (actual data amount).

Preferably, the uncompressed video data is delayed by a predetermined delay time, and the detected P picture is
After the compression, the predetermined picture type sequence is changed so as to become an I picture, and the data after compression is determined based on the data amount of the generated first compressed video data and the predicted first compressed video data. An amount target value is generated, and the uncompressed video data delayed by the predetermined compression method is changed to the predetermined picture type sequence so that the data amount after compression becomes the generated target value. Compress.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First Embodiment Hereinafter, a first embodiment of the present invention will be described. By compression encoding video data such as the MPEG method, when compression encoding video data with a high degree of difficulty, such as a pattern with many high frequency components or a pattern with a lot of motion, distortion due to compression is likely to occur. Become. For this reason, it is necessary to compress and encode video data having a high degree of difficulty at a low compression ratio. For compressed video data obtained by compressing and encoding data having a high degree of difficulty, a compressed image of video data having a pattern having a low level of difficulty is obtained. It is necessary to allocate a larger amount of target data than data.

As described above, the two-pass encoding method shown as a conventional technique is effective for adaptively allocating a target data amount to the difficulty of video data. However, the two-pass encoding method is not suitable for real-time compression encoding. The simplified two-pass encoding method shown as the first embodiment has been made to solve the problem of the two-pass encoding method, and the compressed video data obtained by preliminary compression-encoding the non-compressed video data The difficulty level of the uncompressed video data is calculated from the difficulty level data, and the FI level is calculated based on the difficulty level calculated by the preliminary compression encoding.
The compression rate of the uncompressed video data delayed by a predetermined time by the FO memory or the like can be adaptively controlled.

FIG. 1 is a diagram showing a configuration of a video data compression device 1 according to the present invention. As shown in FIG. 1, the video data compression apparatus 1 includes a compression encoding unit 10 and a host computer 20. The compression encoding unit 10 includes an encoder control unit 12, a motion estimator 14,
Simple 2-pass processing unit 16, second encoder (encoder) 1
8, and the simple two-pass processing unit 16 includes a FIFO memory 160 and a first encoder 162. The video data compression device 1 uses these components to generate uncompressed video data VI input from external devices (not shown) such as an editing device and a video tape recorder device.
For N, the above-described simple two-pass encoding is realized.

In the video data compression apparatus 1, a host computer 20 controls the operation of each component of the video data compression apparatus 1. Also, the host computer 20
The data amount of the compressed video data generated by the encoder 162 of the simple two-pass processing unit 16 preliminarily compression-encoding the non-compressed video data VIN, the value of the DC component (DC component) of the DCT-processed video data, and the DC component The power value of the (AC component) is received via the control signal C16, and the difficulty of the picture of the compressed video data is calculated based on the received values. Further, the host computer 20 based on the calculated difficulty, assigned to each picture of the target amount of data T _j of the compressed video data encoder 18 is generated via a control signal C18, the quantization circuit 166 of the encoder 18 (FIG. 3 ), And the compression rate of the encoder 18 is adaptively controlled on a picture basis.

The encoder controller 12 determines whether or not there is a picture of the uncompressed video data VIN by the host computer 20.
And performs a pre-process for compression encoding for each picture of the uncompressed video data VIN. That is, the encoder control unit 12 rearranges the input non-compressed video data in the order of encoding, performs picture / field conversion, and performs 3: 2 pull-down processing (movie processing) when the non-compressed video data VIN is video data of a movie. Of the 24 frames / sec video data into 30 frames / sec video data, and removes the redundancy before the compression encoding.
The FIF of the simple 2-pass processing unit 16 is used as the video data S12.
Output to the O memory 160 and the encoder 162. The motion detector 14 detects a motion vector of the uncompressed video data, and outputs the motion vector to the encoder control unit 12 and the encoders 162 and 18.

In the simple two-pass processing unit 16, the FIFO
The memory 160 converts the video data S12 input from the encoder control unit 12 into, for example, uncompressed video data VIN
Is delayed by the time of L (L is an integer) picture input, and is output to the encoder 18 as delayed video data S16.

FIG. 2 shows a simplified two-pass processing unit 1 shown in FIG.
6 is a diagram illustrating a configuration of a sixth encoder 162. FIG. The encoder 162 includes, for example, as shown in FIG.
4. DCT circuit 166, quantization circuit (Q) 168, variable length coding circuit (VLC) 170, inverse quantization circuit (IQ)
172, inverse DCT (IDCT) circuit 174, addition circuit 1
Is a general video data compression encoder composed of the video data S12 and the motion compensation circuit 178. The input video data S12 is compression-coded by the MPEG method or the like, and the amount of compressed video data for each picture is determined. Output to the host computer 20.

The addition circuit 164 subtracts the output data of the addition circuit 176 from the video data S12,
6 is output. The DCT circuit 166 includes the addition circuit 1
For example, the video data input from 64 is converted to 16 pixels ×
Discrete cosine transform (D
CT), converts the data in the time domain into the data in the frequency domain, and outputs the data to the quantization circuit 168. Further, the DCT circuit 166 controls the DCT of the video data after the DCT.
The value of the component and the power value of the AC component are output to the host computer 20. The quantization circuit 168 quantizes the frequency domain data input from the DCT circuit 166 with a fixed quantization value Q, and outputs the quantized data to the variable length coding circuit 170 and the inverse quantization circuit 172. . The variable length coding circuit 170 performs variable length coding on the quantized data input from the quantization circuit 168, and calculates the data amount of the compressed video data obtained as a result of the variable length coding.
Output to the host computer 20 via the control signal C16. The inverse quantization circuit 172 inversely quantizes the quantized data input from the variable length encoding circuit 168, and outputs the inversely quantized data to the inverse DCT circuit 174.

The inverse DCT circuit 174 includes the inverse quantization circuit 17
Inverse DCT processing is performed on the inversely quantized data input from 2 and output to the adding circuit 176. Addition circuit 1
76 is the output data of the motion compensation circuit 178 and the inverse DC
The output data of the T circuit 174 is added and output to the addition circuit 164 and the motion compensation circuit 178. The motion compensation circuit 178 performs a motion compensation process on the output data of the addition circuit 176 based on the motion vector input from the motion detector 14, and outputs the result to the addition circuit 176.

FIG. 3 is a diagram showing a configuration of the encoder 18 shown in FIG. As shown in FIG.
Has a configuration in which a quantization control circuit 180 is added to the encoder 162 shown in FIG. The encoder 18
With these components, based on the target amount of data T _j set from the host computer 20, the motion compensation process to the L picture delayed by the delayed video data S16 by the FIFO memory 160, DCT processing, quantization processing and The variable-length encoding processing is performed to generate compressed video data VOUT in the MPEG format or the like, and output the same to an external device (not shown).

In the encoder 18, the quantization control circuit 180 sequentially monitors the data amount of the compressed video data VOUT output from the variable length quantization circuit 170, and finally starts from the j-th picture of the delayed video data S 16. data amount of the compressed video data generated is, so as to approach the target amount of data T _j set from the host computer 20,
The quantization value Q _j to be set in the quantization circuit 168 is sequentially adjusted. The variable-length quantization circuit 170 outputs the compressed video data VOUT to the outside, and also outputs the delayed video data S1
6 is output to the host computer 20 via the control signal C18, the actual data amount _Sj of the compressed video data VOUT obtained by compression-encoding 6.

Hereinafter, a simple two-pass encoding operation of the video data compression device 1 according to the first embodiment will be described. FIGS. 4A to 4C are diagrams illustrating the operation of the simple two-pass encoding of the video data compression device 1 according to the first embodiment. The encoder control unit 12 controls the video data compression device 1
4A, the encoder control unit 12 performs preprocessing such as rearranging the pictures in the encoding order, and as shown in FIG. 4A, the FIFO memory 160 and the encoder 162. Note that the encoder control unit 12
, The order of picture encoding shown in FIG. 4 and the like differs from the order of display after decompression decoding.

The FIFO memory 160 delays each picture of the input video data S12 by L pictures and outputs it to the encoder 18. Encoder 16
Reference numeral 2 denotes a data amount of compression-encoded data obtained by compression-encoding a picture of the input video data S12 in a preliminary and sequential manner, and compression-encoding a j-th (j is an integer) picture; And outputs the DC component value and AC component power value of the video data to the host computer 20.

