JPH03295378A - Picture decoding device - Google Patents

Abstract

PURPOSE:To realize the reproduction picture of high quality by intra-frame-encoding the low pass component of an input picture signal being most important in a picture and adaptively decoding an encoding picture signal outputted by means of inter-frame- encoding other band pass components in accordance with respective band pass components. CONSTITUTION:An input separation means 12 separates the inputted encoding picture signal into the low band component and the other band components and they are respectively given to an intraframe decoding means 17 and inter-frame decoding means 18, 19 and 110. The intra-frame decoding means 17 interframe-decodes the low pass component of the separated encoding picture signal and the inter-frame decoding means 18, 19 and 110 inter-frame-decode the band components except for the low pass component of the encoding picture signal. Then, a synthesis means 115 synthesizes an output signals from the intra-frame-decoding means 17 and output signal from the inter-frame-decoding means 18, 19 and 110 and obtains a reproduction picture signal. Thus, the influence of the abolishment of a cell can be ceased within a frame by applying the intra-frame encoding/decoding to the low pass component being the important component on the picture, and the deterioration of picture quality can be kept minimum.

Description

Detailed Description of the Invention [Industrial Application Field] The present invention provides an image encoding device that performs intra-frame encoding on the low-frequency component of an input image signal, which is the most important component of an image, and The present invention relates to an image decoding device that performs interframe encoding on components and decodes an output encoded image signal.

[Prior Art] Fig. 2 shows a schematic diagram of an image packet encoding device including a conventional image encoding device and an image packet decoding device including a conventional image decoding device (rDCT by Nomura et al. J, 1988 Image Coding Symposium, pp. 85-86).

Figure 2 (A> shows the encoding device. Figure 2 (A)
In the image signal S input from the input terminal 21,
After each frame is divided into blocks consisting of several nXn pixels, a discrete cosine transformer (DCT transformer) 22
Discrete cosine transform is performed block by block. From the obtained discrete cosine coefficient, the subtracter 23 subtracts the discrete cosine coefficient of the corresponding block one frame before, which is stored in the frame memory 24, to obtain a difference value d. The difference value d is converted into a code i by a quantizer 25 and output to a scan unit 26 .

Further, this code i is sent to an inverse quantizer 27 having an inverse characteristic to that of the quantizer 25, and after being inversely quantized, an adder 20 uses the discrete value of the previous frame used to obtain the difference value d. The coefficients are added to the cosine coefficients, converted into locally reproduced discrete cosine coefficients, and output to the frame memory 24. This locally recovered discrete cosine coefficient is used for encoding the next frame.

With such a processing system, a code i obtained by predictively coding the discrete cosine coefficients is obtained.

As shown in FIG. 3, the scanner 26 sequentially scans the input code i by repeatedly reciprocating in the diagonal direction block by block. Note that FIG. 3 shows the scanning order of the code i of an 8×8 pixel block. At this time, the code i is hierarchically divided by dividing the scan into a low frequency component i[ and a high frequency component iH. The hierarchical code components iL and iH are respectively sent to the corresponding variable length encoders 28 and 29 and subjected to run-length Huffman encoding, and then sent to the packet assembling unit 210 into priority cells and non-priority cells, respectively. Layered packets of the same type (
cells) and sent out onto a transmission path.

That is, information indicating priority/non-priority is added to each cell as header information. A hierarchical structure with priorities is introduced in the network, and when cells are discarded during network congestion, priority cells containing low-frequency elements important for image signals are not discarded, but non-priority cells are discarded first. Image quality deterioration due to disposal is kept to a minimum.

FIG. 2(B) shows the configuration of the decoding device.

