JP2011004342A - Encoding preprocessor, encoding device, decoding device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an encoding preprocessor for adaptively encoding a moving image in a motion scene; and to provide an encoding device, a decoding device, and a program.SOLUTION: The encoding preprocessor 1 includes: a block division part 21 for dividing a certain frame image into a plurality of pixel blocks by a predetermined block size; a motion search-processing part 22 for calculating a motion vector in each of the pixel blocks; a motion vector clustering-processing part 23 for performing clustering for classifying all motion vectors so as to identify as the predetermined number of motion vector candidate clusters and calculating a moving speed in the image frames; and a wide-area and high-speed motion determination part 24 for determining whether the image frame is a wide-area and high-speed motion frame or not, and when determining that the image frame is the wide-area and high-speed motion frame, determining the frame image as a frame to be forcedly subjected to intra-frame encoding.

Description

  The present invention relates to an adaptive encoding technique in a motion scene, and more particularly to an encoding preprocessing device, an encoding device, a decoding device, and a program in a high-speed motion scene.

  Ultra-high definition moving images with 16 times higher resolution such as Super Hi-Vision (SHV) called 8K is a moving image system aiming at a larger screen than conventional Hi-Vision, and has a horizontal viewing angle at a standard viewing distance. Since it has a wide field of view, it is possible to enjoy a moving image with a sense of presence that seems to be wrapped in an image. However, since the frame rate is different from that of a conventional standard image quality or high-definition screen, when moving images are captured at the same angle of view, a difference occurs in the motion speed between frames.

  For example, as shown in FIG. 8A, a high-definition screen is 1920 pixels × 1080 lines, and if the viewpoint is at a distance of 3H with respect to the screen height (H), the horizontal viewing angle is 30 degrees. On the other hand, as shown in FIG. 8B, the super high-definition screen has 7680 pixels × 4320 lines, and if the viewpoint is at a distance of 0.75H with respect to the screen height (H), the horizontal viewing angle is Is 100 deg. In this case, when the super high definition video signal for the super high-definition screen and the video signal for the high vision screen are compared, the super high definition video signal is quadrupled in both the horizontal and vertical resolutions. In the case of rate, the motion speed between frames is also quadrupled.

JP 2008-131321 A

  Wide-field ultra high-definition moving images not only tend to have a higher motion speed between frames than conventional moving images, but also increase the amount of motion blur per pixel in the moving region. Even if the block size is simply increased or the motion search range is expanded in a motion detection method using block matching such as H.264 (MPEG-4 AVC) or MPEG-2, the motion vector detection accuracy is not so high. Image quality degradation due to inter-coding tends to be noticeable. This is particularly noticeable in the case of a moving scene.

  Therefore, an object of the present invention is to provide a pre-encoding device, an encoding device, a decoding device, and a program for adaptively encoding a motion scene.

  The present invention relates to a high-speed motion scene of an ultra-high definition moving image with a wide field of view. In order to encode with an encoding method using an intra-frame encoding process (I frame / I slice) and an inter-frame encoding process (P, B frame / P, B slice) such as H.264 or MPEG-2, When motion detection is performed in a relatively large block size and a large motion search range, and the motion vectors are clustered to determine whether the motion region is dominant and the motion speed is high, and it is determined that the motion speed is high Provides a pre-encoding device for sending auxiliary information for performing intra-frame encoding, as well as an encoding device, a decoding device, and a program.

  That is, the pre-encoding apparatus of the present invention performs pre-processing for performing pre-processing in order to adaptively encode a moving image in a motion scene with an encoding method using intra-frame encoding processing and inter-frame encoding processing. A processing device, a block dividing unit that divides a certain frame image in a moving image frame sequence into a plurality of pixel blocks with a predetermined block size, and another frame image different from the frame image in the image frame sequence As a search range, a motion search processing unit that calculates a motion vector in each of the pixel blocks, clustering that classifies all the calculated motion vectors so as to be identifiable as a predetermined number of motion vector candidate clusters, and the image frame A motion vector clustering processing unit for calculating a motion speed between the predetermined number of motion vectors Whether each frame is a wide area / high-speed motion frame is determined from each motion vector of the complementary cluster, and when it is determined that the frame is a wide area / high-speed motion frame, the frame image is used regardless of the encoding method. And a wide area / high-speed motion determination unit that forcibly determines a frame to be subjected to intra-frame coding, and the wide area / high-speed motion determination unit allocates among the predetermined number of motion vector candidate clusters. When the number of motion vectors obtained is larger than the number specified as a ratio to the total motion vectors and the value of the motion vector is larger than the specified value, it is determined that the frame has a high-speed motion in a wide area. It is characterized by that.