For example, the delay input to the encoder 18
The video data S16 is stored in the L memory by the FIFO memory 160.
As shown in FIG. 4 (B),
As described above, the encoder 18 determines the j-th
(J is an integer) picture (picture of FIG. 4B)
-A), the encoder 162
Are L-pins from the j-th picture of the video data S12.
The (j + L) -th picture ahead of the kuture (FIG. 4
This means that picture b) of (B) is compression-encoded.
You. Therefore, the encoder 18 transmits the delayed video data S16.
When starting the compression encoding of the j-th picture,
The encoder 162 is configured to j-th to
(J + L-1) -th picture (range of FIG. 4B)
c) the compression encoding has been completed and these pictures
Difficulty data D after compression encoding_j, D _{j + 1}, D_{j + 2},
…, D_{j + L-1}Is already calculated by the host computer 20.
Has been issued.

The host computer 20 uses the following formula 1
As a result, the encoder 18 sets the j-th
A target data amount T _j to be allocated to the compressed video data obtained by compression-coding the third picture is calculated, and the calculated target data amount T _j is set in the quantization control circuit 180.

[0037]

(Equation 1)

Where D _j is the video data S
12 is the actual difficulty data of the 12 th picture,
R ′ _j is the average of the target data amount that can be allocated to the j-th to (j + L−1) -th pictures of the video data S12 and S16, and the initial value of R ′ _j (R ′
₁ ) is a target data amount that can be allocated to each picture of the compressed video data on average, and is expressed by the following equation (2). Each time the encoder 18 generates one picture of the compressed video data, Updated as shown.

[0039]

(Equation 2)

[0040]

(Equation 3)

Note that the numerical bit rate (Bit rat
e) is based on the transmission capacity of the communication line and the recording capacity of the recording medium.
Data amount per second (bit amount)
The picture rate (Picture rate) is
Number of pictures per second (30 pictures / sec.
(NTSC), 25 sheets / second (PAL))
_{j + L}Is a picture determined according to the picture type
Shows the average amount of data per group. DC of encoder 18
The T circuit 166 is configured to output the delayed video data S16
DCT processing is performed on the j-th picture, and a quantization circuit 168
Output to The quantization circuit 168 is a DCT circuit 1
Frequency domain of the j-th picture input from
Of the target data amount T by the quantization control circuit 180._j
Quantized value Q adjusted based on_jQuantized by
Output to the variable length coding circuit 170 as encoded data.
You. The variable length coding circuit 170
Variable quantized data of the input j-th picture
After long encoding, the target data amount T_jData volume close to
And outputs the compressed video data VOUT.

Similarly, as shown in FIG. 4B, when the encoder 18 compresses and encodes the (j + 1) -th picture (the picture a 'in FIG. 4C) of the delayed video data S16. , The encoder 162 completes the compression encoding of the (j + 1) -th to (j + L) -th pictures (range c ′ in FIG. 4C) of the video data S12,
The actual difficulty data D _{j + 1} , D _{j + 2} , D of these pictures
_{j + 3} ,..., D _{j + L} have already been calculated by the host computer 20.

The host computer 20 uses the following equation (1).
The encoder 18 determines the (j + 1) th of the delayed video data S16.
A target data amount T _{j + 1} to be allocated to compressed video data obtained by compression-encoding the third picture is calculated and set in the quantization control circuit 180 of the encoder 18.

The encoder 18 is connected to the host computer 2
From 0, the (j + 1) -th picture is compression-encoded based on the _scale data amount _Tj set in the quantization control circuit 180, and compressed video data VOUT having a data size close to the target data size _{Tj + 1} is generated. And output. In the same manner, the video data compression device 1 similarly converts the k-th picture of the delayed video data S16 into a quantized value Q _k (k = j + 2,
j + 3,...) are changed for each picture, and are sequentially compression-encoded and output as compressed video data VOUT.

As described above, according to the video data compression apparatus 1 shown in the first embodiment, the degree of difficulty of the pattern of the non-compressed video data VIN is calculated in a short time, and the compression ratio according to the calculated degree of difficulty is calculated. Thus, the non-compressed video data VIN can be adaptively compression-encoded. That is, according to the video data compression apparatus 1 shown in the first embodiment, unlike the two-pass encoding method, the non-compressed video data VI
N based on the degree of difficulty of the picture
IN can be compression-encoded, and can be applied to applications requiring real-time performance such as live broadcasting. In addition to the data multiplexing apparatus 1 according to the present invention, the data multiplexing apparatus 1 according to the present invention uses the data amount of the compressed video data compressed and encoded by the encoder 162 as difficulty data as it is, Various configurations can be adopted, such as simplification.

Second Embodiment According to the simple two-pass encoding method shown in the first embodiment, it is possible to perform compression encoding processing on uncompressed video data adaptively in real time according to the difficulty of a picture. . However, when the simple two-pass encoding method shown in the first embodiment is used, when strict real-time performance is required, the delay time of the FIFO memory 160 cannot be increased, and a truly appropriate calculation is difficult for the target data amount T _j, the quality of the image obtained compressed video data VOUT to expansion decoding is likely to decrease.

In the second embodiment, the processing content of the host computer 20 is changed by using the video data compression apparatus 1 (FIG. 1) shown in the first embodiment, and the delay time of the FIFO memory 160 is increased. In order to obtain an appropriate value of the target data amount _Tj without performing the above processing, the j-th picture to the L-th picture of the uncompressed video data and the j-th picture of the compressed video data obtained by preliminary compression encoding are used. (J + L
-1) Actual difficulty data D _{j to} D _{j + L-1 of the first} picture
From the (j + L) -th picture to the (j + L + B) -th picture (B is an integer) of the compressed video data, the difficulty data (prediction difficulty data) D _{j + L to} D _{j + L + B} are calculated. Difficulty data D _{j to} D _{j + L-1} (actual difficulty data) and difficulty data D ′ obtained by prediction
_{Based on j + L to} D ′ _{j + L + B} , the target data amount T is more appropriate than the simple two-pass encoding method shown in the first embodiment.
_The compression encoding method (simple prediction 2
The path encoding method will be described.

First, a simplified predictive two-pass encoding method described in the second embodiment will be conceptually described. Simple prediction 2
In the path encoding method, the picture becomes gradually more difficult, that is, the picture of the non-compressed video data, in which the high frequency components after the DCT processing in the compression encoding gradually increase and the movement becomes faster, becomes more difficult. On the contrary, it is assumed that the pattern of the uncompressed video data, in which the pattern gradually becomes difficult (simplifies), can be predicted to be further simplified.

That is, in the simple predictive two-pass encoding method, the host computer 20 uses the
If the picture is expected to become more difficult,
In preparation for a picture with a more difficult picture, the target data amount to be allocated to the picture currently being compression-encoded is saved, and conversely, if the picture is expected to become simpler, the compression will be performed at that point. The compression rate of the encoder 18 is controlled so as to increase the target data amount allocated to the picture being coded.

Further, a conceptual description of the simple predictive two-pass encoding method will be continued. Video data generally has high correlation in the time direction and the spatial direction, and compression coding of video data is performed by focusing on these correlations and removing redundancy. The high correlation in the time direction means that the difficulty level of the picture of the current uncompressed video data is close to the difficulty level of the picture of the subsequent uncompressed video data. In addition, the tendency of the increase and decrease of the difficulty level up to the present time often continues thereafter.

As a specific example, the non-compressed video data of the case where the camera starts rotating slowly in the horizontal direction from the stationary state, and finally rotates at a constant rotational speed while photographing a stationary object. Think about the design. At first, since the camera is in a stopped state, a still image is captured, and the difficulty of the picture is reduced. Next, assuming that the rotation speed becomes constant after one to two seconds from starting to rotate the camera, the difficulty of the picture tends to increase from one to two seconds after starting to rotate the camera. When this state is viewed from the video data compression device 1 side, several GOPs
During the generation of the compressed video data, the pattern of the input non-compressed video data tends to be more difficult.

Therefore, in the case shown in this specific example, when the difficulty of the pattern of the non-compressed video data shows a tendency to increase, it is predicted that the difficulty of the pattern after that shows a tendency to increase. Reasonable. The simple predictive two-pass encoding method described below positively utilizes the temporal correlation of the difficulty and the increasing / decreasing tendency of the difficulty to apply the first embodiment to each picture of the compressed video data. Simple 2 shown
It is intended to allocate a more appropriate target data amount than in the path encoding method.

The operation of the predictive simple two-pass encoding of the video data compression device 1 according to the second embodiment will be described below. 5A to 5C are diagrams illustrating the operation of the video data compression device 1. As in the first embodiment, the encoder control unit 12 performs pre-processing such as rearranging pictures in the coding order by the encoder control unit 12 on the uncompressed video data VIN input to the video data compression device 1. Is performed, and as shown in FIG.
2 as FIFO memory 160 and encoder 162
Output to

As in the first embodiment, the FIFO memory 160 delays each picture of the input video data S12 by L pictures, and
Output to As in the first embodiment, the encoder 162 preliminary compresses and encodes the picture of the input video data S12 sequentially, and compresses and encodes the j-th (j is an integer) picture. Data amount of compression encoded data, D of video data after DCT processing
The value of the C component and the power value of the AC component are output to the host computer 20. The host computer 20
Based on these values input from the encoder 162, sequentially, to calculate the real difficulty data D _j.

