JPH06209469A - Picture coder, picture coding method, picture decoder, picture decoding method and picture recording medium - Google Patents

Abstract

PURPOSE:To improve the bit precision by applying conversion processing to quantization data of a difference between picture data and representative data for each prescribed block. CONSTITUTION:When the motion compensation mode is the in-picture prediction mode and a dynamic range of a block of picture data in 10-bit precision from a computing element 4 exceeds 8 bits, a block quantization device 7 converts a 10-bit block into data of 8-bit precision from the computing element 4 and outputted to a difference signal coder 8. On the other hand, in the case of the inter-picture prediction mode, a representative value calculation circuit 5 detects a maximum value and a minimum value of an arithmetic operation output of the computing element 3 and its mean value is outputted to computing elements 4, 11 as representative data of a block of the picture data in 10-bit precision outputted from the computing element 3. A difference between the block of the picture data in 10-bit precision from the computing element 3 and a mean value of the block from the circuit 5 is taken and outputted to a quantization device 7. Thus, the picture signal with high precision is coded and decoded.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus suitable for use in recording an image signal on a recording medium such as a magneto-optical disk or a magnetic tape and reproducing it.
The present invention relates to an image coding method, an image decoding device, an image decoding method, and an image recording medium.

[0002]

2. Description of the Related Art In a system for transmitting a moving image signal to a remote place such as a video conference system or a video telephone system, for example, line correlation or frame (or frame) of an image signal (or The image signal is compressed and encoded by utilizing the correlation between fields.

By utilizing the line correlation (two-dimensional correlation), the image signal can be subjected to DCT (discrete cosine transform) processing, for example, to concentrate the signal power on a specific frequency component, thereby compressing. be able to.

Further, by utilizing the correlation between frames (or fields), it becomes possible to further compress and encode the image signal. That is, since the images of frames that are temporally adjacent do not usually have such a large change,
When the difference between the two is calculated, the difference signal has a small value. Therefore, if this difference signal is encoded, the code amount can be compressed.

However, the original image cannot be restored by transmitting only the difference signal. Therefore, the image of each frame is set to any one of three types of pictures of I picture, P picture, and B picture,
The image signal is compressed and encoded.

That is, for example, image signals of 17 frames from frames F1 to F17 are grouped into a group of pictures (Gr
oup of picture, hereinafter referred to as GOP), and is a unit of processing. The image signal of the first frame F1 is encoded as an I picture, the second frame F2 is a B picture, and the third frame F3 is
Are processed as P pictures respectively. Hereinafter, the fourth and subsequent frames F4 to F17 are alternately processed as a B picture or a P picture.

As the image signal of the I picture, the image signal for one frame (or field) is transmitted as it is. On the other hand, as the image signal of the P picture, basically, the difference from the image signal of the I picture or P picture that precedes it in time is transmitted. Further, as the image signal of the B picture, basically, the difference from the average value of both the temporally preceding frame and the subsequent frame is obtained, and the difference is encoded.

As described above, when the moving image signal is encoded, the first frame F1 is processed as an I picture, and therefore, is transmitted as it is to the transmission path as transmission data (intra-picture encoding). On the other hand, the second frame F
Since 2 is processed as a B picture, the difference between the frame F1 temporally preceding and the average value of the frame F3 temporally following is calculated, and the difference is transmitted as transmission data.

However, there are four types of processing as the B picture, which will be described in more detail. The first process is to transmit the data of the original frame F2 as it is as transmission data (intra coding) (intra-picture prediction coding), and is the same process as in the I picture. The second process is to calculate the difference from the frame F3 that is temporally later and transmit the difference (backward predictive coding). The third process is to transmit the difference from the frame F1 that precedes in time (forward predictive coding). Further, the fourth processing is to generate a difference between an average value of a temporally preceding frame F1 and a temporally subsequent frame F3 and transmit this as transmission data (bidirectional predictive coding).

Of these four methods, the method with the least amount of transmission data is adopted.

When transmitting the difference data, the motion vector x1 between the image of the frame (prediction image) for which the difference is to be calculated (motion vector between the frames F1 and F2) (in the case of forward prediction). , Or x2 (motion vector between frames F3 and F2) (for backward prediction),
Or both x1 and x2 (for bidirectional prediction) are transmitted with the difference data.