  In the pre-encoding apparatus of the present invention, the motion speed between frames is an average value of all motion vector amounts.

  In the pre-encoding apparatus of the present invention, the predetermined block size is larger than a motion detection block size defined by the encoding method.

  In the pre-encoding apparatus of the present invention, a search range searched by the motion search processing unit is larger than a search range defined by the encoding method.

  In the pre-encoding apparatus of the present invention, the clustering is performed according to a k-means method.

  Furthermore, the encoding device of the present invention determines whether or not the pre-coding device of the present invention and the moving image frame sequence are wide-area / high-speed motion frames determined by the pre-coding device. And an encoder that performs encoding according to the encoding method according to the attached information.

  Furthermore, the decoding device of the present invention is characterized in that the moving picture frame sequence encoded by the encoding device of the present invention performs decoding of the encoding method according to the attached information.

  Furthermore, the present invention provides a computer configured as a pre-encoding device that performs pre-processing to adaptively encode a moving image in a motion scene by an encoding method using intra-frame encoding processing and inter-frame encoding processing. A step of dividing a certain frame image in the moving image frame sequence into a plurality of pixel blocks with a predetermined block size, and using another frame image different from the frame image in the image frame sequence as a search range, A step of calculating a motion vector in each of the blocks; a step of performing clustering for classifying all the calculated motion vectors so as to be identifiable as a predetermined number of motion vector candidate clusters; and calculating a motion speed between the image frames; From each motion vector of the predetermined number of motion vector candidate clusters, It is determined whether the frame is a region / high-speed motion frame. When it is determined that the frame is a wide region / high-speed motion frame, the frame image is forcibly encoded regardless of the encoding method. And a step for determining a frame to be subjected to.

  Furthermore, the present invention is also characterized as a program for causing a computer configured as a decoding apparatus of the present invention to execute decoding of the predetermined encoding method according to the attached information.

  According to the present invention, it is possible to prevent image quality deterioration in a high-speed motion scene that is likely to cause a problem in a wide-field ultra-high-definition moving image, and to provide a simple and highly effective encoding technique.

It is a figure which shows the encoding apparatus of one Example by this invention. It is a figure which shows the decoding apparatus of one Example by this invention. It is a flowchart which shows operation | movement of the encoding apparatus of one Example by this invention. It is a figure which shows the block division in the encoding apparatus of one Example by this invention. It is a figure which shows the motion search in the encoding apparatus of one Example by this invention. It is explanatory drawing of the Eugrid distance in the encoding apparatus of one Example by this invention. It is a figure which shows the clustering in the encoding apparatus of one Example by this invention. (A) is a figure which shows an example of a high-definition screen, (b) is a figure which shows an example of a super high-definition screen. It is a figure which shows the mode of the high spatial frequency component at the time of DCT of the motion blur area | region in a high-speed motion scene.

  Hereinafter, an encoding apparatus and a decoding apparatus according to an embodiment of the present invention will be described.

  First, in order to facilitate understanding of the present invention, a high-speed motion scene will be described. In a high-speed motion scene, the motion region is dominant in the frame, and the motion speed between the frames increases, so that it can be said that most of the images on the screen cause motion blur. Therefore, in the encoding apparatus of the present embodiment, intra-frame encoding (I slice / picture) processing is forcibly performed in a high-speed motion scene in which the moving region is dominant.