For example, since the delayed video data S16 input to the encoder 18 is delayed by L pictures by the FIFO memory 160, as shown in FIG. j
When the third picture (picture a in FIG. 5B) is compression-encoded, the encoder 162 performs L-pictures from the j-th picture of the video data S12 in the same manner as in the first embodiment. This means that the preceding (j + L) -th picture (picture b in FIG. 5B) has been compression-encoded.

Therefore, when the encoder 18 starts compression-encoding the j-th picture of the delayed video data S16, the encoder 162 sets the (j) -th picture of the video data S12.
-A) -th to (j + L-1) -th pictures (FIG. 5)
(B), where FIG. 5 shows the case where A = 0), completes the compression encoding, the data amount of these pictures after compression encoding, and the D of the video data after DCT processing.
The value of the C component and the power value of the AC component are output to the host computer 20. Host computer 20
_Are based on these values input from the encoder 162, based on the difficulty data (actual difficulty data, range d in FIG. 5B) D _jA , D _{j-A + 1} ,..., D _j , D _{j + 1} , D _{j + 2} ,
.., D _{j + L−1} has already been calculated. Note that A is an integer, and may be either positive or negative.

The host computer 20 stores actual difficulty data D _jA , D _{j-a + 1} ,..., D _j , D _{j + 1} , D _{j + 2 ,.
Based on j + L-1} , the difficulty data after the compression encoding of the (j + L) -th to (j + L + B) -th pictures of the video data S12 (predicted difficulty data, range e in FIG. 5B)
D ′ _{j + L} , D ′ _{j + L + 1} , D ′ _{j + L + 2} ,..., D ′ _{j + L + B,} and the j-th of the delayed video data S16 Target data amount T after compression encoding of the picture
Calculate _j . Therefore, in order to calculate the target amount of data T _j of the compressed encoding of the j-th picture of the delayed video data S16, including the real difficulty data of the predictive difficulty data, the range c shown in FIG. 5 (B) The difficulty data for (A + L + B + 1) pictures is used.

[0058]

(Equation 4)

Note that each symbol in Equation 4 is the same as each symbol in Equation 1.
The same. Encoder 18 is the same as in the first embodiment.
And the quantization control circuit 18 by the host computer 20.
Target data amount T set to 0_jBased on the goal date
Volume T_jProduces compressed video data VOUT with a data amount close to
And output. Further, the host computer 20
As in the operation shown in FIG. 5B, the delayed video data S1
6 (j + 1) -th picture (picture in FIG. 5C)
(Char + '), the (j +
L + 1) th picture (picture in FIG. 5C)
b ') The actual difficulty data D in the range d' in FIG.
_{j-A + 1}, D_{j-A + 2}, ..., D_j, D_{j + 1}, D_{j + 2}, ..., D
_{j + L}, And the prediction difficulty data shown in a range e ′ in FIG.
Data, D '_{j + L + 1}, D '_{j + L + 2}, D '_{j + L + 3}, ..., D '
_{j + L + B + 1}In other words, the actual difficulty shown in the range c 'in FIG.
Delay video data based on the
After compression encoding of the (j + 1) th picture in S16
Target data amount T_{j + 1}Is calculated. The encoder 18
Scale data amount T calculated by the host computer 20_{j + 1}
(J + 1) -th of the delayed video data S16 based on
Is compressed and coded, and the target data amount T_{j + 1}Close to
A large amount of compressed encoded data VOUT is generated. What
Note that the prediction simple 2-pass engine of the above video data compression apparatus 1
The code operation is the (j + 1) th of the delayed video data S16.
The same applies to eye pictures.

Hereinafter, the operation of the video data compression apparatus 1 according to the second embodiment will be summarized and described with reference to FIG. FIG. 6 is a flowchart showing the operation of the video data compression device 1 (FIG. 1) in the second embodiment. FIG.
As shown in step 102, in step 102 (S102),
The host computer 20 converts the numerical values j and R ′ ₁ used in Expression 1 and the like into j = − (L−1), R ′ ₁ = (Bit rate
× (L + B)) / Picture rate

In step 104 (S104), the host computer 20 determines whether or not the numerical value j is larger than 0. If the value j is larger than 0, the process proceeds to S106, and if it is smaller, the process proceeds to S110. In step 106 (S106), the encoder 162
Compresses and encodes the (j + L) -th picture of the video data S12 to generate actual difficulty data D _{j + L.}

In step 108 (S108), the host computer 20 increments the numerical value j (j = j + 1). In step 110 (S110), the host computer 20 transmits the delayed video data S16
It is determined whether the j-th picture exists.
If the j-th picture exists, the process proceeds to S112; otherwise, the compression encoding process ends.

In step 112 (S112), the host computer 20 determines whether or not the numerical value j is larger than the numerical value A. When the numerical value j is larger than the numerical value A, the process proceeds to S114, and when the numerical value j is smaller, the process proceeds to S116. In step 114 (S114), the host computer 20 _executes the actual difficulty data D _{jA to} D _{j + L-1.}
, The predicted difficulty level data D ′ _{j + L to} D ′ _{j + L + B} are calculated. At step 116 (S116), the host computer 20 is the real difficulty data _{D 1 ~D j + L-1} ,
The prediction difficulty data D ′ _{j + L to} D ′ _{j + L + B} are calculated.

In step 118 (S118), the host computer 20 calculates the target data amount T
_j is calculated and set in the quantization control circuit 180 of the encoder 18. Further, the encoder 18 compression-encodes the j-th picture of the delayed video data S16 based on the target data amount T _j set in the quantization control circuit 180, and compresses the compressed picture actually obtained from the j-th picture. The data amount _Sj of the video data is output to the host computer 20. In step 120 (S120), the host computer 20 stores the data amount _Sj from the encoder 18, and further stores the data amount _Sj of the video data S12.
L) Output the actual difficulty data D _{j + L} of the picture.

In step 122 (S122), the encoder 18 outputs the compressed video data VOUT obtained by compression-coding the j-th delayed video data S16 to the outside. In step 124 (S124), the host computer 20 calculates the numerical value F _{j + L} used in Expression 3 according to the picture type. Step 12
6 (S126), the host computer 20
The operation shown in Equation 3 (R ′ _{j + 1} = R ′ _j −S _j + F _{j + L} )
I do.

As described above, according to the prediction simple two-pass encoding by the video data compression apparatus 1 shown in the second embodiment, the difficulty of the pattern of the uncompressed video data VIN is calculated in a short time. The uncompressed video data VIN can be adaptively compression-encoded by further using the degree of difficulty predicted based on the degree of difficulty, and a more appropriate target data amount can be set for each picture of the compressed image data as compared with the simple two-pass encoding method. Can be assigned to Therefore, when the compressed video data is expanded and decoded by the predictive simple two-pass encoding method, a higher quality video can be obtained as compared with the case where the compressed video data is expanded and decoded by the simple two-pass encoding method.

Third Embodiment Hereinafter, as a third embodiment of the present invention, a plurality of uncompressed video data (hereinafter, also referred to as “scene”) are connected by editing processing to form one uncompressed video data. Uncompressed video data (edited video data) is used, and the edited video data composed of the plurality of scenes is compressed by a simple two-pass encoding method using the video data compression device 1 (FIG. 1) shown in the first embodiment. A method for converting the data will be described.

FIGS. 7A to 7C show pictures before and after a scene change by the simple predictive two-pass encoding method according to the second embodiment and the improved simple predictive two-pass encoding method according to the third embodiment. FIG. 4 is a diagram illustrating compression encoding for. The simplified predictive two-pass encoding method shown in the second embodiment utilizes temporal correlation between pictures included in input video data as shown in FIG. Predict the amount of each data. However, if a scene change occurs at the timing shown in FIG. 7B, there is no correlation between pictures before and after the scene change, and therefore, as shown in FIG. It becomes possible to calculate the target amount of data T _j for pictures after the scene change based on the previous difficulty data, not only it is impossible to obtain the effect of the prediction simplified two pass encoding system shown in the second embodiment, rather However, there is a possibility that the quality of the video after the decompression decoding is deteriorated.