In FIG. 2(B), cells sent from the encoding device are decomposed by a packet decomposition unit 211 and divided into encoded image data and parameters necessary for decoding. Encoded image data that constituted priority cells and non-priority cells (
The data for the low-band code i [and the data for the high-band code iH) are respectively processed by the corresponding variable length decoding unit 212
．． 2] Run 1 at -3. /'Huffman decoded. After that, the data synthesis unit 214 reconstructs the image codes (low-band code i[, high-band code iH) that were divided into priority/non-priority, and then performs the reverse operation of scanning in the encoding device for each block. The reproduction code iS is obtained.

The reproduced code is is subjected to inverse quantization, which is the inverse processing of the quantizer 25, in an inverse quantizer 215, and then added to the reproduced discrete cosine coefficient of the corresponding block one frame before, which is stored in the frame memory 216. 21-7, and converted into reproduced discrete cosine coefficients. The inverse DCT transformer 218 performs inverse discrete cosine transform on the reproduced discrete cosine coefficients, and the reproduced image signal SS obtained thereby is output from the output terminal 219 and is also used for processing the next frame. is stored in the frame memory 216.

[Section B to be invented or solved] By the way, asynchronous transfer mode (AT) is used as a cell transmission network.
M>Net is being considered. It is known that a signal deterioration factor in this asynchronous transfer mode network is cell discard due to network interference.

In the image packet encoding/decoding device described above, priority can be assigned to each cell based on the importance of data within the cell, and the network can also be divided into hierarchical lines such as priority channels and non-priority channels. This is intended to suppress image quality deterioration by protecting cells with high priority from being discarded, but the control configuration for this purpose has become complicated.

Furthermore, in some cases, due to the introduction of a priority concept, the priority channel may run out of capacity, and the priority cell may be discarded. At this time, the influence on the reproduced image quality becomes very large.

Furthermore, if an error occurs in the transmission path, such as cell discard,
There is also a problem that the contents of the frame memory 24 used for interframe encoding and the frame memory 216 used for interframe decoding become inconsistent, making it impossible to decode appropriately and this state propagating for a long time.

Therefore, an image encoding device that does not assume the use of a layered network and that encodes in a way that prevents large negative effects from propagating for a long time even if an error occurs in the transmission system is developed. Separately proposed by the applicant.

The present invention aims to provide an image decoding device that can appropriately decode encoded image signals from such an image encoding device.

[Means for Solving the Problems] According to the present invention, an image encoding device performs intra-frame encoding on the low-frequency component of an input image signal, which is the most important in an image, and performs frame encoding on other band components. The present invention relates to an image decoding device that decodes an encoded image signal outputted by performing inter-encoding, and consists of the following components.

That is, an input separating means that separates an input encoded image signal into a low frequency component and other band components; an intraframe decoding means that intraframe decodes the low frequency component of the separated encoded image signal; An interframe decoding means for interframe decoding of band components other than the low frequency components of the encoded image signal, and a reproduced image by combining the output signal from the intraframe decoding means and the output signal from the interframe decoding means. and a synthesis means for obtaining the signal.

Here, the intraframe decoding means may have a configuration that combines intraframe decoding and other decoding depending on the intraframe encoding method, and the interframe decoding means may have a configuration that combines intraframe decoding and other decoding. Depending on the encoding method, the configuration may be a combination of interframe decoding and other decoding.

Note that it is preferable to provide a leakage section for performing a leakage operation in a loop for forming a predicted signal from a decoded signal in the interframe decoding means.

[Operation] In the present invention, the input separation means separates the input encoded image signal into a low frequency component and other band components, and supplies them to the intraframe decoding means and the interframe decoding means, respectively. The intraframe decoding means performs intraframe decoding on the low frequency components of the separated encoded image signal, and the interframe decoding means performs interframe decoding on the band components other than the low frequency components of the encoded image signal. The combining means combines the output signal from the intra-frame decoding means and the output signal from the inter-frame decoding means to obtain a reproduced image signal.

Here, the intra-frame encoding/decoding method is adopted for the low-frequency components that are important in the image, because even if there is a drop in the encoded image signal in the transmission system, the negative effects will be avoided. 1
This is because it stays within the frame.