  In a high-speed motion scene, image quality degradation is less when performing intra-frame coding (I slice / picture) processing than when performing inter-frame coding (P or B slice / picture processing). Furthermore, since motion blur occurs in most of the images in a high-speed motion scene, it is sufficient to perform I slice / picture processing with a small bit amount. At this time, most of the high spatial frequency components are lost in DCT (Discrete Cosine Transform) and quantization in the I slice picture processing, but most of the high spatial frequency components are originally lost due to motion blur. The image degradation caused by the image does not need to be a problem. For example, as shown in FIG. 9, most of the high spatial frequency components at the time of DCT in the motion blur region are lost.

  FIG. 1 shows an encoding apparatus according to an embodiment of the present invention. The encoding apparatus 1 according to the present embodiment includes a high-speed motion scene determination unit 2 and an encoding unit 3. The high-speed motion scene determination unit 2 includes a block division unit 21, a motion search processing unit 22, a motion vector clustering processing unit 23, and a wide area / high-speed motion determination unit 24. Data used for each process of the encoding device 1 according to an embodiment of the present invention can be appropriately stored in a storage unit (not shown) included in the encoding device 1.

  The block division unit 21 stores the frame image F (t) at an arbitrary time position t in the moving image frame sequence in the storage unit and divides the frame image F (t) into a plurality of pixel blocks with an arbitrary block size. To do.

  The motion search processing unit 22 performs motion detection using the entire screen of another frame image (for example, the subsequent frame image F (t + 1)) as a search range, calculates a motion vector in each pixel block, and calculates all the calculated values. The motion vector is sent to the motion vector clustering processing unit 23.

  The motion vector clustering processing unit 23 performs clustering for all motion vectors calculated by the motion search processing unit 22 and calculates a motion speed between frames. Clustering is a process of classifying each motion vector so as to be identifiable as a motion vector candidate cluster. The motion speed between frames can be, for example, an average value of all motion vector amounts.

  The wide area / high-speed motion determination unit 24 has a predetermined number of motion vector candidate clusters in which the number of allocated motion vectors is greater than a specified number and the value of the motion vector is greater than a specified value. It is determined that the frame region is a frame image F (t) in which the moving area in the frame is dominant and the movement speed between frames is large. In other words, the wide area / high-speed motion determination unit 24 is positive in determining whether the number of assigned motion vectors is greater than a specified number and whether the value of the motion vector is greater than a specified value. It is determined that the frame image is a wide area / high-speed motion frame image (for example, a frame image of a high-speed motion scene). The wide area / high-speed motion determination unit 24 sends the determination result to the encoder 3 as auxiliary information indicating the slice type.

  The encoder 3 refers to auxiliary information indicating whether or not the frame image F (t) is a wide-area / high-speed motion frame image, and encodes an arbitrary moving image (eg, H.264, for example). Encoding based on MPEG-2) is performed. When the encoder 3 refers to the auxiliary information to indicate that the frame image is a wide area / high-speed motion frame, the frame image F (t) is compulsorily framed regardless of the encoding method. An inner coding (I slice / picture) process is performed. On the other hand, when the auxiliary information indicates that it is not a wide area / high-speed motion frame image, encoding according to the encoding method, that is, intra-frame encoding (I slice / picture) processing or inter-frame encoding (P or B) A coded image is generated by performing slice / picture processing. The above auxiliary information is given to this encoded image. The encoded image is stored in the storage unit or transmitted to an external device.

  FIG. 2 shows the decoding device 4 of the present embodiment. The decoding device 4 according to the present embodiment reads the encoded image and auxiliary information stored in the storage unit included in the decoding device 4 or receives the encoded image and auxiliary information to be transmitted. A decoding process according to the encoding method is executed. Note that the decoding device 4 of the present embodiment refers to the auxiliary information, performs intra-frame coding (I slice / picture) processing or inter-frame coding (P or B slice / picture processing), and decodes the decoded frame. An image F ′ (t) is generated.

  Hereinafter, the operation of the encoding apparatus according to an embodiment of the present invention will be described in more detail with reference to FIG.

  In step S1, the block dividing unit 21 of the encoding device 1 inputs the frame image F (t) to be processed and the subsequent frame image F (t + 1). In step S2, the block dividing unit 21 uses the frame image F (t + 1). F (t) is divided into a plurality of pixel blocks.