That is, to give a specific example, in the predictive simple two-pass encoding method, when a scene change occurs while a scene with a simple pattern is input and the scene is replaced with a scene with a difficult pattern, the host computer 20 Even after the scene change, despite the fact that the value of the difficulty data of the input edited video data is predicted to be small, a picture with a difficult picture is actually input, and the amount of data allocated to each picture in the subsequent scene is insufficient. Would. As described above, when the data amount to be allocated is insufficient, remarkable coding distortion occurs in the compressed video data in the scene change portion, and the quality of the video obtained by decompression decoding is significantly reduced.

The simplified prediction two-pass encoding method (improved simplified prediction two-pass encoding method) shown in the third embodiment is as follows.
From such a viewpoint, when the temporal correlation of the edited video data is lost before and after a scene change, etc., difficulty data generated in a portion where the temporal correlation of the edited video data is lost It is an object of the present invention to eliminate an adverse effect caused by the data amount allocation based on the prediction of the above, further accurately predict the code amount to be allocated to the picture immediately after the scene change, and perform efficient compression encoding.

In order to achieve this object, the improved simplified predictive two-pass encoding method is to improve the simplified predictive two-pass encoding method using the video data compression device 1 (FIG. 1) shown in the second embodiment. Detects scene changes and uses the actual difficulty data obtained after the scene change instead of the actual difficulty data before the scene change, which can no longer be used to calculate the amount of data allocated to pictures of compressed video data. Then, the difficulty of a predetermined number of pictures is predicted.

First, the improved prediction simple two-pass encoding method will be conceptually described with reference to FIGS. FIG.
(A) to (C) show a process of changing the order of pictures of edited video data by the encoder control unit 12 (FIG. 1).
FIG. 9 is a diagram illustrating a process of changing a picture type (picture type) by the host computer 20. FIG. 9 is a diagram exemplifying a change with time of the value of the actual difficulty data near the scene change portion of the edited video data. In FIG. 9, I picture, P picture, and B picture indicate the picture types after the edited video data is compression-encoded.

When a scene change of edited video data occurs in a picture that becomes a P picture after compression encoding (hereinafter, a "picture that becomes a P picture after compression encoding" or the like is also simply referred to as a "P picture"). As shown in FIGS. 8A and 8B, the control unit 12 (FIG. 1) executes the actual difficulty data D generated by the encoder 162 and the host computer 20 from the video data S12 in which the order of the pictures of the edited video data is rearranged. The value of _j changes, for example, as shown in FIG. That is, immediately after the scene change, the real difficulty data D _j of the P picture of the head of the edited video data, P-picture of the compressed video data generated from the picture is increased because it is not possible to refer to the front of the picture, It is generated by the same processing as that of the I picture. Therefore, the value of the actual difficulty data D _j of the P picture at the head of the scene is, for example, about the same as the difficulty data D _j of the I picture.

[0074] Therefore, the host computer 20, based on the picture type sequence of compressed video data encoder 162 produces monitors the temporal change in the value of the real difficulty data D _j, for example, the real difficulty data of the P picture The value of D _j is the actual difficulty data D of the immediately preceding P picture.
_j , 1.5 times or more, the actual difficulty data D _j of the immediately preceding I picture, 0.7 times or more, or in the predictive simple two-pass encoding method shown in the second embodiment. If the value of the actual difficulty data is 1.5 times or more as compared with the value predicted by the host computer 20 in the same manner as above, it is determined that a scene change has occurred in the picture of the edited video data corresponding to the P picture. You can judge.

[0075] However, when a scene change of edited video data is generated in a picture which becomes the I picture after compression coding, the value of the real difficulty data D _j by the host computer 20 to generate is sometimes hardly changes. Conversely, the scene edited when the video data pattern is simple in the like after change, rather, there is a possibility that the value of the real difficulty data D _j is reduced. Also, when the pattern of the edited video data before the scene change is complicated and the pattern of the edited video data after the scene change is flat, or when the edited video data before and after the scene change has a very large movement, the value of the real difficulty data D _j of the P-picture may not increase significantly. However, practically, because immediately after the scene change can not be referred to only a rear picture, the value of the real difficulty data D _j of the B-picture immediately after the scene change, to the same extent as the value of the real difficulty data D _j of the P picture Increase.

[0076] Therefore, the host computer 20 monitors the temporal change in the value of the real difficulty data D _j, for example, B
When the value of the actual difficulty data D _j of the picture is 1.5 times or more the actual difficulty data D _j of the immediately preceding B picture,
Alternatively, if the actual value of the real difficulty data D _j compared with predicted values is equal to or greater than 1.5 times, scene change picture of edited video data corresponding to the I-picture and P-picture immediately before the B-picture Can be determined to have occurred. By using a method of detecting a scene change based on a change in the actual difficulty data D _j of the P picture and a method of detecting a scene change based on a change in the actual difficulty data D _j of the B picture in combination, The computer 20 can reliably detect a scene change.

On the other hand, when a scene change occurs, the correlation between the picture before the scene change of the edited video data and the picture after the scene change is lost. Predicted difficulty data D ′ _j for the picture after the scene change using the previous actual difficulty data D _j has no meaning. However, the number of sheets of pictures immediately after a scene change of the edited video data has a subsequent picture and a sufficient correlation, therefore, based on the real difficulty data D _j of the number of sheets of pictures immediately after a scene change, it it is possible to predict the value of the difficulty data D _j of the picture after a predetermined number.

Further, in the simple predictive two-pass encoding method shown in the second embodiment, the target data amount _Tj is calculated as shown in Expression 4. Therefore, the target data amount T
_In order to calculate _j , it is sufficient to use the sum value Sum _j defined in Expression 5 shown below, and it is not always necessary to calculate individual prediction difficulty data D ′ _j .

[0079]

(Equation 5)

Using the sum Sum _j defined in Equation 5, Equation 4 can be rewritten as Equation 6 shown below.

[0081]

(Equation 6)

[0082] That is, the host computer 20, rather than the individual predictive difficulty data D _'j, if only able to predict the sum value Sum _j, it is possible to calculate the target amount of data T _j.

In the improved prediction simple two-pass encoding method according to the third embodiment, the host computer 20
Predicts the sum value Sum _j based on the actual difficulty data D _j generated immediately after the scene change, and calculates the predicted sum value Su.
Based on m _j , the target data amount T _j is accurately calculated. Subsequently, while a predetermined number of pictures of edited video data are input, the host computer 20 sequentially corrects the value of the sum Sum _j based on the actual difficulty data D _j generated thereafter. Further, the host computer 20
Scene change subsequent further input a predetermined number of picture, after generating the real difficulty data D _j of sufficient number,
The target data amount _Tj is generated by the same method as in the simple predictive two-pass encoding method shown in the second embodiment.

Next, the operation of the video data compression device 1 (FIG. 1) in the third embodiment will be described. For simplicity of description, also in the third embodiment, as shown in FIG. 7, the video data compression device 1 uses the same picture type sequence (N = 15,
M = 3; N is the number of pictures included in one GOP, M is P
The edited video data is compression-encoded to (the number of B pictures between pictures), and is 15 bits as in the second embodiment.
An example will be described in which predicted difficulty data D ′ _j of the next 15 pictures are generated from actual difficulty data D _j of the pictures.

The encoder control unit 12 performs the same processing as in the first and second embodiments, for example, for the non-compressed video data input in the picture type sequence shown in FIG. Change the picture order
As shown in FIG. 8B, an order suitable for compression encoding in the encoder 162 and the encoder 18, that is,
The B picture is rearranged in the order following the immediately following I picture or P picture, and is output to the encoder 162 and the FIFO memory 160 as video data S12. Therefore, for example, as shown in FIG. 8A, even if a scene change between the data of the first scene and the data of the second scene is a picture to be compression-encoded into a B picture, The first picture type of the subsequent scene input to the encoder 162 and the encoder 18 is always a P picture or an I picture. The FIFO memory 160 outputs the input edited video data to the encoder 18 with a delay of 15 pictures, for example, as in the first and second embodiments.

The encoder 162 converts the picture data S12 into the picture type sequence I, B, B, P, B, B, irrespective of the presence or absence of a scene change, as in the first and second embodiments. P, B, B,
P, B, B, P, B, B, P, B, B
And outputs to the host computer 20 to generate the real difficulty data D _j. Temporal change in the value of the real difficulty data D _j of the encoder 162 generates, for example, is as shown in FIG. 9, generally behind just after the scene change occurs scene of the first P-picture The value of the actual difficulty data is larger than the values of the actual difficulty data of the other P pictures.

