JPH04341066A - Picture coder and decoder - Google Patents

Abstract

PURPOSE:To suppress the effect of an error in coefficient information and to enhance the reliability of picture reproduction by dividing one pattern into plural blocks and adding coefficient information relating to a quantity step of coded picture information to each block. CONSTITUTION:In the picture coder in which a conversion data obtained by converting picture information into a frequency area is quantized and the quantized conversion data is subject to variable length coding, a re-SYNC multiplexer 9 divides one pattern of a picture element inputted from a variable coding circuit (VLC) 8 into plural areas and coefficient information relating to a quantization step of the coded picture information is added to each picture signal for each split area and the result is outputted to an ECC coding circuit 10. Through the configuration above, even when an error takes place in the coefficient information on a transmission line or the like, the effect of the error of the coefficient information is suppressed very small by replacing the information data properly thereby improving the reliability of picture reproduction.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding apparatus for quantizing image information and variable length encoding of the quantized transformed data, and a decoding apparatus for variable length decoding of the encoded data.

0002

[Prior Art] In recent years, adaptive DCT (discrete cosine transform) encoding has attracted attention as a color image signal encoding method, and Cable, which was established as an international standardization organization for this type of encoding, A certain JPEG (Joint
Photographic Expert Gro
The DCT encoding method is also adopted as the encoding method in UP).

[0003] The basic system of this type of encoding system will be briefly explained below.

FIG. 5 is a block diagram for explaining a schematic configuration example of a conventional encoding device using DCT transform, and FIG. 7 (A
) to (D) are diagrams for explaining the processing of the encoding method shown in FIG. 5. Reference numeral 46 denotes an input terminal for a digital image signal to be encoded, into which a raster scan digital image signal is input.

The image signal input to the terminal 46 is 8×8
It is input to the blocking circuit 47, where it is two-dimensionally (8
×8) The image is divided into pixel blocks each consisting of pixels, and sent to the subsequent stage in units of pixel blocks.

48 performs DCT transformation on the image signal from this blocking circuit 47, and converts it into (8×8) in the frequency domain.
This is a DCT conversion circuit that outputs a data matrix. That is, image data D11 to D88 as shown in FIG. 7(A)
The pixel block consisting of
It is converted into a data matrix consisting of X11 to X88 as shown in FIG.

Here, X11 represents a direct current (DC) component in the horizontal and vertical directions of the pixel block, that is, the average value of this pixel block. This X11~X8
8 is generally represented by Xij, the larger i is, the higher the frequency is in the vertical direction, and the larger j is, the higher the frequency is in the horizontal direction.

The data matrix output from the DCT conversion circuit 48 is input to a linear quantization circuit 49. on the other hand,
The quantization matrix generation circuit 57 generates each DCT coefficient X11.
Quantization matrices W11 to W88 (shown in FIG. 7C) indicating weighting of the quantization step size to X88 are generated, and the generation circuit 56 generates a coefficient C.

[0009] Also, an AC component G1 other than the DC component G11
2 to G88, the quantized output is as described above.
Encoding is performed while zigzag scanning from low frequency components to high frequency components, and significant coefficients whose quantized output is not 0 are classified into groups based on their values, and are sandwiched between the group identification number and the immediately preceding significant coefficient. The quantization output is combined with the run length of the number of invalid coefficients of 0 and subjected to Huffman encoding, and then an equal length code is added to indicate which value in the group it is.

[0010] Generally, since high frequency components in the diagonal direction of an image have a low probability of occurrence, it is expected that the latter half of Gij after zigzag scanning will often be all zero. Therefore, the variable length code obtained in this way can be expected to have a very high compression rate, and assuming an average compression rate of about a fraction,
Images can be restored with almost no image quality deterioration.

On the other hand, the transmission capacity per unit time of a transmission path is generally determined, and when one screen must be transmitted every predetermined period, such as when transmitting a moving image, the output code It is desired that the number of bits be fixed for each screen or pixel block.

Here, if the coefficient C mentioned above is made large, Gi
The probability that j becomes 0 increases, and the total number of bits NB of encoded data decreases. The relationship between the coefficient C and the total number of bits NB varies depending on the image, but in any case it is a simple decreasing function, and it is known that for an average image it becomes a logarithmic curve as shown in FIG. 7(D).