  Note that a block of 16 × 16 pixels is usually used in consideration of a case where a high-definition screen is encoded by MPEG-2. However, the block size divided by the block dividing unit 21 is, for example, a 16 × 16 pixel block four times in the horizontal and vertical directions, that is, a 64 × 64 pixel block, so that it can be applied to a super high-definition screen. It is preferable to use a larger pixel block than that. In other words, the block size divided by the block dividing unit 21 is preferably larger than the block size for motion detection defined by the encoding method of the encoder 3.

  For practical purposes, it is assumed that the entire screen pans or a high-speed motion scene in which an extremely large moving object moves at a high speed. Therefore, as shown in FIG. 4, a relatively large pixel block size is used. Alternatively, a large block of about 1/4 or 1/16 of the frame image F (t) is prepared in the center of the screen, and the entire frame image F (t + 1) screen is set as one search range. It is effective to detect only the motion vector. That is, preferably, the search range searched by the motion search processing unit 22 is larger than the search range defined by the encoding method of the encoder 3.

  In the following description, a case where the frame image F (t) is divided into a plurality of pixel blocks B (Bx, By) as shown in FIG. 4 will be described. Bx is a block position number in the horizontal direction, and By is a block position number in the vertical direction.

  In step S3, the motion search processing unit 22 of the encoding device 1 generates, for example, a frame image F (t + 1) for each pixel block B (Bx, By) divided into a plurality of blocks as shown in FIG. The motion search is executed with reference to the reference.

  Since the motion search process takes time, it is possible to search for a representative pixel block position while skipping several pixel blocks instead of searching all pixel blocks. By the processing of the motion search processing unit 22, a motion vector Mv_B (Bx, By) in each pixel block B (Bx, By) is obtained.

  In step S4, the motion vector clustering processing unit 23 of the encoding device 1 calculates the clustering of the motion vector Mv_B (Bx, By) and the motion speed between frames. As a clustering method, a k-means method that is well-known as a non-hierarchical clustering algorithm is used.

  The algorithm of the k-means method will be briefly described.

  First, k motion vector candidate cluster centers set in advance as initial values are set. The motion vector candidate cluster center is searched in the 360 degree direction of the motion vector (for example, every 45 degrees search), and a position having a large value is set as an initial candidate for the cluster center of the motion vector candidate cluster.

Next, each motion vector Mv_B (Bx, By) is assigned to the motion vector candidate cluster center with the shortest Eugrid distance. The Eugrid distance can be defined as an average value of vector quantities u ⁱ from the motion vector candidate cluster center O to each pixel value (Bx, By) as shown in FIG. 6, for example.

  Next, the cluster center of each motion vector candidate cluster is recalculated for each motion vector candidate cluster. If the change values of all cluster centers are equal to or less than a certain threshold value Th, the process is terminated. Otherwise, the cluster is assigned again as a motion vector candidate cluster center.

  In the motion vector clustering, clustering is performed with the size and orientation of each motion vector Mv_B (Bx, By) with the origin at the center. For example, as shown in FIG. 7A, for the frame image F (t), a “person” that moves to the right and a “background” that also moves to the right at a slightly slower speed than “person”, When three clusterings are calculated on the assumption that there are three “other” objects other than “person” and “background”, clustering is performed with the size and orientation of each motion vector Mv_B (Bx, By), As shown in FIG. 7B, the motion vector candidate cluster representing “person” is classified as MVset1, the motion vector candidate cluster representing “background” is classified as MVset2, and the vector candidate cluster MVset3 representing “other”. be able to.

  Since the determination processing of k motion vector candidate clusters can be converged by the above processing, all motion vectors Mv_B (Bx, By) can be assigned to k motion vector candidate clusters.

  In step S5, the wide region / high-speed motion determination unit 24 of the encoding device 1 is a frame in which the motion region in the frame is dominant and the motion speed between the frames is high for the frame image F (t). If it is affirmative, it is determined that the frame image is a wide area / high-speed motion frame (for example, a high-speed motion scene), and the process proceeds to step S6. It determines with it not being an image, and progresses to step S7.