Note that if a leakage section that performs a leakage operation is provided in the loop for forming a predicted signal from the decoded signal in the interframe decoding means, this leakage operation will prevent image deterioration due to information loss over time. You will be able to return with it.

[Embodiment] Prior to describing an image packet decoding device incorporating an image decoding device according to an embodiment of the present invention, a corresponding image packet decoding device will be described.

FIG. 4 shows the configuration of this image packet encoding device, and FIGS. 5 and 6 show detailed configurations of the intraframe encoding section and interframe encoding section, respectively.

In FIG. 4, the image signal inputted from the input terminal 41 is a QMF (Qua
(Mirror Filter) section 42
Accordingly, each frame is divided into two bands, a high frequency component and a low frequency component, in each of the horizontal and vertical directions, and the frame as a whole is divided into four bands of signal components in a two-dimensional spatial frequency domain.

The image signal components of each of the four bands obtained by the division are processed by corresponding decimation units (thinning units) 43, 44, .
45.46, the samples are thinned out by 172 in each of the horizontal and vertical directions, and the image signal components SLL, SLU, S, and 5U are thinned out to a total of 1/4.
U (subscripts are in the horizontal and vertical directions, L indicates a low frequency component, and U indicates a high frequency component) is obtained.

The image signal component SLL, which is a low frequency component in both the horizontal and vertical directions, is supplied to an intraframe encoding unit 47 whose detailed configuration is shown in FIG.
8 and 8 are respectively applied to corresponding motion compensated interframe encoders 48, 49, and 410 whose detailed configuration is shown in FIG. Each encoding unit 47, 48, 49, 410 encodes the given image signal component, and its encoded value iLL, 1
The variable length encoder 4 corresponding to f-U, ill, 1u (J
11.412.413.414.

Each variable length encoding unit 411.412.413.414 is
For example, the input code values iLL, i group, i Kano, ill are run-length/Huffman encoded and MUX unit (multiplex unit) 415
Output 4.

The MUX unit 415 multiplexes the input code into one series of signals and provides it to the packet assembly unit 416, and sends it to the bucket 1.
Assemble this into cells of a certain length, for example A
Send to the TM transmission network. At this time, the type of information in the cell (which band the signal is in, etc.) and the parameters necessary for decoding (block average value, motion vector MVLυ, MVKo, MVUU, etc.) are added as header information. .

Here, the reason why intraframe coding is used only for the image signal component SLL, which is a low frequency component in both the horizontal and vertical directions, is that the image signal component S1. ,
Since L is an important component for images, it is undesirable for this image signal component SLL to be unstable for a long time due to cell discards, etc. By adopting intra-frame coding, the effects of cell discards are suppressed within the frame. This is a lick that is kept to a minimum and does not affect the next frame.

Frame, 4 tongue.

Next, the intraframe encoding unit 47 that encodes the image signal component SLL, which is a low frequency component in both the horizontal and vertical directions, will be described.

In FIG. 5, an input image signal component SLL is provided to a block dividing section 51. In FIG. The block division section 51 divides each frame into several blocks each consisting of nXn pixels, and calculates the image signal components of each block by means of an average value calculation section 52.
and is given to the subtracter 53. The average value calculation unit 52 calculates the intra-block average @h and provides it to the subtracter 53. Subtractor 53
provides a clear value dLL obtained by subtracting the intra-block average value from the image signal component to an orthogonal transform unit 54, such as a discrete cosine transformer. The orthogonal transform unit 54 converts the difference value d
ll- is orthogonally transformed and provided to the quantizer 55, and the quantizer 5
5 quantizes this orthogonal transform coefficient to obtain an output code iLL.

Note that the intra-block average value obtained by the average value calculation unit 52 is transmitted to the image packet decoding device as a parameter necessary for the decoding process, as described above.