The host computer 20 includes the encoder 1
Monitoring the temporal change in the value of the real difficulty data input from 62, as described above in the third embodiment, the value of the real difficulty data D _j is the real difficulty data D immediately before the P-picture _{j- 1} , for example 1.5 times (practically 1.4 times ~
It is preferable to set a value between 1.8 times.) It is determined that a scene change has occurred in the P picture by a method such as detecting a P picture showing the above value. When a scene change is detected, the host computer 20 further transmits the information shown in FIG.
As shown in (C), the first P picture of the subsequent scene is changed to an I picture that does not refer to the last picture of the previous scene, and the last I picture of the previous scene is changed to a P picture. , And controls the encoder 18 to change the picture type sequence when the part before and after the scene change of the edited video data is compression-encoded.

It should be noted that even if a scene change occurs, a large change does not always occur in the data amount of the I picture itself. However, as described above in the third embodiment, the host computer 20 monitors a temporal change in the value of the actual difficulty data of the B picture, and for example, 1.5 times the actual difficulty data of the immediately preceding B picture. It is possible to determine that a scene change has occurred in the I picture by a method such as detecting a B picture having the actual difficulty data of the value.

FIG. 10 shows that the host computer 20
Real difficulty when scene change occurs in the collected video data
Degree data D₁~ D_FifteenBased on the prediction difficulty data D '₁₆~
D ' ₃₀And how to calculate the
Prediction difficulty data D 'when no change occurs₁₆~
D '₃₀It is a figure showing the method of calculating. Host computer
Data 20 indicates that a scene change has occurred in the edited video data.
If not, the data obtained from encoder 162
The actual difficulty data D indicated by a circle in FIG.₁~ D_FifteenRaw
Actual difficulty data D₁~ D_FifteenBased on the figure
Predicted difficulty data D 'indicated by a cross in 10₁₆~ D '₃₀The
It is calculated for each type of picture (picture type).

That is, when no scene change occurs in the edited video data, the host computer 20
Picture actual difficulty data D ₂ , D ₃ ,..., D ₁₃ , D ₁₄
The values, extrapolated linearly approximated by the dotted line A in FIG. 10, B-picture predictive difficulty data _{D '16, D' 17,} ...,
D ′ ₂₉ and D ′ ₃₀ are generated, and the actual difficulty data D ₄ of the I picture and, if necessary, the values of the actual difficulty data D _j of the previous I picture are extrapolated by linear approximation to obtain the I picture. Predicted difficulty data D ′ ₁₈ is generated, and the actual difficulty data D ₁ , D ₇ ,..., D ₁₂ of the P picture and, if necessary, the actual difficulty data D _j of the previous P picture are linearly approximated. extrapolated Te, predictive difficulty data D _'15, D' ₂₁ P-picture, ..., to produce a D _'27. Further, the host computer 20 uses these real difficulty data D _j and the predicted difficulty data D _'j, the predicted simplified two pass method shown in the second embodiment calculates the target amount of data T _j.

Hereinafter, the processing content when the host computer 20 detects a scene change of edited video data in a P picture will be described in stages. The first step host computer 20, when it is detected that scene change P picture is generated, from only the real difficulty data D ₁₅ of the P-picture indicated by ● in Fig. 10, left and right by the amount or the like of the motion between the picture It is impossible to predict the difficulty of the B picture and the P picture to be performed.
Therefore, the host computer 20 uses the ratio (i: p: b) of the values of the actual difficulty data of the I picture, the P picture, and the B picture obtained in advance by an experiment or the like to calculate the total sum Sum _j defined in Expression 5. Ask for.

That is, the host computer 20 executes the j-th
In order to calculate the target data amount for the + 1st (j = 1 in FIG. 10) picture, for example, the ratio of the previously obtained I, P, and B picture actual difficulty data values (i: Substituting the actual difficulty data D _{j + 15} of the P picture in which the scene change has occurred into Equation 7 using p: b), and using it for calculating the target data amount T _{j + 1} for the (j + 1) th picture Sum S
um _{j + 1} is predicted, and the predicted sum value Sum _{j + 1 is} further predicted.
Is substituted into Equation 4 to calculate a target data amount T _{j + 1} for the (j + 1) -th picture.

[0093]

(Equation 7)

In the equation 7, the value of the actual difficulty data D _{j + 15} of the P picture in which the scene change has occurred is, as described above in the third embodiment, the value of the actual difficulty data D _{j + 18} of the immediately succeeding I picture. On the premise that they are equal, the host computer 20 first calculates the ratio (i: p: b) determined in advance and the coefficient multiplied by the number of I-pictures, P-pictures, and B-pictures included in one GOP after the scene change. Actual difficulty data D of the calculated P picture
_This means that the sum Sum _{j + 1} is calculated by multiplying _{j + 15} and further adding a predetermined constant α.

In equation (7), the constant α takes a predetermined value obtained in advance by an experiment or the like, and is immediately after the (j + 15) th P picture in FIG. ) -Th and (j + 1) -th
7) Since the B-th picture is generated only by forward prediction or backward prediction, it has a meaning as a margin in anticipation that the data amount is larger than other B pictures.

Assuming that the host computer 20 changes the linear prediction of the (j + 15) -th to (j + 30) -th difficulty data using the sum value Sum _j obtained by the equation 7, the prediction difficulty data D ′ The values of _{j + 15 to} D' _{j + 30} increase due to the scene change and become the values indicated by the dotted line B in FIG. However, in order to calculate the target data amount T _j , only the value of the sum Sum _j needs to be predicted. As described later, the value of the constant α is the sum Sum Sum for the (j + 2) -th picture. Since the correction is made when calculating _{j + 1} , unlike the case where no scene change occurs, the host computer 20 does not dare to predict difficulty data for each picture type (picture type) when a scene change occurs. .

[0097] The second stage host computer 20, the (j + 2) th when calculating the target amount of data T _{j + 2} are for picture, the (j + 16) th B picture real difficulty data D _{j + 16}
Is calculated. In the example shown in FIG. 10, the (j + 16) -th B picture belongs to the subsequent scene, but as shown in FIGS. 8A and 8B, the encoder control unit 12 changes the order of the pictures. Since it has been replaced, the (j + 16) -th B picture may belong to the previous scene, and is generated only by forward prediction or backward prediction. ) The actual difficulty data D _{j + 16} of the B-picture and the sum Sum _{j + 2} for calculating the target data amount T _{j + 2} for the (j + 2) -th picture
Cannot be used to predict

However, in equation (7), using the value of the actual difficulty data D _{j + 16} of the first one of the two B pictures in consideration of the margin as the constant α, the constant α in equation (7) Can be corrected. Therefore, the host computer 20 calculates the constant α ′ by correcting the constant α in Expression 7 based on the actual difficulty data D _{j + 16} as shown in Expression 8 below, and further calculates the sum S
um _{j + 2} can be predicted. The host computer 20 substitutes the predicted sum value Sum _{j + 2} into Expression 4, and
Target data amount T for the (j + 2) th picture
Calculate _{j + 2} .

[0099]

(Equation 8)

Third Stage When the host computer 20 calculates the target data amount T _{j + 3} for the (j + 3) -th picture, the actual difficulty data D _{j + 17 of} the (j + 17) -th B picture is calculated.
Is calculated. Therefore, in Equation 7, both sets of two B pictures in consideration of the margin as the constant α, that is, one set sandwiched between the I picture and the P picture in the picture type sequences shown in FIGS. Actual difficulty data D _{j + 16 for} all B pictures of
Since the value of D _{j + 16} has been found, the constant α in Expression 7 or the constant α ′ in Expression 8 is unnecessary as shown in Expression 9 below.

[0101]

(Equation 9)

[0102]Fourth stage The host computer 20 receives the (j + 4) th picture
Target data amount T for_{j + 3}When calculating
Actual difficulty data D of the (j + 18) th I picture_{j + 18}
Is calculated. At this stage, the example shown in FIG.
For all types (pictures) after the scene change
Actual difficulty data D of type) picture _iThe value of
You. Therefore, the previously calculated values used in Expressions 7 to 9 are obtained.
The value of the ratio (i: p: b) obtained is
0 is the actual difficulty data of the I picture actually calculated
D_{j + 18}, P picture actual difficulty data D_{j + 15}And P
Kucha's actual difficulty data D_{j + 16}(D_{j + 17})
It becomes possible.

As described above, the host computer 20
Equation 9 in which the ratio (i: p: b) obtained in advance is replaced with the actual ratio [D _{j + 18} : D _{j + 15} : D _{j + 16} (D _{j + 17} )]
, The sum value Sum _{j + 18} is more accurately predicted,
The target data amount T _{j + 4} for the (j + 4) -th picture is calculated by substituting into Equation 4.