[0013] Therefore, a method of predicting the coefficient C0 to obtain the desired total number of bits NB0 has been proposed by the above-mentioned JPEG and the like. That is, a certain coefficient C1 is first encoded, and the total number of bits NB1 of the code thus obtained is determined. The predicted value of C0 is calculated based on this NB1 and C1. This calculation is performed so that the logarithmic curve shown in FIG.
, NB1).

By repeating the above operations, the coefficient C0 is determined for each frame. The coefficient information indicating the coefficient C0 and the compressed image data (encoded code) output from the VLC 51 are outputted by a frame multiplexer 52 with one coefficient information added to the compressed image data for one frame.

[0015] These data are error correcting codes (ECC)
The signal is error-corrected encoded in the encoding circuit 53, modulated in a form suitable for the characteristics of the transmission path in the modulation circuit 54, and output as a transmission signal (transmission data string) from the output terminal 56 to the transmission path.

Next, referring to FIG. 6, the outline of a decoding device corresponding to the encoding device of FIG. 5 will be briefly explained.

FIG. 6 is a block diagram for explaining a schematic configuration example of a conventional decoding device.

In FIG. 6, a transmission signal on a transmission path is inputted from an input terminal 59, demodulated by a demodulation circuit 60, and converted into an EC signal.
The signal is input to the C decoding circuit 61.

The ECC decoding circuit 61 corrects code errors in the transmission signal according to a predetermined algorithm. Said EC
The transmission signal subjected to error correction by the C decoding circuit 61 is transmitted to the coefficient decoder 52, where it is distinguished into image information and coefficient information.

The coefficient information identified by the coefficient decoder 62 is sent as is to the multiplier 70.

The image information identified by the coefficient decoder 62 is input to a variable length decoding circuit (IVLC) 63 where it is variable length decoded and subjected to subsequent zigzag scanning 54, linear inverse quantization 65 and inverse DCT (IDCT). ) The processing operations up to conversion 66 are the zigzag scan 5 described in the encoding device of FIG.
0, linear quantization 49 and DCT transformation 48 are performed.

The image information subjected to the above processing is subjected to digital-to-analog conversion in a D/A converter 67 and is supplied to an output terminal 68. The signal is input from the output terminal 68 to an image display means such as a monitor.

[0023]

As described above, in the conventional technology, one coefficient information for quantization is added to one frame of image information.

Since the coefficient information determines the quantization characteristics of the image, it is a very important parameter that affects the quality of the image. Therefore, if an error occurs in the coefficient information on the transmission path, for example, one frame worth of information will be lost. There was a problem in that the quantization characteristics of the image were destroyed, resulting in a visually very unsightly image.

In light of the above-mentioned background, it is an object of the present invention to provide an image encoding device and a decoding device that solve the conventional problems and improve the reliability of image reproduction.

[0026]

[Means for Solving the Problems] An image encoding device according to the present invention quantizes transformed data obtained by transforming image information into a frequency domain, and variable-length encodes the quantized transformed data. The encoding device is characterized in that one frame is divided into a plurality of blocks, and coefficient information related to the quantization step determined for each screen is added to each block.

[0027] The image decoding device also quantizes the transform data obtained by converting the image information into the frequency domain, variable-length encodes the quantized transform data, and divides one frame into a plurality of blocks. An image decoding device that adds coefficient information related to the quantization step to the encoded transform data for each block, and performs variable length encoding on the variable length encoded data, wherein the coefficient information is added for each block. The present invention is characterized by comprising a detection means for detecting an error in information, and a replacement means for replacing coefficient information of a block detected by the detection means with coefficient information of another block.

[0028]

[Operation] According to the above invention, even when an error occurs in the coefficient information on a transmission path, the influence of the error in the coefficient information can be minimized by replacing the information data appropriately, and image reproduction is reliable. can improve sexual performance.

[0029]

[Examples] Examples of the present invention will be described below.

FIG. 1 is a block diagram of an image encoding device of an image transmission system that is an embodiment of the present invention, FIG. 2 is a block diagram of an image decoding device of an image transmission system that is an embodiment of the present invention, and FIG. FIG. 4 is a diagram illustrating one screen according to an embodiment of the present invention. FIG. 4 is a diagram showing a form of a transmission signal of an image transmission system according to an embodiment of the present invention.