Next, image signal components StU other than the image signal component SLL,
Interframe encoding unit 48.4 that encodes S elbow and SUU
9.410 will be explained.

In FIG. 6, the input image signal component S [υ, S or 8
The dish is provided to the block dividing section 61. The block dividing unit 61- divides the input image signal components into nXn frames.
The subtractor 62 divides the blocks into several blocks each consisting of pixels.
The subtracter 62 subtracts the predicted value of this signal component given from the motion compensation unit 63 from this (S) component, and converts the difference value dLU, d(几 or dlJIJ) into an orthogonal The orthogonal transform unit 64 orthogonally transforms the difference values d, d, or dUU for each block, and supplies them to the quantizer 65, which quantizes the orthogonal transform coefficients. , the obtained code i group, ill,
Or output ill.

The code between itυ, NJL, or 5 is given to an inverse quantization unit 66, where the inverse processing of the quantization unit 65 is performed, and it is converted into orthogonal transform coefficients for local reproduction. 67. The inverse orthogonal transform section 67 performs the inverse processing of the orthogonal transform section 64 on the locally reproduced orthogonal transform coefficients to obtain a locally reproduced difference value and supplies it to the adder 68 . The adder 68 receives the motion-compensated local reproduction orthogonal transform coefficients (predicted values) of one frame before, which are output from the motion compensation unit 63.
The adder 68 adds the locally reproduced orthogonal transform coefficient and this predicted value to obtain a locally reproduced image signal component, and supplies the locally reproduced image signal component to the leakage section 69. The leakage unit 6 multiplies the locally reproduced image signal component by a leakage coefficient α (less than 1), and then supplies it to the frame memory 610 for storage.

The image signal components stored in the frame memory 610 are used for processing the next frame, and are used in the motion compensation unit 6.
3 and is then motion compensated and given to the subtracter 62 as a predicted value as described above.

Note that the motion compensation unit 63 is supplied with an image signal component from the block division unit 61, although the signal line is not shown in the figure, and this image signal component and the previous frame stored in the frame memory 610 signal and the motion vector from (
For example, a pair of motion vectors in the horizontal and vertical directions) is detected, and motion compensation is performed based on this motion vector. As mentioned above, this motion vector information MVLU, MVUL, MVUULt, decoding process 4. :8 Transmitted to the image packet decoding device as necessary parameters.

Next to Packet 1, an image packet decoding device incorporating an image decoding device according to an embodiment of the present invention will be described.

FIG. 1 shows the configuration of this image packet decoding device, and FIGS. 7 and 8 show detailed configurations of the intra-frame decoding section and inter-frame decoding section, respectively.

In Fig. 1, cells from the transmission path are packet disassembled by the packet disassembly unit 1.
1 and decomposed. The packet disassembly unit 11 also detects the cell discard position.

The data decomposed by the packet decomposition unit 11 is provided to a DEMUX unit (demultiplexing unit) ]-2.

The DEMUX unit 12 divides this into encoded image information and parameters required for decoding (motion vectors, etc.), and further divides it into signal components of each band and sends them to corresponding variable length decoding units 13, 14, . Give on 15.16.

The variable length decoding unit 13.14.1-5.16 performs the inverse processing of the variable length encoding unit 411.412.413.414 described above to generate reproduced codes YLL, Y[U, Y Kano, YUU(
each i[1-1i 11. i Ul-i LJU4
This correspondence) is obtained and the corresponding decoding unit 17.18.19.1
Give 10. The reproduction code YLL relating to low frequency image signal components in both the horizontal and vertical directions is given to the intra-frame decoding unit 17 whose detailed configuration is shown in FIG. The signal is applied to interframe decoding units 18, 19, and 110 whose detailed configuration is shown in FIG.

Each of the decoding units 17.1-8.19.110 decodes the reproduction codes YLL, YLU, YUL, and yuu, as described later, and reproduces the reproduction image signal components S' [[, s ' tu, s', and s'uu are respectively provided to corresponding interpolation units (sample interpolation units) 111, 112, 113, ]14.

