JP2961828B2 - Image data decoding apparatus and decoding method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a decoding device and a decoding method for decoding compression-encoded image data.

[Conventional technology]

The applicant of the present application obtains a dynamic range, which is a difference between the maximum value and the minimum value of a plurality of pixels included in a two-dimensional block, as described in JP-A-61-144989.
An adaptive coding device that performs coding (referred to as ADRC) adapted to the dynamic range has been proposed. In addition, the present applicant has proposed an ADRC and a variable-length ADRC for a three-dimensional block formed from pixels in an area included in each of a plurality of frames.

In the above-mentioned ADRC, redundancy in the level direction can be removed and the number of bits per pixel can be reduced, so that the amount of data to be transmitted can be greatly reduced. Such ADRC is suitable for application to a digital VTR. In particular, variable length ADRC is
The compression ratio can be increased. But variable length ADRC
However, since the amount of transmission data varies depending on the content of an image, buffering processing is necessary when using a fixed-rate transmission path such as a digital VTR that records a predetermined amount of data as one track.

As a method of buffering variable-length ADRC, the present applicant has formed a frequency distribution of an integral type dynamic range as described in JP-A-63-111781,
A set of threshold values prepared in advance is applied to this frequency distribution to determine the amount of generated data for a predetermined period, for example, one frame period, and to control the generated data amount so as not to exceed the target value. is suggesting.

FIG. 5 shows, for simplicity, pixels P that are aligned horizontally.
ADRC using a one-dimensional block consisting of 1, P2, P3, and P5 as an example
It is to explain. In FIG. 5, the waveform of the solid line represents the original signal.

In the case of non-edge matching where the number n of bits allocated to the block is 2 bits, as shown in FIG. 5A, the dynamic range DR which is the difference between the minimum value MIN and the maximum value MAX is used.
Is divided into four equal parts, and is quantized into two bits according to the level range (having a quantization step width) to which the value (level) after removal of the minimum value of each pixel belongs. In the decoding device, as indicated by x, decoding is performed to a representative level that is the median of each level range.

In the case of edge matching, as shown in FIG. 5B, the values of the pixels P1 to P5 are encoded such that the values of the minimum value MIN and the maximum value MAX themselves become the representative level.

[Problems to be solved by the invention]

In the above-described ADRC, the decoding device obtains a decoded signal indicated by a broken line. However, since the true value of each pixel can exist within the width of each level range, always setting the median value as the decoded value is a case where quantization distortion, which is the difference between the true value and the decoded value, becomes large. Occurs.

Accordingly, the present invention provides an image data decoding apparatus and a decoding method capable of reducing distortion due to quantization by adaptively taking the most probable decoded value, judging from the situation around the target pixel to be decoded. Is to do.

[Means for solving the problem]

An image data decoding apparatus for receiving encoded image data that has been compressed and encoded by quantization and transmitted, and decoding the image data from the encoded image data, comprising: Extracting means for extracting peripherally coded data spatially and / or temporally adjacent to, and comparing the peripherally coded data extracted by the extracting means with the coded data of interest, and in a predetermined direction, Detecting means for detecting a change point at which the same state of the coded data and the coded data of interest becomes n times consecutively and then becomes the same; and correcting the correction value according to the slope of the change point to the code of interest before and after the change point. An image data decoding apparatus characterized in that decoded data is obtained based on an addition result by adding to decoded data.

The invention of claim 2 is a decoding method in which decoded data is obtained by adding the correction value to the encoded data of interest.

[Action]

Pixel data a0 spatially adjacent to the coded data x of interest
Aa8 or temporally adjacent pixel data a9 is supplied to the filtering means. The most likely value x 'of the encoded data x is obtained from the filtering means. By decoding the encoded data x 'corrected by this filtering, a decoded image with less quantization distortion can be obtained.

〔Example〕

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 shows the configuration of a decoding apparatus according to this embodiment, in which a transmission path (for example, digital
The data received via the VTR recording / playback process) is supplied.