Fifth Step Similarly to the fourth step, the target data amount T _{j + 3} for several (for example, 6 to 9) pictures after the (j + 5) -th picture.
Is calculated, and the host computer 20 obtains the actual difficulty data D _i in a sufficient quantity for the calculation of the predicted difficulty data D ′ _i , and then, as in the case where the scene change does not occur, the predicted difficulty data D _i is obtained by linear approximation. D ′ _i is calculated, and the calculated prediction difficulty data D ′ _i is substituted into Equation 4 to calculate a target data amount T _i .

[0105] The host computer 20 is, as described above in the third embodiment, based on the change in the real difficulty data D _i of I-picture, if it is determined that a scene change has occurred in the I-picture, the scene in P picture The same process as determining that a change has occurred,
By performing the processing of the first to fifth stages described above,
The target data amount T _i for each picture can be calculated.

[0106] On the other hand, when the host computer 20, as described above in the third embodiment, based on the change in the value of the real difficulty data D _i of B channels, it is determined that a scene change has occurred in the I-picture, The host computer 20 cannot perform the first-stage or second-stage processing when determining that a scene change has occurred in the P picture. Therefore, if the real difficulty data D _i of a scene change in the I-picture based on the change in the value of the B-channel is determined to have occurred, the host computer 20, a second when it is determined that a scene change occurs in the P picture The processing of the third or third stage is performed to calculate the target data amount T _i for each picture.

The processing for predicting the total sum Sum _i and calculating the target data amount T _i described above will be further described with reference to flowcharts. FIG. 11 and FIG.
FIG. 2 is a flowchart showing the processing contents related to the prediction of the sum Sum _i and the calculation of the target data amount T _i in the improved prediction simple two-pass encoding method in the third embodiment.

In FIGS. 11 and 12, the data SC_Flag indicates the position of a scene change when a scene change has occurred within the past 15 pictures, and is set to 0 in other cases. In addition, in the picture type sequences shown in FIGS. 8A to 8C, the value of the data I_Flag is 1 immediately after the I picture and until the processing for 3 pictures is completed, and is 0 otherwise. Become. Also, the coefficient Ith
1, Ith2, Pth, and Bth indicate coefficients used to determine the values of the I picture, P picture, and B picture when detecting a scene change.

As shown in FIG. 11, step 100 (S
100), the host computer 20 obtains predetermined data from the encoder 162, and obtains the actual difficulty data _Di.
Generate In step 102 (S102), the host computer 20 determines whether or not the value of the data SC_Flag is 0. If the value of the data SC_Flag is 0, the process proceeds to S200 (FIG. 12).
If it is not 0, the process proceeds to S104.

In step 104 (S104), the host computer 20 determines the type (picture type) of the i-th picture, and if the i-th picture is a B picture, a P picture, or an I picture, The process proceeds to S106, S120, and S128, respectively. In step 106 (S106), the host computer 20 determines whether or not the value of the data I_Flag is 0. If the value of the data I_Flag is 0, the process proceeds to S110; otherwise, the process proceeds to S110.
Proceed to 108. At step 108 (S108), the host computer 20, the real difficulty data D _i of the B picture is determined whether predictive difficulty data D _'i × or Bth larger, the greater the flow proceeds to the processing of S112, if it is smaller The process proceeds to S110.

In step 110 (S110), the host computer 20 performs the same processing as when no scene change occurs, and calculates the predicted difficulty data D′ _i . In step 112 (S112), the host computer 20 sets the value of the data SC_Flag to 1. In step 114 (S114), if the i-th picture is the first B picture after the scene change, the host computer 20 calculates the total sum Sum _i by equation 8, and In the case of the second B picture, the sum S
um _i is calculated.

In step 116 (S116), the host computer 20 substitutes the predicted sum value Sum _i or the predicted difficulty data D ′ _i into Equation 4 to obtain a target data amount T _i (target bit amount) for the i-th picture. ) Is calculated. In step 118 (S118), the host computer 20 increments the data i.

[0113] In step 120 (S220), the host computer 20, the real difficulty data D _i of P-picture is determined whether predictive difficulty data D _'i × or Pth larger, the greater the flow proceeds to the processing of S122 If it is smaller, the process proceeds to S110. Step 122 (S
At 122), the host computer 20 substitutes the data i for the data SC_Flag. Step 124
In (S124), the host computer 20 sets the value of the data I_Flag to 0. Step 126 (S
At 126), the host computer 20 predicts the total sum Sum _i using Expression 7.

[0114] In step 128 (S220), the host computer 20, the real difficulty data D _i of I picture is determined whether outside predictive difficulty data D _'i × Ith1~ predictive difficulty data D' _i × Ith2 If the value is out of the range, the process proceeds to S130.
Proceed to step 10. In step 130 (S130), the host computer 20 transmits the data SC_Flag
Is substituted for data i. In step 132 (S132), the host computer 20 transmits the data I_Fla
The value of g is set to 1, and the process proceeds to S126.

As shown in FIG. 12, step 200 (S
200), the host computer 20 determines that the value obtained by subtracting the data SC_Flag from the data i is 1, 2, 3
S202 and S20, respectively, when the number is
The process proceeds to steps S4, S206, and S210. Step 202
In (S202), the host computer 20 predicts the total sum Sum _i by using the equation 8, and S116 (FIG. 11)
Proceed to processing. In step 204 (S204),
The host computer 20 calculates the sum Sum _i
And the process proceeds to S116 (FIG. 11).

In step 206 (S206), the host computer 20 replaces the ratio (i: p: b) obtained in advance in equation 9 with the calculated actual difficulty data.
In step 208 (S208), the host computer 20 predicts the total sum Sum _i using Expression 9 in which the ratio (i: p: b) is replaced with the calculated actual difficulty data.

In step 210 (S210), the host computer 20 sets the picture (i-SC_Fl
ag) A straight line approximation is performed using the actual difficulty data for
The sum Sum _i (predicted difficulty data D ′ _i ) is calculated.
In step 212 (S212), the host computer 20 determines whether (i-SC_Flag) = 15. If (i-SC_Flag) = 15, the process proceeds to S214, and (i-SC_Flag)
If not = 15, the process proceeds to S110 (FIG. 11).

The host computer 20 has been described above.
Target data amount T generated by processing _jAnd encoder 1
8 is set in the quantization control circuit 180. Encoder 18
Is the same as in the first and second embodiments.
The target data set from the host computer 20
Volume T_jBased on the above, as shown in FIG.
The first P picture of the scene is the last P picture of the previous scene.
Change to I-picture so as not to refer to the culture,
Change last I picture of previous scene to P picture
Compression encoding, and output as compressed video data VOUT
I do.

As described above, according to the improved predictive simplified two-pass encoding method shown in the third embodiment, a larger amount of data can be allocated to video data including scene changes and camera flashes, and compression encoding can be performed. In addition, encoding distortion occurring before and after a scene change or a camera flash can be significantly reduced. Therefore, it is possible to improve the quality of the video obtained by decompressing and decoding the compressed video data generated by the improved simplified simple two-pass encoding method shown in the third embodiment.

In the third embodiment, N = 1
Equations (7) to (9), which are suitable for the processing for the picture sequence of 5, M = 3, have been exemplified. By doing so, it is possible to apply the improved prediction simple 2-pass encoding to other picture sequences.

Fourth Embodiment Hereinafter, as a fourth embodiment of the present invention, a modified example of the scene change detection method of the improved prediction simple 2-pass encoding method shown in the third embodiment will be described. First, the principle of the scene change detection method according to the fourth embodiment of the present invention will be described.

The video data compression apparatus 1 (FIG. 1) converts the edited video data in the vicinity of a scene change from the simplified predictive two-pass encoding method and the improved predictive simple two-pass method shown in the second and third embodiments, respectively. In the encoding method, the predicted difficulty data D _j ′ generated by using the temporal correlation between pictures of the video data reflects the tendency of the change in the difficulty of the video data before the actual difficulty data D _j−1. cage, error between the real difficulty data D _j becomes very small unless a scene change. For example, FIG.
In the case of 0, the prediction difficulty data D ₁₆ ′
This is a value obtained by predicting the difficulty of the next picture based on 15 pieces of actual difficulty data D _{1 to} D _15. If there is no scene change, it can be expected that the accuracy is extremely high.

FIG. 13 shows actual difficulty data D before and after a scene change occurs in a P picture.
FIG. 9 is a diagram illustrating the relationship between _j (○) and prediction difficulty data D ′ _j (×) in the order of compression encoding. On the other hand, FIG.
As shown in FIG. 7, when a scene change occurs in a P picture, the actual difficulty data D _j of the P picture immediately after the scene change often becomes impossible to perform compression encoding with reference to the preceding picture. Data D _j '
The value is much larger than

Conversely, the actual difficulty data D _j of the P picture in the scene change portion is the predicted difficulty data D _j when the pattern after the scene change is flatter than the pattern before the scene change, for example. May be significantly smaller than '. Further, the value of the real difficulty data D _j of the B-picture immediately after the scene change, in order to be compressed and encoded with reference to only the rear of the picture, predictive difficulty data D
It is much larger than _j ′, for example, as large as a P picture.