In FIG. 1, 1 is an input terminal for an analog image signal, and the image signal inputted from the input terminal 1 is digitized into 8 bits by an A/D converter 2.

The digitized image signal is input to a two-dimensional spatial filter (SPF) 3, and high frequency components that cannot be converted into data are removed during DCT transformation, which will be described later.

The processed digital image signal is input to the blocking circuit 4, where it is two-dimensionally (8×8)
The image is divided into pixel blocks each consisting of pixels, and sent to the subsequent stage in units of pixel blocks.

Next, the image signal from the blocking circuit 4 is D
It is input to the CT conversion circuit 5. That is, the pixel block consisting of image data D11 to D88 as shown in FIG.
is converted into a data matrix consisting of

Here, X11 represents a direct current (DC) component in the horizontal and vertical directions of the pixel block, that is, the average value of this pixel block. This X11~X8
8 is generally represented by Xij, the larger i is, the higher the frequency is in the vertical direction, and the larger j is, the higher the frequency is in the horizontal direction.

The data matrix output from the DCT conversion circuit 5 is input to a linear quantization circuit 6. On the other hand, the quantization matrix generation circuit 13 generates each DCT coefficient X11 to
The coefficient generation circuit 14 generates a coefficient C by generating a quantization matrix W11 to W88 (shown in FIG. These quantization matrices W11 to W88 and coefficient C are supplied to the multiplier 15.
is input to. The multiplier 15 calculates (Wij×C), and the quantization step of the linear quantization circuit 6 is performed by this multiplier 15.
is determined according to the outputs Q11 to Q88.

[0037] Here, the coefficient C is a positive value, and the coefficient C
The image quality and amount of generated data are controlled by the value of .

Actually, in the linear quantization circuit 6, Xij/
Qij is calculated and output. This linear quantization circuit 6
The outputs of G11 to Q88 are assumed to be G11 to Q88. The quantized conversion data G11 to G88 are sent out in order from the DC component by the zigzag scanning circuit 7. That is, from the zigzag scanning circuit 7, G11 to G88 are G11, G12, G21, G31.
,G22,G13,G14,G23,G32,G41...
,G85,G86,G77,G68,G78,G87,
The signals are supplied to the variable length coding circuit (VLC) 8 in the order of G88.

In VLC8, for example, the DC component G1
1, a predicted value is calculated between pixel blocks located in the vicinity, and a prediction error with this predicted value is Huffman encoded.

On the other hand, the transmission capacity per unit time of a transmission path is generally determined, and when one screen must be transmitted every predetermined period, such as when transmitting a moving image, the output code It is desired that the number of bits be fixed for each screen or pixel block.

Here, if the coefficient C mentioned above is made large, Gi
The probability that j becomes 0 increases, and the total number of bits NB of encoded data decreases. The relationship between the coefficient C and the total number of bits NB varies depending on the image, but in any case it is a simple decreasing function, and it is known that for an average image it becomes a logarithmic curve as shown in FIG. 7(D).

[0042] Therefore, a method of predicting the coefficient C0 to obtain the desired total number of bits NB0 has been proposed by the above-mentioned JPEG and the like. That is, a certain coefficient C1 is first encoded, and the total number of bits NB1 of the code thus obtained is determined. A predicted value of C0 is calculated based on this NB1 and C1. This calculation shows that the logarithmic curve shown in Figure 6(D) is (C1
, NB1).

Coefficient C0 and VLC obtained by the above calculation
The variable-length coded image signal output from the resync multiplexer 9 is input to the resync multiplexer 9 . The resync multiplexer 9 divides one screen (one frame) into a plurality of regions, and outputs the image signals of each divided region with information on the coefficient C0 added thereto.

The divided areas are hereinafter referred to as resync blocks, and FIG. 3 is a diagram illustrating the relationship between resync blocks in one frame in this embodiment.

In FIG. 3, one screen is 1280 pixels horizontally,
Assuming that the vertical pixels are 1088 pixels, and the image is A/D converted with 8 bits each, the amount of data per screen is 12
80×1088×8=11,141,120 bits.

[0046] Since the DCT conversion block is 8 pixels horizontally x 8 pixels vertically, and the amount of data for one image is also fixed, the resync block is a collection of 40 DCT blocks with a well-cut image. Block.