In the frame F3 of the P picture, the frame F1 temporally preceding is used as a prediction image, the differential signal from this frame and the motion vector x3 are calculated, and this is transmitted as transmission data (forward prediction code). )).
Alternatively, the data of the original frame F3 is directly transmitted as transmission data (intra coding) (intra-picture predictive coding). Which method is used for transmission is B
As with pictures, the one with less transmitted data is selected.

FIG. 20 is a block diagram showing the structure of an example of a conventional image coding apparatus for coding a moving image signal based on the above-mentioned principle. First, for example, 8 × 8 pixels or 1
The 8-bit precision image data divided into blocks of 6 × 16 pixels is input to and stored in the field memory group 115 in block units.

Then, the motion prediction circuit 2 uses the image data of each frame (or field) stored in the field memory group 115 as an I picture, a P picture, or a B picture according to a preset predetermined sequence. To process. Which of I, P, and B pictures to process the images of sequentially input frames is determined in advance (for example, as described above, GO configured by the frames F1 to F17).
P is treated as I, B, P, B, P, ... B, P).

The signal of each picture stored in the field memory group 115 is read therefrom, and the operation unit 3 performs intra-picture prediction, forward prediction, backward prediction, or bidirectional prediction. Which of these processes is to be performed is determined in accordance with the prediction error signal (difference between the reference image to be processed and the predicted image corresponding thereto). Therefore, the motion prediction circuit 2 generates the sum of absolute values (or the sum of squares) of the prediction error signals used for this determination as follows.

That is, as the sum of the absolute values of the prediction errors of the intra-picture prediction, the sum Σ | Aij of the absolute value | ΣAij | of the sum ΣAij of the signal Aij of the block of the reference image and the absolute value | Aij | of the signal Aij of the block. Find the difference |. Also, as the sum of absolute values of the prediction errors in the forward prediction, the absolute value | Ai of the difference Aij-Bij between the signal Aij of the block of the reference image and the signal Bij of the block of the predicted image.
The sum Σ | Aij−Bij | of j−Bij | is calculated. Further, the sum of absolute values of the prediction errors of the backward prediction and the bidirectional prediction is also obtained in the same manner as in the case of forward prediction (the predicted image is changed to a predicted image different from that in forward prediction).

Then, the motion prediction circuit 2 calculates the sum of absolute values of prediction errors of forward prediction, backward prediction and bidirectional prediction.
The smallest one is selected as the sum of absolute values of prediction errors in inter-picture prediction. Further, the sum of absolute values of prediction errors of inter-picture prediction is compared with the sum of absolute values of prediction errors of intra-picture prediction, the smaller one is selected, and the mode corresponding to the selected sum of absolute values is motion compensated. Select as the mode. That is,
If the sum of absolute values of prediction errors in intra-picture prediction is smaller, the intra-picture prediction mode is set. If the sum of absolute values of prediction errors in inter-picture prediction is smaller, the motion compensation mode with the smallest corresponding absolute value sum is set among the forward prediction, backward prediction, and bidirectional prediction modes.

Further, the motion prediction circuit 2 detects the motion vector between the predicted image and the reference image corresponding to the set motion compensation mode, and outputs it to the variable length coding (VLC) circuit 117 and the motion compensation circuit 14. . As described above, the motion vector that minimizes the sum of absolute values of the corresponding prediction errors is selected.

When the I-picture image data is read from the field memory group 115, the motion prediction circuit 2
The intra-picture prediction mode (mode without motion compensation) is set as the motion compensation mode. As a result, the image data of the I picture passes through the arithmetic unit 3 (without arithmetic processing being performed in the arithmetic unit 3) and is input to the differential signal encoder 8.

The differential signal encoder 8 includes a DCT circuit 111.
And a quantizer 112. Quantizer 1
12 passes through the arithmetic unit 3 and the field memory group 11
I-picture image data supplied from
(Discrete cosine transform) process and transform into DCT coefficients.
This DCT coefficient is input to the quantizer 112, and VLC
After being quantized with a quantization width S corresponding to the data storage amount (buffer storage amount) of a transmission buffer (not shown) provided in the subsequent stage of the circuit 117, it is input to the VLC circuit 117.

The VLC circuit 117 converts the image data (in this case, I picture data) supplied from the quantizer 112 into a variable length code such as Huffman code, and a transmission buffer provided in the subsequent stage. Output to.