FIG. 14 shows actual difficulty data D before and after a scene change occurs in an I picture.
FIG. 9 is a diagram illustrating the relationship between _j (○) and prediction difficulty data D ′ _j (×) in the order of compression encoding. FIG.
As shown in the figure, when a scene change occurs in the j (16) th I picture, there is no temporal correlation between the I pictures before and after the scene change. error between the _j 'and the real difficulty data D _j occurs.

However, since the I picture is originally compressed and encoded without referring to other pictures, the prediction difficulty data D _j ′ and the actual difficulty data D _j ′ are compared with the case where a scene change occurs in the P picture. The difference from _j is small.
On the other hand, the value of the actual difficulty data D _j of the B picture immediately after the scene change is much larger than the predicted difficulty data D _j ′, as in the case where a scene change occurs in the P frame.

[0127] difficulty Thus, the 'even if a large error in the value of the difficulty data D _j is not generated, predictive difficulty data D _j of the B-picture itself' predictive difficulty data D _j of the P picture and I picture If a large error in the value of the data D _j occurs, it can be determined that the I-picture or P-picture at the scene change immediately before occurred.

The scene change detection method shown in the fourth embodiment utilizes the relationship between the actual difficulty data D _j and the predicted difficulty data D _j ′ described above, and the improvement shown in the third embodiment, respectively. In the simple two-pass encoding method,
This enables more accurate scene change detection. That is, the scene change detection method shown in the fourth embodiment is
In the improved simplified simple two-pass encoding method using the video data compression device 1 shown in the third embodiment, a scene change is accurately detected by comparing the values of the prediction difficulty data D _j ′ and the actual difficulty data D _j. It is supposed to.

More specifically, the detection of a scene change in the fourth embodiment is based on the actual difficulty data D _{jI of the} I picture.
_Value of the prediction difficulty data D _jI ′ with respect to (D _jI /
D _jI ′) and the ratio of the predicted difficulty data D _jp ′ to the actual difficulty data D _jp of the P picture (D _jp / D _jp ′)
Is outside the range of the predetermined threshold [Th _I1 <(D _j /
D _j ') or (D _jP / D _jP ') <Th _I2 , Th _p1 <
(D _jP / D _jP ') or (D _j / D _j ') <Th _p2 . However, Th _I1 >1> Th _I2 > 0, Th _p1 >1> Th _p2 >
0], the occurrence of a scene change is detected in the picture. However, usually, the value of the ratio of the predicted difficulty data D _jp ′ to the actual difficulty data D _jp of the P picture to the actual difficulty data D _jp of the P picture (D _jp / D _jp ′) rarely becomes less than the adjustment value Th _P2 .

The scene change detection method according to the fourth embodiment _uses the prediction difficulty data D _jI ′, and the actual difficulty data D _jI , D _jP of the I picture and the P picture.
Even if the value of the ratio of D _jP ′ is within the range of the predetermined threshold value, the value of the ratio (D _jB / D _jB ′ ₎ of the predicted difficulty data D _jB ′ to the actual difficulty data D _jB of the B picture. ) Is out of the predetermined range, [Th _B <(D _jB / D _jB ′).
However, Th _B> 1], the occurrence of a scene change, the B
It is detected that a scene change has occurred in the I picture or P picture immediately before the picture.

Next, the operation of the video data compression device 1 (FIG. 1) in the fourth embodiment will be described. As in the first to third embodiments, the encoder control unit 12 displays pictures of uncompressed video data in, for example, the order shown in FIG. Swap in order. FIFO memory 160 according to the first embodiment
As in the third embodiment, for example, input edited video data is delayed by 15 pictures. The encoder 162, as well as in the first embodiment to the third embodiment, regardless of the presence or absence of a scene change, compresses and encodes the video data S12, generates the real difficulty data D _j.

The host computer 20 has the encoder 1
Actual difficulty data D input from 62 _jAnd forecast difficulty data
D_j′, As described above in the fourth embodiment.
P-picture and I-picture prediction difficulty data
TA D_j’Actual difficulty data D_jThe value of the ratio to, and
B picture prediction difficulty data D_j’Actual difficulty data D
_jAt a position where the value of the ratio to
Detects that a traffic change has occurred.

When a scene change is detected, the host computer 20 further changes the first P picture of the subsequent scene to an I picture that does not refer to the last picture of the previous scene, as in the third embodiment. (FIG. 8C), the last I picture of the previous scene is set to P
The picture type sequence is changed like changing to a picture.

[0134] The host computer 20, as well as in the third embodiment, when the edited video data scene change does not occur, generates real difficulty data D _j from the data obtained from the encoder 162, the prediction difficulty Data D' _{16 to} D' ₃₀ are calculated for each picture type. Further, when a scene change occurs, the host computer 20 loses the correlation between the pictures before and after the scene change. Therefore, as in the third embodiment, the actual difficulty level of a predetermined number of pictures immediately after the scene change is changed. From the data D _j , the total sum Sum _j (Equation 5) is calculated according to Equation 6, and the target data amount T _j is calculated based on the calculated total sum Sum _j . Encoder 12, the amount of compressed data encoded, compressed and encoded non-compressed video data S16 delayed to be closer to the value indicated by the target amount of data T _j by the host computer 20 is generated, compressed video data VO
Output as UT.

Hereinafter, the details of the scene change detection processing by the host computer 20 of the video data compression apparatus 1 shown in the fourth embodiment will be described with reference to flowcharts. FIG. 15 is a flowchart showing the content of the scene change detection process by the host computer 20 of the video data compression device 1 (FIG. 1) in the fourth embodiment.

As shown in FIG. 15, step 300 (S
In 300), the host computer 20 calculates the j-th real difficulty data D _j. Step 302 (S
In 302), the host computer 20 determines whether or not there is a j-th picture. If there is a j-th picture, the process proceeds to S304; otherwise, the process ends. Step 304 (S304)
In, the host computer 20 determines the picture type of the j-th picture. If the picture type of the j-th picture is a B picture, an I picture, or a P picture,
6, the process proceeds to S316 and S320.

In step 306 (S306), the host computer 20 increments the numerical value B_count. In step 308 (S308),
The host computer 20 determines whether the value of the numerical value B_count is 1 or not. If the value of the numerical value B_count is 1, the process proceeds to S312, where the numerical value B_c
If the value of “out” is not “1”, the process proceeds to S310.

In step 310 (S310), the
The strike computer 20 determines whether a scene change has occurred.
Judge that Step 312 (S312)
The host computer 20 generates the B picture
Predicted difficulty data D_j’And the actual difficulty data D_jOf the ratio of
Calculate the value, D_j> Th_B× D_j’(D_jB/ D_jB’> T
h_B) Is determined. D _j> Th_B× D_j’
In step S310, the process proceeds to step S310, where D_j> Th_B× D
_jIf not, the process proceeds to S314. Step 31
4 (S314), the host computer 20
The immediately preceding I picture or P picture [(j-1)
Judge that a scene change has occurred
I do.

At step 316 (S316), the host computer 20 clears the value of the numerical value B_count to zero. In step 318 (S318),
The host computer 20 calculates a ratio value between the predicted difficulty data D _j ′ generated from the P picture and the actual difficulty data D _j, and calculates D _j > Th _P1 × D _j ′ or D _j <Th _P2 × D
_j 'is determined. If a _{D j> Th P1 × D j} ' or _{D j <Th P2 × D j} ', the flow proceeds to the processing of _{S324, D j> Th P1 ×} D j ' or D _j <Th _P2 × D
If it is not _j ′, the process proceeds to S310.

At step 320 (S320), the host computer 20
The value of the numerical value B_count is cleared to zero. Step 3
22 (S322), the host computer 20
Calculates the ratio value between the predicted difficulty data D _j ′ generated from the I picture and the actual difficulty data D _j, and calculates D _j > Th _I1 ×
It is determined whether D _j ′ or D _j <Th _I2 × D _j ′. D _j > Th _I1 × D _j ′ or D _j <Th _I2 ×
If D _j ′, the process proceeds to S324, where D _j > Th
_{If I1} × D _j ′ or D _j <Th _I2 × D _j ′, S
Proceed to 310.

In step 324 (S324), the host computer 20 determines that a scene change has occurred in the j-th picture. Step 326
In (S326), the host computer 20, using up real difficulty data D _j, next prediction difficulty data D
Calculate _{j + 1} . In step 328 (S328), the host computer 20 increments the numerical value j.