  For example, among k motion vector candidate clusters, the number of allocated motion vectors Mv_B (Bx, By) is larger than a prescribed number (more than the prescribed number as a ratio to all motion vectors), and the motion vector A candidate whose value is larger than a specified value is determined to be a frame in which the moving area is dominant and the moving speed is high. As a specific example, if all the motion vectors Mv_B (Bx, By) are assigned at least 1/4 and the motion speed is equal to or higher than [the horizontal resolution / frame rate of the frame image F (t)], it corresponds. The frame image F (t) is determined to be a frame in which the moving area is dominant and the moving speed is high. For example, in a moving image signal of a super high-definition screen having a horizontal resolution of 7680 pixels and a frame rate of 60 Hz, this corresponds to 128 pixels / frame as a moving speed across the screen in one second. The wide area / high-speed motion determination unit 24 sends the determination result to the encoder 3 as auxiliary information indicating the slice type.

  In step S6, the wide area / high-speed motion determination unit 24 of the encoding device 1 forcibly sets the positive determination result as the slice type regardless of the encoding method as auxiliary information indicating the slice type. An I-slice picture flag indicating that it should be processed as an I-slice picture is set valid.

  On the other hand, in step S7, the wide area / high-speed motion determination unit 24 of the encoding device 1 sets the negative determination result as the slice type in order to process the negative determination result as auxiliary information indicating the slice type according to the encoding method. The I slice / picture flag indicating that it should not be forcibly processed as an I slice / picture is set invalid.

  For this forced I-slice / picture processing (hereinafter referred to as “IF slice / picture”), it is necessary to control the rate because the encoding bit rate is determined normally. Since it is determined that the image F (t) includes a lot of motion blur, there is no problem of image degradation even if the rate of the IF slice picture is lowered.

  For example, when Closed GOP = 15, the number of P and B slice pictures is about 1/10 of the number of I slice pictures inserted at the beginning of the GOP. Here, when the IF slice picture is set to be about 1/5 of the I slice picture and the rate control is performed, a so-called intra flicker or the like is not conspicuous and a good image result is obtained. However, IF slices / pictures are likely to be continuous.

  Finally, in step S8, the encoder 3 of the encoding device 1 performs predetermined processing with reference to auxiliary information indicating whether or not the frame image F (t) is a wide region / high-speed motion frame image. Are encoded based on the moving image encoding method (for example, H.264, MPEG-2).

  As described above, according to the encoding device and the decoding device of the present embodiment, motion detection is performed with a relatively large block size and a large motion search range, and the motion vectors are clustered, and the motion region is dominant. In addition, it is determined whether or not the motion speed is high, and if it is determined that the motion speed is high, a control instruction for performing intra-frame encoding is transmitted as auxiliary information, for example, a super-high definition video with a wide field of view. It is possible to prevent image quality deterioration in a high-speed motion scene that is likely to cause a problem with an image, and to provide a simple and highly effective encoding technique.

  The present invention searches for a frame in which the motion region is dominant in the frame and the motion speed between the frames is large. It may be performed within the block matching process of H.264 (MPEG-4 AVC) or MPEG-2. However, by adopting a larger block size and a larger motion search range than those used in these encoding methods, it becomes possible to more reliably discriminate a high-speed motion scene or pan frame.

  Each of the encoding device 1 and the decoding device 4 according to the present embodiment can be configured as a computer. A program for realizing each function of these devices is stored in a storage unit of a computer, and the program is appropriately read and executed by a central processing unit (CPU). Functions can be realized.

  Similarly, the high-speed motion scene determination unit 2 that functions as an encoding preprocessing device according to the present embodiment can also be configured as a computer. A program for realizing each function of the high-speed motion scene determination unit 2 is stored in a storage unit of a computer, and the program is appropriately read out and executed by a central processing unit (CPU), so that Functions can be realized.