Here, the transmission buffer provided in the subsequent stage of the VLC circuit 117 increases the quantization width S of the quantizer 112 by the quantization control signal when the remaining data amount increases to the allowable upper limit value. , Reduce the amount of quantized data. On the contrary, when the data remaining amount decreases to the allowable lower limit value, the transmission buffer reduces the quantization width S of the quantizer 112 by the quantization control signal to reduce the data amount of the quantized data. Increase. In this way, overflow or underflow of the transmit buffer is prevented.

The VLC circuit 117 also includes a quantizer 11
2, the quantization width S is input, the motion prediction circuit 2 inputs a motion compensation mode (a mode that indicates which of intra-picture prediction, forward prediction, backward prediction, or bidirectional prediction has been set), and a motion vector. Is also variable-length coded.

Then, the data output from the VLC circuit 117 is temporarily stored in the transmission buffer at the subsequent stage, read at a predetermined timing, output to the transmission path, and recorded in, for example, a recording medium.

On the other hand, the I picture data output from the quantizer 112 is also input to the differential signal decoder 9. The differential signal decoder 9 is composed of an inverse quantizer 113 and an inverse DCT circuit 114, and the inverse quantizer 113 is
The I-picture data output from the quantizer 112 is inversely quantized with the same quantization width as the quantization width S in the quantizer 112. The output of the inverse quantizer 113 is the inverse DCT
After being input to the circuit 114 and subjected to inverse DCT processing, it is supplied to the field memory group 116 via the arithmetic unit 12 and stored therein.

The motion prediction circuit 2 receives the image data of each frame, which is sequentially input, as I, B,
When processed as P, B, P, B ... pictures, respectively, after processing the image data of the first input frame as an I picture, before processing the image of the next input frame as a B picture. Further, the image data of the frame input next is processed as a P picture. This is because a B picture is accompanied by backward prediction and cannot be decoded unless a P picture as a backward predicted image is prepared in advance.

Then, the motion prediction circuit 2 starts the processing of the image data of the P picture stored in the field memory group 115 after the processing of the I picture. Then, similarly to the case described above, the intra-picture prediction or forward prediction motion compensation mode is set in correspondence with the sum of absolute values of the inter-frame difference (prediction error) in block units.

When the intra-picture prediction mode is set, the arithmetic unit 3 outputs the data from the field memory group 115 without performing arithmetic processing as described above. Therefore, this data is transmitted to the transmission line via the differential signal encoder 8, the VLC circuit 117, and the transmission buffer, similarly to the I picture data. Further, this data is supplied to and stored in the field memory group 116 via the differential signal decoder 9 and the arithmetic unit 12.

In the forward prediction mode, image data (in this case, an I-picture image) data stored in the field memory group 116 is read out, and the motion compensation circuit 14
Motion compensation is performed according to the motion vector output from the motion prediction circuit 2. That is, when the motion prediction circuit 2 issues an instruction to set the forward prediction mode, the motion compensation circuit 14 sets the read address of the field memory group 116 to the position corresponding to the position of the block currently output by the motion prediction circuit 2. Then, the data is read out by shifting by the amount corresponding to the motion vector, and predicted image data is generated.

The predicted image data generated by the motion compensation circuit 14 is supplied to the arithmetic unit 3 via the field memory group 116. In this case, the calculator 3 subtracts the predicted image data corresponding to this block supplied from the motion compensation circuit 14 from the data of the block of the reference image supplied from the field memory group 115, and calculates the difference (prediction error). ) Is output. This difference data is transmitted to the transmission line via the difference signal encoder 8, the VLC circuit 117, and the transmission buffer. The difference data is locally decoded by the difference signal decoder 9 and input to the calculator 12.

The calculator 12 is also supplied with the same data as the predicted image data supplied to the calculator 3. The calculator 12 adds the prediction image data output by the motion compensation circuit 14 to the difference data output by the difference signal decoder 9. As a result, the image data of the original (decoded) P picture is obtained. The image data of the P picture is supplied to and stored in the field memory group 116.

After the I-picture and P-picture data are stored in the field memory group 116 in this way, the motion prediction circuit 2 next executes the B-picture processing. That is, the motion compensation mode is set to one of the intra-picture prediction mode, the forward prediction mode, the backward prediction mode, and the bidirectional prediction mode, corresponding to the magnitude of the sum of absolute values of the inter-frame differences in block units.

In the intra-picture prediction mode or the forward prediction mode, the same processing as in the case of the P picture as described above is performed and the data is transmitted.