Therefore, one screen is made up of a total of 544 resync blocks divided into 4 horizontal by 136 vertical blocks.

Note that the amount of data per resync block is 40×8×8×8=20,480 bits.

Next, FIG. 4 shows the transmission signal format output from the resync multiplexer 9.

In FIG. 4, 30 is a vertical sync code;
31, 34, 37, 40 and 43 are resync codes, 3
2, 35, 38, 41 and 44 are coefficient information codes. The coefficient information code is the coefficient C0 in this embodiment.
This is the information code.

33, 36, 39, 42 and 45 are variable length coded image data. The resync multiplexer 9 outputs the transmission signal in the above-described format.

The signal output from the resync multiplexer 9 is error-corrected encoded in an ECC circuit 10, modulated in a form suitable for the characteristics of the transmission path in a modulation circuit 11, and sent out from an output terminal 12 to the transmission path.

Next, the configuration and operation of an image decoding apparatus according to an embodiment of the present invention will be explained with reference to FIG.

The transmission signal (transmission data string) on the transmission path is
The signal is input from the input terminal 16, demodulated by the demodulation circuit 17, and input to the ECC decoding circuit 18.

The ECC decoding circuit 18 corrects code errors in the transmission data string according to a predetermined algorithm. Also,
If data determined to be uncorrectable occurs, the error flag output for each code word of each error correction code is
1''. This error flag is input to the resync block error determination circuit 18a.

The resync block error determination circuit 18
In a, a block in which coefficient information is included in the code word for which the error flag is "1" is determined, and if there is a block in which the coefficient information is included, resync is output for each resync block of image information. Set the block error signal to “1”
”. This resync block error signal is input to the coefficient selector 27.

The transmission signal error-corrected by the ECC decoding circuit 18 is input to the coefficient decoder 19, where it is distinguished into an image information signal and a coefficient information signal.

On the other hand, the coefficient information identified by the coefficient decoder 19 is input to the coefficient selector 27. At this time, if "1" is not input as a resync block error signal from the resync block error determination circuit 18a to the coefficient selector 27, the coefficient information identified by the coefficient decoder 19 is input as is to the multiplier 28.

The image information signal identified by the coefficient decoder 19 is input to the IVLC 20 and variable length decoded,
The subsequent processing operations of zigzag scanning 21, linear inverse quantization 22, inverse quantization matrix generation circuit 29, and IDCT conversion 23, zigzag scanning 7, linear quantization 6, and quantization matrix explained in the encoding device of FIG. Calculations and processing that are completely opposite to those of the generation circuit 13 and the DCT conversion 5 are performed.

The image information subjected to the above-described processing is buffered in the frame memory 24 and sent to the D/A converter 25.
The signal is digital-to-analog converted and supplied to the output terminal 26. The signal is input from the output terminal 26 to an image display means such as a monitor.

On the other hand, if "1" is input to the coefficient selector 27 as the resync block error signal, the coefficient selector 27 ignores the coefficient information signal added to the uncorrectable resync block and performs multiplication. vessel 28
is replaced with coefficient information added to a resync block before or after the uncorrectable resync block.

Furthermore, if three or more uncorrectable resync blocks occur in one frame, the frame being processed is ignored, and the previous or subsequent frames are frozen. In other words, the coefficient selector 27 has a built-in resync block counter, and when the number of error blocks in the resync block exceeds a certain threshold in one frame, it sends a freeze signal to the frame memory 24, and the frame memory 24 follows the freeze signal. The image information signal of the current frame is not output, but the image information of the previous or subsequent frame is output. That is, the image information of the current frame is replaced.

The processing operation of the coefficient selector 27 will be explained in more detail with reference to FIG.

Assume that an error occurs in the transmission signal of FIG. 4 on the transmission path, and the error flag becomes "1" in section 1 of FIG. 4 in the ECC decoding circuit 18 of FIG. Here, the error flag is set to "1" for a code word for which error correction is not possible, and this error flag is input to the resync block error determination circuit 18a, and this error flag is set to "1".
It is determined whether the code word that has become 1'' contains coefficient information.