On the transmitting side, the image signal is compression-coded, for example, ADRC-coded. The ADRC is composed of, for example, 2 frames (4 lines × 4 pixels) (8 lines × 8 pixels) formed by subdividing one frame.
For each dimension block, detect the maximum value MAX and minimum value MIN,
A dynamic range DR of (MAX−MIN = DR) is detected. For example, in the case of 4-bit fixed-length non-edge matching, the dynamic range DR is divided into 16 level ranges each having a quantization step width Δ, and a 4-bit code signal is generated according to the value of each pixel. . From such an encoding device, a dynamic range DR, a minimum value MIN, and a code signal DT for each pixel are generated for each block.

The present invention can be applied not only to the fixed-length ADRC but also to a variable-length ADRC in which buffering for controlling the transmission data amount during a predetermined period such as one frame to be constant is performed. In this case, in the same manner as previously proposed, a set of a plurality of thresholds that can make the amount of transmission data not exceed the target value is determined according to the frequency of occurrence of the dynamic range DR, The number of bits allocated to the block is determined from the set of thresholds and the dynamic range DR. A dynamic range DR, a minimum value MIN, a code signal DT of each pixel, and information for identifying a set of threshold values for each block are transmitted.

Further, the present invention relates to DCT (Discrete cosine) other than ADRC
transform), and compression coding such as DPCM.

The quantization of an n-bit ADRC can be represented by the following equation.

256 ÷ 2 ⁿ = Δ: quantization step width ai = Li ÷ Δ (rounded down to an integer) ai: code signal of pixel i Li: original level of pixel i In frame decomposition circuit 2, additional code AD (ADRC:
Dynamic range DR and minimum value MIN) and code signal DT
And are separated. The code signal DT is supplied to the peripheral pixel extraction circuit 3 and the comparison circuit 4. As shown in FIG. 2, code signals a1 to a8 adjacent to each other in the up, down, left, right, and diagonal directions around the code signal x of the target pixel and the code at the same position as the target pixel one frame earlier in time. The signal a9 is extracted by the peripheral pixel extraction circuit 3. The output of the peripheral pixel extraction circuit 3 is supplied to the comparison circuit 4 and the selection circuit 5. In the comparison circuit 4, the values of the code signals a1 to a8 of the peripheral pixels and the code signal x of the target pixel are compared, and the result of the comparison is held.

The code signal x of the pixel of interest is supplied to the correction circuit 6, and the output signal x 'of the correction circuit 6 is supplied to the decoder 7. The dynamic range DR from the frame decomposition circuit 2 is supplied to the decoder 7, and the corrected code signal x ′ is decoded by the decoder 7. Although not shown, the decoding level Li 'taken out at the output terminal 8 is returned to the scanning order by the block decomposition circuit. The correction circuit 6 is supplied with a code signal (one of a1 to a9) of the peripheral pixel selected by the selection circuit 5. The correction circuit 6 generates a correction code signal corresponding to the value of the selected peripheral pixel. In the decoder 7, the code signal x 'is decoded as described below, and the decoded value Li' is taken out to the output terminal 8.

Li ′ = Δ × (x + c + 0.5) (rounded off) c: Correction value The comparison circuit 4 monitors the presence or absence of a change in the code signal three-dimensionally. Detection of a horizontal change made between the code signal a4 or a5 and the code signal x, the code signal a2 or
Detection of vertical changes made between a7 and code signal x, detection of oblique changes made between code signals a1, a3, a6 or a8 and code signal x, code signal a9 and code signal Detection of a change in the time direction with respect to x is performed. The presence or absence of these detected changes in each direction is held. Then, when a change is detected after the same code signal has been repeated a predetermined number of times in a certain direction, a correction value c corresponding to the slope of the change with respect to the code signal x of the target pixel is detected.
Is added. When there is no predetermined number of consecutive times or when there is no change, the correction value c is 0.