On the other hand, in the backward prediction mode, image data (in this case, P picture image) data stored in the field memory group 116 is read out, and the motion compensation circuit 14 causes the motion prediction circuit 2 to read. Is motion-compensated corresponding to the motion vector output by. That is, the motion compensation circuit 14
When the backward prediction mode setting is instructed by the motion prediction circuit 2, the read address of the field memory group 116 corresponds to the motion vector from the position corresponding to the position of the block currently output by the motion prediction circuit 2. The data is read out by shifting by an amount, and predicted image data is generated.

The predicted image data generated by the motion compensation circuit 14 is supplied to the arithmetic unit 3 via the field memory group 116. The computing unit 3 includes a field memory group 11
The predicted image data supplied from the motion compensation circuit 14 is subtracted from the data of the block of the reference image supplied from No. 5, and the difference is output. This difference data is transmitted to the transmission line via the difference signal encoder 8, the VLC circuit 117, and the transmission buffer.

In the bidirectional prediction mode, the images stored in the field memory group 116 (in this case, I
Picture image data, and image (in this case, P
(Picture image) data is read out, and the motion compensation circuit 1
4, the motion compensation is performed according to the motion vector output from the motion prediction circuit 2. That is, when the motion prediction circuit 2 instructs the motion compensation circuit 14 to set the bidirectional prediction mode, the motion compensation circuit 14 sets the read address of the field memory group 116 to the position corresponding to the position of the block currently output by the motion prediction circuit 2. The data is read out by shifting by the amount corresponding to the motion vector (in this case, there are two motion vectors for the forward prediction image and the backward prediction image), and predicted image data is generated.

The predicted image data generated by the motion compensation circuit 14 is supplied to the arithmetic unit 3 via the field memory group 116. The calculator 3 subtracts the average value of the predicted image data supplied from the motion compensation circuit 14 from the reference image block data supplied from the motion prediction circuit 2,
The difference is output. This difference data is transmitted to the transmission line via the difference signal encoder 8, the VLC circuit 117, and the transmission buffer.

Note that the B picture image is not used as a predicted image of another image, and therefore, the field memory group 1
It is not stored in 16.

In the above processing, the timing of reading the image data from the field memory groups 115 and 116 is controlled based on the timing control signal generated by the memory controller 16 in response to the synchronizing signal of the input image. It

Next, FIG. 21 is a block diagram showing the structure of an example of an image decoding apparatus for decoding the image coded by the image coding apparatus shown in FIG. The encoded image data is received by a receiving circuit (not shown) via a transmission path, reproduced by a reproducing device from a recording medium, and supplied to a variable length decoding (inverse VLC) circuit 124. Reverse VLC circuit 1
Reference numeral 24 denotes variable-length decoding of the encoded image data, the motion vector and motion compensation mode to the motion compensation circuit 27, and the quantization width S, and the decoded (variable-length decoded) image data to a differential signal. Output to the decoder 22 respectively.

The differential signal decoder 22 is similar to the differential signal decoder 9 in FIG.
The inverse quantizer 121 includes a T circuit 122.
Is the image data supplied from the inverse VLC circuit 124,
Similarly, inverse quantization is performed according to the quantization width S supplied from the inverse VLC circuit 124, and the result is output to the inverse DCT circuit 122. Data output from the inverse quantizer 121 (DCT coefficient)
Is subjected to inverse DCT processing by the inverse DCT circuit 122 and supplied to the calculator 25.

When the image data supplied from the inverse DCT circuit 122 is I-picture data, the data is output from the calculator 25 as it is, and the image data (P or B-picture data) input later to the calculator 25 is input. Field memory group 12 for generating predicted image data
3 and is stored. Further, this data is output as decoded image data via the field memory group 123.

When the image data supplied from the inverse DCT circuit 122 is the data of the P picture in which the image data of the immediately preceding frame is the predicted image data and the data of the forward prediction mode, it is stored in the field memory group 123. The stored image data of one frame before (data of I picture) is read out, and the motion compensation circuit 27 performs motion compensation corresponding to the motion vector output from the inverse VLC circuit 124. Then, in the calculator 25, the inverse DCT
Image data supplied from the circuit 122 (difference data)
Is added and output. This added data, that is, the decoded P picture data, is stored in the field memory group 123 in order to generate predictive image data of the image data (B picture or P picture data) input later to the calculator 25. Supplied and stored.