Since section 1 includes coefficient information 38 and 41, the resync block error determination circuit 18a outputs "1" as a resync block error signal to the image information of the resync block corresponding to the coefficient information 38 and 41. ” is input to the coefficient selector 27.

The coefficient selector 27 receives the resync block, ignores the coefficient information 38 and 41 in section 1, does not send the coefficient information 38 and 41 to the multiplier 28, and outputs the coefficient information 38 and 41 to the multiplier 28. 41, the coefficient information added to the previous resync block is sent. For example, it is coefficient information 35.

The coefficient selector 27 has a built-in resync block error counter, and when the number of error blocks in the resync block exceeds a certain threshold, the image of this frame is considered to have low reliability, and the image of the previous frame is Replace with In this embodiment, the threshold value for the number of errors in a resync block is set to two.

A specific explanation will be given using FIG. 4 as an example.

For example, when an error occurs in the coefficient information in section 2, the number of error blocks in the resync block is three, so the coefficient selector 27 sends a freeze signal to the frame memory 24 to freeze the image of the previous frame. to freeze the previous frame image.

Note that when replacing coefficient information or image information, previously processed information is replaced, but it is obvious that subsequent coefficient information or image information may be used.

[0071]

[Effects of the Invention] As explained above, the image signal encoding device and decoding device of the present invention quantizes the transformed data obtained by converting image information into the frequency domain, and makes the quantized transformed data variable. An encoding device that performs long encoding and a decoding device that decodes the encoded image information, comprising: 1
When decoding the encoded image information by dividing a frame into a plurality of blocks and adding coefficient information related to the quantization step of the encoded image information to each block, Since it has a detection means for detecting an error in the coefficient information for each block, and a replacement means for replacing the coefficient information of the block detected by the detection means with the coefficient information of another block, When an error occurs in the coefficient information, the error coefficient information can be ignored and effective coefficient information can be substituted for other blocks to improve the reliability of image reproduction.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an image encoding device according to an embodiment.

FIG. 2 is a block diagram of an image decoding device according to the present embodiment.

FIG. 3 is a configuration diagram of one frame according to this embodiment.

FIG. 4 is a diagram showing the form of a transmission signal according to the present embodiment.

FIG. 5 is a block diagram of a conventional image encoding device.

FIG. 6 is a block diagram of a conventional image decoding device.

FIG. 7 is a diagram for explaining processing of the encoding method shown in FIGS. 1 and 5. FIG.

[Explanation of symbols]

9 Sync multiplexer 18 ECC decoding circuit 18a Resync block error determination circuit 19 Coefficient decoder 27 Coefficient selector

Claims (5)

[Claims] 1. An image encoding device that quantizes transformed data obtained by transforming image information into a frequency domain, and encodes the quantized transformed data with variable length encoding, the apparatus comprising: converting one frame into a plurality of blocks; An image encoding apparatus characterized in that the image is divided and coefficient information related to the quantization step determined for each screen is added to each block. 2. Transform data obtained by converting image information into the frequency domain is quantized, the quantized transform data is variable-length coded, one frame is divided into a plurality of blocks, and each block is coded. An image decoding device that adds coefficient information related to the quantization step to the quantized transform data and performs variable length decoding of the variable length encoded data, the image decoding device detecting errors in the coefficient information for each block. An image decoding device comprising: a detection means for detecting a block detected by the detection means; and a replacement means for replacing coefficient information of a block detected by the detection means with coefficient information of another block. 3. The image decoding device according to claim 2, wherein when the number of blocks detected by the detection means is equal to or greater than a predetermined number within one frame, that frame is replaced with another frame. . 4. An image encoding device that quantizes image information and variable-length encodes the quantized transform data, the device divides one frame into a plurality of blocks, and divides one frame into a plurality of blocks and encodes the quantized data determined for each screen. An image encoding device characterized in that coefficient information related to the step of converting is added to each block. 5. The quantization step of quantizing image information, variable length encoding the quantized transform data, dividing one frame into a plurality of blocks, and converting the encoded transform data into each block. An image decoding apparatus that performs variable length decoding of the variable length coded data, the image decoding apparatus comprising: a detection means for detecting errors in the coefficient information for each block; and a detection means for detecting errors in the coefficient information for each block; 1. An image decoding device, comprising: replacement means for replacing coefficient information of a block with coefficient information of another block.