Each interpolation part 111.112.113.1
14 are given reproduced image signal components S'LL, respectively.
, S' LU, S' Kano, and s'uu are upsampled by a total of 4 times in both the horizontal and vertical directions to 0M.
Give 1 part to 115. The 0M1 unit 115 filter-links each of the upsampled image signal components, then synthesizes the four-band signals and outputs a reproduced image signal from the output terminal 1-16.

Next, details of the above-mentioned intra-frame decoding unit 17 that decodes the reproduction code Y)-[- will be explained.

In FIG. 7, the reproduction code YLL is applied to a switch circuit 71, and is selected by this switch circuit 71 and applied to an inverse quantizer 72 except when a cell is discarded. This reproduction code Y
LL is subjected to the inverse processing of the quantization unit 55 (FIG. 5) in the inverse quantization unit 72, converted into reproduced orthogonal transform coefficients, and provided to the inverse orthogonal transform unit 73. The inverse orthogonal transform section 73 performs the inverse processing of the orthogonal transform section 54 on the reproduced orthogonal transform coefficients to obtain a reproduced difference value d'[[ and provides it to the adder 74. The adder 74 is given the average value outputted from the average value calculation section 52 of the encoding device and given to the decoding device, and the adder 74 calculates the reproduced difference value dll and this average value. is added to obtain the reproduced image signal component S'[[, and this reproduced image signal component S'[1- is added to the interpolation unit 111
Output to.

In this way, intraframe decoding processing during normal reception is performed.

When cell discard is detected, the above-mentioned switch circuit 71 is switched to the other input terminal side, and the block information (reproduction code) lost due to cell discard is set to zero. At this time,
Since the difference value given from the inverse orthogonal transformer 73 to the adder 74 also becomes 0, the average value is output as is as the compensated reproduced image signal component S'LL, ensuring the minimum reproduced image quality.

Tongue 2 of frame a'1 Next, details of the interframe decoding unit 18, 19 or 110 for the reproduced code YLU, YY, or YUu will be explained.

In FIG. 8, the reproduced code ytυ, Yko, or YUU is applied to a switch circuit fiJ81, and is selected by this switch circuit 81 and applied to an inverse quantization unit 82 except when a cell is discarded. The reproduced codes YLU, YY, or YUU are subjected to the inverse processing of the quantizer 65 (FIG. 6) in the inverse quantizer 82, converted into reproduced orthogonal transform coefficients, and provided to the inverse orthogonal transform unit 83. The inverse orthogonal transform unit 83 performs the inverse processing of the orthogonal transform unit 64 on the reproduced orthogonal transform coefficients to obtain a reproduced difference value d.
'LU, d' or d'UU is obtained and given to the adder 84. The adder 84 receives one frame before the motion compensation unit 85 based on the motion vector information MVLU, MVK, or MVUU output from the motion compensation unit 63 of the encoding device and given to the decoding device. Regenerated orthogonal transform coefficients (
predicted value) is given. The adder 84 reproduces the reproduction difference value d
By adding 'LU, d' or d'UU and this predicted value, reproduced image signal components s'tu, S' or S'UU are obtained, and the reproduced image signal components s'tu, s' are Or S'U
LI to interpolation part 112, 113 or 11
Output to 4.

The reproduced image signal components S', S', or s'uu are provided to the leakage section 86. The leakage section 86 multiplies the reproduced image signal components S', S', or S' by a leak coefficient α, which is equal to the leak coefficient in the encoding device, and then supplies the resultant to the frame memory 87 for storage.

The signal components stored in the frame memory 87 are used for processing the next frame, are motion compensated via the motion compensator 85, and are provided as predicted values to the adder 84 as described above.

In this way, normal interframe decoding processing is performed.