Priority is set for the change in each direction, and detection in the horizontal direction is given the highest priority. Hereinafter, detection in the vertical direction, detection in the diagonal direction, and detection in the time direction are given priority. FIG. 3 shows the detection of a horizontal change, ie,
As a value of the code signal DT, 1 (a code of 01 in the case of 2 bits) is repeated three times, and then a state is shown in which the value has changed to 0. The change width is 1, and the code signal before and after the change is corrected by the correction value c of (0 <c <0.5). For example, if (c = 0.2), the code signal x of the target pixel before the change point is set to (x−0.2), and 0.2 is added to the next code signal. The code signal x 'to which the correction value c has been added is decoded. Therefore, the decoded level of the code signal of the target pixel becomes smaller than 1, and the decoded level of the next code signal becomes larger than 0. As a result, quantization distortion of the decoded output can be reduced.

In the vertical direction, the diagonal direction, and the time direction other than the horizontal direction, a change is similarly detected. The number of continuous times, which is a condition under which the correction is made, is set to about two to three times. In addition, in the example of FIG. 3, since the width of the change is 1, (0 <
Although the correction value c of c <0.5) was used, the value of the correction value c is also increased corresponding to the larger width of the change.

The above-described selection circuit 5 and correction circuit 6 constitute a three-dimensional (spatiotemporal) filter 9A for the pixel of interest.

Here, since ADRC is block coding, even if the code signals DT included in different blocks have the same code, the decoded values are different. Therefore, special processing is required for the target pixel near the block boundary.

In FIG. 4A, a broken line indicates a boundary between blocks, and a code signal of a plurality of pixels that is in contact with the boundary is locally decoded. Δ indicates a decoded pixel. When the difference between the decoded value of the target pixel and the decoded values of the peripheral pixels of another block is equal to or smaller than 1/2 of the quantization step width Δ of the block of the target pixel, it is considered that the code signal is continuous, and 1 / If greater than two, the code signal is considered to have changed.

The other boundary processing is to locally decode the code signal near the boundary of the block outside the block of interest indicated by □ in FIG. 4B, and to determine the decoded value as the quantization step width Δ and the minimum value MIN of the block of interest. And re-encode. By performing the decoding and encoding processes, the code signals of the other blocks can be treated in the same manner as the code signals in the target block.

〔The invention's effect〕

According to the present invention, when decoding a compression-encoded code signal, a more probable code is obtained by referring to a spatially or temporally close code signal without simply using the median value of the quantization step width Δ as a decoded value. Since the signal is corrected to a signal and the corrected code signal is decoded, quantization distortion can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is a schematic diagram used for explaining peripheral pixels, FIG. 3 is a schematic diagram used for explaining a code signal correction, and FIG. FIG. 5 is a schematic diagram used for explaining the boundary processing, and FIG. 5 is a schematic diagram used for explaining the ADRC. Explanation of main symbols in the drawing 1: input terminal for received data, 3: peripheral pixel extraction circuit, 7: decoder, 9A, 9B, 9C: filter.

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-62-214788 (JP, A) JP-A-1-200885 (JP, A) JP-A-63-104586 (JP, A) (58) Survey Field (Int.Cl. ⁶ , DB name) H04N 7/24-7/68 H04N 1/41-1/419

Claims (2)

(57) [Claims] An image data decoding apparatus for receiving encoded image data which has been compressed and encoded by quantization and transmitted and decodes the image data from the encoded image data. Extracting means for extracting neighboring encoded data spatially and / or temporally adjacent thereto, and comparing the peripheral encoded data extracted by the extracting means with the encoded data of interest, and in a predetermined direction, Detecting means for detecting a change point at which the same state of the peripheral coded data and the coded data of interest becomes n times successively non-identical; and correcting values corresponding to the slope of the change point before and after the change point Wherein the decoded data is added to the encoded data of interest to obtain decoded data based on the addition result. 2. An image data decoding apparatus for receiving encoded image data which has been compressed and encoded by quantization and transmitted, and for decoding image data from the encoded image data, comprising: Extracting peripherally coded data spatially and / or temporally adjacent to the extracted coded data; comparing the extracted coded data with the coded data of interest; Detecting a change point in which the same state as the coded data of interest becomes n times consecutively and then non-identical; and correcting the correction value according to the gradient of the change point with the coded data of interest before and after the change point. And obtaining decoded data based on the addition result.