Even in the case of P-picture data, the intra-picture prediction mode data is stored in the field memory group 123 as it is without any special processing by the calculator 25, as in the case of I-picture data.

Since this P picture is an image to be displayed next to the next B picture, it is not yet output from the field memory group 123 at this point (as described above, the P input after the B picture is input). Pictures have been processed and transmitted before B pictures).

When the image data supplied from the inverse DCT circuit 122 is B picture data, the I picture stored in the field memory group 123 corresponding to the motion compensation mode supplied from the inverse VLC circuit 124. Image data (in the forward prediction mode), P-picture image data (in the backward prediction mode), or both image data (in the bidirectional prediction mode),
In the motion compensation circuit 27, motion compensation corresponding to the motion vector output from the inverse VLC circuit 124 is performed,
A predicted image is generated. However, when motion compensation is not required (in the case of the intra-picture prediction mode), the predicted picture is not generated.

The data thus motion-compensated by the motion compensation circuit 27 is processed by the calculator 25 in the inverse DC.
It is added to the output of the T circuit 122. This addition output is output as it is via the field memory group 123.

Since this addition output is B picture data and is not used for predictive image generation of other images, it is not stored in the field memory group 123.

After the B picture image is output, the P picture image data stored in the field memory group 123 is read and output.

The data output from the field memory group 123 is converted from the block format to the frame format. Then, D / A conversion circuit etc.
After being A-converted, it is output and displayed on a display (not shown) such as a CRT, for example.

If the image is a color image, the above processing is performed on the image luminance signal and color signal (for example, color difference signal).
Is applied to both. However, in this case, the motion vector used for the luminance signal is sometimes halved in the vertical and horizontal directions.

In the above processing, the memory controller 2 outputs the timing pulse at the timing when the inverse VLC circuit 124 performs the variable length decoding of one image (one screen).
8 and the memory controller 28 supplies a timing control signal to the field memory group 123 corresponding to the timing pulse. As a result, the timing of reading image data from the field memory group 123 is controlled.

The hybrid image coding method combining the DCT processing and motion compensation as described above is CCIT which is a moving image coding standard used for videophones.
H.T (International Telegraph and Telephone Advisory Committee) 261 or ISO-IEC / which is a moving image coding standard for storage media.
It is used in JTC1 / SC29 / WG11 (so-called MPEG) and the like.

As a format for processing an image signal as a digital signal (converting into a digital signal), for example, CCIR (International Radiocommunication Advisory Committee) Recommendation 601 (hereinafter referred to as Rec. 601), etc. It has been known.

In Rec. 601, an image signal as a digital signal is a 4: 2: 2 component signal with 8-bit precision (color difference signals are thinned out from every other line, and two luminances are obtained from each line. The signal is handled in a digital image signal format in which one color difference signal is associated with the signal.

For example, in MPEG, the Rec.601 format is adopted as the input / output image signal format. Therefore, in the above-mentioned image coding apparatus and image decoding apparatus, for example, the motion prediction circuit 2 and the differential signal code are used. Digitizer 8 (DCT circuit 111, quantizer 112), differential signal decoder 9 (inverse quantizer circuit 121, inverse DCT circuit 1)
22), 22 (inverse quantization circuit 121, inverse DCT circuit 12)
The hardware (IC or LSI), etc., which constitutes the central block of the device such as 2) is designed, manufactured and marketed on the assumption that an 8-bit precision digital image signal is input to or output from the device. There is.

That is, for example, when the motion compensation mode is the intra-picture prediction mode, the quantizer 112 of the differential signal encoder 8 converts the image data, which can be represented by 8 bits of 0 to 255, into the motion compensation mode. In the case of inter-picture prediction mode (either forward prediction, backward prediction or bidirectional prediction mode), code bits are added to 8 bits to -255 to 25
It is designed, manufactured, and commercially available so as to be able to quantize image data that can be represented by 5 9 bits.

[0058]

By the way, with the progress of digital signal processing technology, an image signal with higher precision than an image signal with 8-bit precision is required. That is, for example, a movie film is recorded on an HDTV (High Definition
For example, in the case of performing telecine on the signal of the ition TV) or CT used in the medical field and the like, an image signal having higher precision than 8-bit precision, for example, 10-bit precision is required.