  In the above embodiment, one frame image of a moving image has been representatively described, but the present invention can also be applied to compression such as Motion JPEG. Further, the encoding apparatus 1 has been described as including the high-speed motion scene determination unit 2, but the high-speed motion scene determination unit 2 is a separate apparatus, and the high-speed motion scene determination unit 2 and the encoder 3 are separate apparatuses. Can be configured. Accordingly, the present invention is not limited to the above-described embodiments, but is limited only by the description of the scope of claims.

  According to the present invention, there is a problem in moving image coding of a high-speed motion scene that occurs with a higher probability than a conventional moving image system in an ultra-high-definition moving image signal for a super high-definition screen that is an ultra-high-definition moving image with a wide field of view. Because it can be solved by a relatively simple method, the use of coding technology such as high-speed panning such as soccer and basketball broadcasts, where similar problems are likely to occur not only in future Super Hi-Vision but also in current Hi-Vision Useful for.

DESCRIPTION OF SYMBOLS 1 Encoding apparatus 2 High speed motion scene determination device 3 Encoder 4 Decoding apparatus 21 Block division part 22 Motion search process part 23 Motion vector clustering process part 24 Wide area | region and high speed motion determination part

Claims (9)

An encoding pre-processing device that performs pre-processing to adaptively encode a moving image in a motion scene by an encoding method using intra-frame encoding processing and inter-frame encoding processing,
A block dividing unit for dividing a certain frame image in the moving image frame sequence into a plurality of pixel blocks with a predetermined block size;
A motion search processing unit that calculates a motion vector in each of the pixel blocks using a different frame image different from the frame image in the image frame sequence as a search range;
Performing clustering to classify all the calculated motion vectors so as to be identifiable as a predetermined number of motion vector candidate clusters, and a motion vector clustering processing unit for calculating a motion speed between the image frames,
From each motion vector of the predetermined number of motion vector candidate clusters, it is determined whether or not the frame is a wide area / high speed motion frame. Regardless, the frame image includes a wide area and high-speed motion determination unit that determines the frame image as a frame to be subjected to the intra-frame encoding.
The wide area / high-speed motion determination unit is configured such that, among the predetermined number of motion vector candidate clusters, the number of allocated motion vectors is greater than a number defined as a ratio to all motion vectors, and the value of the motion vector An encoding pre-processing apparatus that determines that the frame has a high-speed motion in a wide area when is larger than a specified value.   The pre-coding apparatus according to claim 1, wherein the motion speed between the frames is an average value of all motion vector amounts.   The pre-encoding apparatus according to claim 1 or 2, wherein the predetermined block size is larger than a block size for motion detection defined by the encoding method.   The encoding preprocessing apparatus according to any one of claims 1 to 3, wherein a search range searched by the motion search processing unit is larger than a search range defined by the encoding method. .   The encoding preprocessing apparatus according to claim 1, wherein the clustering is performed according to a k-means method. An encoding pre-processing device according to any one of claims 1 to 5;
An encoder that performs encoding of the encoding scheme in accordance with attached information indicating whether or not the moving image frame sequence is a wide-area high-speed motion frame determined by the encoding pre-processing device;
An encoding device comprising:   7. A decoding apparatus, wherein the moving picture frame sequence encoded by the encoding apparatus according to claim 6 is decoded by the encoding method according to the attached information. In a computer configured as a pre-encoding processing device that performs pre-processing to adaptively encode a moving image in a motion scene with an encoding method using intra-frame encoding processing and inter-frame encoding processing,
Dividing a certain frame image in the moving image frame sequence into a plurality of pixel blocks with a predetermined block size;
Calculating a motion vector in each of the pixel blocks using a different frame image different from the frame image in the image frame sequence as a search range;
Performing clustering to classify all the calculated motion vectors so as to be identifiable as a predetermined number of motion vector candidate clusters, and calculating a motion speed between the image frames;
From each motion vector of the predetermined number of motion vector candidate clusters, it is determined whether or not the frame is a wide area / high speed motion frame. Regardless, the step of determining the frame image as a frame to be forcibly intra-frame encoded,
A program for running   The program for making the computer comprised as a decoding apparatus of Claim 7 perform performing the decoding of the said encoding system according to the said attached information.

Images