When cell discard is detected, the above-mentioned switch circuit B81 is switched to the other input terminal side, and the block information (reproduction code) lost due to cell discard is set to zero. At this time,
Since the difference value given from the inverse orthogonal transformer 83 to the adder 84 is also 0, the reproduced image signal components S', S', which are the reproduced image signal components of one frame before via the motion compensator 85 are compensated as they are. It is output as Kano or S'tJtJ.

By the way, in general, when an error occurs during transmission in interframe encoding, the frame memory 610 in the encoder and the frame memory 87 in the decoder become unsynchronized.
There is a problem with errors or propagation over time. Therefore, in this embodiment, the leakage portion 69, 86 mentioned above is
By adding , it is possible to synchronize the frame memories of the encoder and decoder over time. Is it possible that adding the leakage section 69 and S6 increases the prediction error and reduces the coding efficiency?The leakage section is provided because the image signal components 5LIJ and SU in bands other than the baseband are or 5 (JI
J, and the power is relatively small, so there is no practical problem.

Since intra-frame encoding/decoding is applied to the low-frequency component, which is an important component in the image, the effect of cell discard can be contained within the frame and will not affect the next frame. It is possible to prevent this from occurring. Furthermore, since the average value is used to compensate for frames in which cells are discarded, deterioration in image quality can be kept to a minimum.

By adopting such an encoding/decoding method, there is no need to hierarchize each cell, and there is no need to perform priority control on the transmission path.

The form of transmission from the image encoding device to the image decoding device is not limited to packet transmission. For example, the transmission form may be recording and reproducing on a recording medium.

Various methods can be applied to transmit parameters such as motion vectors and block average values.

Further, as the band division filter, something other than QMF may be applied, and a one-dimensional rather than two-dimensional band division filter may be applied.

In addition to the above-described method of calculating the average value, the intra-frame encoding method may also be a method that uses line correlation or a method that uses correlation with adjacent pixels. According to the invention, an image encoding device performs intra-frame encoding on the low-frequency component of an input image signal, which is the most important in an image, and performs inter-frame encoding on other band components and outputs the resultant image. Since the encoded image signal is appropriately decoded according to each band component, a high-quality reproduced image can be obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an image packet decoding device according to an embodiment of the present invention, FIG. 2 is a block diagram showing a conventional image packet encoding device/decoding device, and FIG. 3 is a block diagram showing a conventional image packet encoding device/decoding device. 4 is a block diagram showing an image packet encoding device corresponding to the image packet decoding device according to the above embodiment, and FIG. 5 is an explanatory diagram showing a method for extracting high frequency components and high frequency components. FIG. 6 is a block diagram showing the interframe coding section of FIG. 4, and FIG. FIG. 8 is a block diagram showing the interframe decoding section. 11... Packet decomposition section, 12... DEMUX section,
13 to 1-6... variable length decoding section, 17...
Intraframe decoding unit, 18.19.110...Interframe decoding unit, 115...QMF unit, 86...Leakage unit. 477 frame 6 encoding unit 7' lock diagram of intra-frame encoding Figure 5 7' lock diagram of frame 6 encoding unit Figure 6

Claims (2)

[Claims] (1) The image encoding device performs intraframe encoding on the low frequency component of the input image signal, which is the most important part of the image,
In an image decoding device that performs interframe coding on other band components and decodes the output coded image signal, an input that separates the input coded image signal into low band components and other band components. Separation means; Intraframe decoding means for intraframe decoding of low frequency components of the separated encoded image signal; and interframe decoding for interframe decoding of band components other than the low frequency components of the encoded image signal. An image decoding device comprising: means for combining an output signal from the intra-frame decoding means and an output signal from the inter-frame decoding means to obtain a reproduced image signal. (2) The image decoding device according to claim 1, wherein a leakage section for performing a leakage operation is provided in a loop for forming a predicted signal from a decoded signal in the interframe decoding means. .