[0142] In the fourth embodiment, the prediction difficulty data D _j 'as a method of predicting, but using a linear approximation shown in the third embodiment, predictive difficulty data D _j' prediction method is Alternatively, for example, based on the difference value of the real difficulty data D _j, it may be adopted a method for calculating a prediction difficulty data D _j 'by predicting a change in the real difficulty data D _j. Further, in the fourth embodiment, when detecting a scene change, whether the preceding picture of the B picture is an I picture or a P picture, the prediction difficulty data D _j ′ and the actual difficulty data of the same B picture are used. D
in the comparison with the _j, but with the same threshold value Th _B, depending on the picture type of the previous picture, the threshold value may be changed.

[0143] According to the method of detecting the scene change described in the fourth embodiment above, by monitoring the change over time of the real difficulty data D _j shown in the third embodiment,
A scene change in an I picture that is difficult to detect, or a scene change in a P picture when a picture before the scene change is difficult and the picture after the scene change is gentle, can be reliably detected. Therefore, the quality of video data after compression encoding can be improved as compared with the case where the scene change detection method described in the third embodiment is employed.

[0144]

As described above, according to the video data compression apparatus and method according to the present invention, video data continuously including a plurality of scenes can be reduced to a predetermined data amount or less without using two-pass encoding. Compressed video data can be generated by compression encoding, and the quality of video obtained by expanding and decoding compressed video data obtained by compressing and encoding boundaries (scene changes) in the time direction of a plurality of continuous scenes Can be held.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a video data compression device according to the present invention.

FIG. 2 is a diagram illustrating a configuration of an encoder of a simple two-pass processing unit illustrated in FIG. 1;

FIG. 3 is a diagram illustrating a configuration of an encoder illustrated in FIG. 1;

FIGS. 4A to 4C are diagrams illustrating an operation of a simple two-pass encoding of the video data compression device according to the first embodiment.

FIGS. 5A to 5C are diagrams illustrating the operation of the video data compression device.

FIG. 6 is a flowchart showing the operation of the video data compression device (FIG. 1) in the second embodiment.

FIGS. 7A to 7C are diagrams for pictures before and after a scene change according to the simplified simplified two-pass encoding method according to the second embodiment and the improved simplified simplified two-pass encoding method according to the third embodiment; FIG. 3 is a diagram illustrating compression encoding.

8A to 8C are encoder control units (FIG. 1)
FIG. 8 is a diagram showing a process of changing the order of pictures in edited video data by a computer and a process of changing a picture type by a host computer.

FIG. 9 is a diagram exemplifying a temporal change in the value of actual difficulty data near a scene change portion of edited video data.

FIG. 10 shows a method in which a host computer (FIG. 1) calculates predicted difficulty data D ′ _{16 to} D ′ ₃₀ based on actual difficulty data D _{1 to} D ₁₅ when a scene change occurs in edited video data; FIG. 11 is a diagram illustrating a method of calculating predicted difficulty data D ′ _{16 to} D ′ ₃₀ when a scene change does not occur in edited video data.

11 is a first flowchart showing a process according to the calculation of the predicted and target amount of data T _i of the summation value Sum _i in the improved predictive simplified two pass encoding system in the third embodiment.

12 is a second flowchart showing the processing contents of the calculation of the predicted and target amount of data T _i of the summation value Sum _i in the improved predictive simplified two pass encoding system in the third embodiment.

FIG. 13 illustrates a relationship between actual difficulty data D _j (Ｄ) and predicted difficulty data D ′ _j (×) before and after a scene change occurs in a P picture in the order of compression encoding. FIG.

FIG. 14 illustrates the relationship between actual difficulty data D _j (Ｄ) and predicted difficulty data D ′ _j (×) before and after a scene change occurs in an I picture in the order of compression encoding. FIG.

FIG. 15 is a flowchart showing the content of a scene change detection process by the host computer of the video data compression device (FIG. 1) in the fourth embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Video data compression apparatus, 10 ... Compression coding part, 14 ...
Motion detector, 16: Simple 2-pass processing unit, 160: FIF
O memory, 162, 18 ... encoder, 20 ... host computer.

Claims (11)

[Claims] 1. After a head of a plurality of uncompressed video data input continuously is compressed into a predetermined picture type sequence composed of a combination of an I picture, a P picture and a B picture by a predetermined compression method. ,
Means for changing the order of pictures so as to be I pictures or P pictures; and compressing the uncompressed video data whose order has been changed by the predetermined compression method to generate first compressed video data. A first compression unit; a delay unit that delays the uncompressed video data whose order has been changed by a predetermined delay time; and the first compression generated from the uncompressed video data corresponding to the predetermined delay time. A prediction unit for predicting a data amount of a predetermined amount of the uncompressed first compressed video data based on a data amount of the video data; a predicted data amount of the first compressed video data; Head detecting means for detecting a head of the non-compressed video data based on a data amount (actual data amount) of the first compressed video data; The beginning of the picture of the reduced image data,
Changing means for changing the predetermined picture type sequence so as not to have a relationship with a picture of another video data after compression; generated first compressed video data; and predicted first compression. A target value generating means for generating a target value of the data amount of the uncompressed video data after compression based on the data amount of the video data; A second compression means for compressing the uncompressed video data into the changed predetermined picture type sequence by the predetermined compression method. 2. The head detecting means according to claim 1, wherein the actual data amount of the I picture and the P picture is such that the predicted ratio of the first compressed video data to the I picture and the P picture is out of a predetermined range. 2. The video data compression apparatus according to claim 1, wherein when the data amount increases, a head of the uncompressed video data is detected at a position corresponding to the P picture whose data amount has increased. 3. The head detecting means according to claim 1, wherein the actual data amount of the B picture is the B data of the predicted first compressed video data.
When the data amount of the uncompressed video data is larger than the data amount of the picture by a predetermined ratio or more, a head of the uncompressed video data is detected at a position of a P picture and an I picture immediately before the B picture whose data amount is increased. 2. The video data compression device according to 1. 4. The method according to claim 1, wherein, in the predetermined picture type sequence, when the head of the uncompressed video data is compressed into a P picture, the head of the uncompressed video data is compressed into an I picture. 2. The video data compression apparatus according to claim 1, wherein said predetermined picture type sequence is changed. 5. The non-compressed video data, wherein when the predetermined picture type sequence is changed so that the head of the non-compressed video data is compressed into an I picture, the uncompressed video data becomes an I picture after neighboring compression. 5. The method of claim 4, wherein the predetermined picture type sequence is further modified such that a picture of data is compressed into a P picture.
2. The video data compression device according to claim 1. 6. After a head of a plurality of uncompressed video data which are continuously inputted is compressed into a predetermined picture type sequence composed of a combination of I picture, P picture and B picture by a predetermined compression method. ,
Generating a first compressed video data by compressing to become an I picture or a P picture; and generating a predetermined amount of the uncompressed video data based on a data amount of the first compressed video data corresponding to the predetermined delay time. The data amount of the first compressed video data, and the predicted data amount of the first compressed video data and the data amount (actual data amount) of the actually generated first compressed video data. A video data compression method for detecting a head of the uncompressed video data based on the video data. 7. A method according to claim 7, wherein said uncompressed video data is delayed by a predetermined delay time, and said predetermined picture type sequence is changed so that a detected P picture becomes an I picture after compression. The first compressed video data and the first
Generating a target value of the compressed data amount based on the data amount with the compressed video data of the compressed video data, and delaying the uncompressed video data so that the compressed data amount becomes the generated target value. 7. The video data compression method according to claim 6, wherein the video data is compressed into the changed predetermined picture type sequence by the predetermined compression method. 8. The position of the I picture and P picture in which the ratio of the actual I picture and P picture to the predicted data amount of the I picture and P picture is out of a predetermined range is used as the first picture. The video data compression method according to claim 7, wherein the detection is performed. 9. The position of an I picture and a P picture immediately before a B picture whose actual data amount has been increased by a predetermined ratio or more than the predicted data amount of a B picture is taken as the head of the first compressed video data. Claim 7 for detecting
2. The video data compression method according to 1. 10. In the predetermined picture type sequence, when the head of the uncompressed video data is compressed into a P picture, the head of the uncompressed video data is compressed into an I picture. 8. The video data compression method according to claim 7, wherein a predetermined picture type sequence is changed. 11. When the predetermined picture type sequence is changed so that the head of the uncompressed video data is compressed into an I picture, the picture of the uncompressed video data that becomes an I picture after nearby compression is: The video data compression method according to claim 10, wherein the predetermined picture type sequence is further changed so as to be compressed into a P picture.