However, as described above, it is premised that the hardware which constitutes the central block of the existing image coding apparatus or image decoding apparatus inputs / outputs an 8-bit precision image signal to / from the apparatus. Therefore, it has been difficult to encode or decode an image signal whose precision exceeds 8 bits by a conventional device.

The present invention has been made in view of such a situation, and makes it possible to encode and decode a high-precision image signal using conventional hardware. .

[0061]

An image coding method according to claim 1 determines representative value data and quantization width of image data for each predetermined block, and calculates a difference between the image data and the representative value data. The difference is calculated, the difference is quantized based on the quantization width, the first quantized data is generated, a predetermined conversion process is performed on the first quantized data, the conversion coefficient is generated, and the conversion coefficient is quantized, It is characterized in that the second quantized data is generated and the representative value data, the quantization width, and the second quantized data are encoded.

Further, according to this image coding method, the difference between the representative value data of the image data in the adjacent blocks is
It can be encoded as representative value data.

In this image coding method, whether the quantization width of the image data in the block coded immediately before and the quantization width of the image data in the block coded now are the same or not. Only when the quantization width of the image data in the immediately previous encoded block and the quantization width of the image data in the currently encoded block are different from each other, the comparison information is further encoded in the currently encoded block. The quantization width of the image data can be encoded.

Further, in this image coding method, the representative value data and the quantization width can be added to the header of the block.

Further, in this image coding method, the representative value data and the quantization width can be added to the header of the layer above the block.

Furthermore, this image coding method is based on size information about the size of the block or accuracy information about the bit precision of the image data in the block.
The quantization width can be determined so that the size information and the accuracy information can be further encoded.

Further, in this image coding method, the representative value data can be determined based on the size information regarding the size of the block, and the size information can be further coded.

According to the image coding method of the eighth aspect, the representative value data and the quantization width of the image data are determined for each predetermined unit area, the difference between the image data and the representative value data is calculated, and the difference is calculated. Quantize based on the quantization width to generate first quantized data, perform predetermined conversion processing on the first quantized data for each predetermined block, generate conversion coefficients,
It is characterized in that the transform coefficient is quantized to generate second quantized data, and the representative value data, the quantized width, and the second quantized data are encoded.

Further, in this image coding method, the unit area and the block size may have an integral multiple relationship.

According to the image coding method of the tenth aspect, the difference between the input image and the predicted image is calculated for each predetermined block, the first difference data is generated, and the representative value data of the first difference data is calculated. And the quantization width are determined, the difference between the first difference data and the representative value data is calculated, the second difference data is generated, the second difference data is quantized based on the quantization width, and the quantized data , Dequantize the quantized data based on the quantization width, generate dequantized data, add the dequantized data and the representative value data, generate locally decoded data, and based on the locally decoded data Then, a new predicted image is generated.

Further, in this image coding method, the difference between the locally decoded data and the representative value data is calculated, the difference is quantized based on the quantization width, and the difference quantized data is generated. The value data and the quantization width are stored in a storage unit such as the memory unit 79, and the difference quantization data, the representative value data, and the quantization width are stored in the memory unit 79.
, And a new predicted image can be generated.

According to the twelfth aspect of the present invention, the image encoding device has a block representative value calculating circuit 5 as a representative value determining means for determining representative value data of image data for each predetermined block, and for each predetermined block. , A block quantization width calculation circuit 6 as a quantization width determination means for determining a quantization width of image data, an arithmetic unit 4 as a difference calculation means for calculating a difference between image data and representative value data, and a difference First quantizing based on the quantization width to generate first quantized data
A block quantizer 7 as a quantizing means for the first quantizer, and a predetermined conversion process on the first quantized data to quantize the first quantized data;
Difference signal encoder 8 as the second quantizing means for generating the quantized data of V, and the representative value data, the quantization width, and V as the encoding means for coding the second quantized data.
And an LC circuit 15.

An image coding apparatus according to a thirteenth aspect of the present invention is an arithmetic unit as a first difference calculating means for calculating a difference between an input image and a predicted image for each predetermined block to generate first difference data. 3, an in-block representative value calculation circuit 5 as a representative value determining unit that determines the representative value data of the first difference data for each predetermined block, and the quantization of the first difference data for each predetermined block. A block quantization width calculating circuit 6 as a quantization width determining means for determining a width, and a second difference calculating means for calculating a difference between the first difference data and the representative value data and generating second difference data. Calculator 4
And quantizes the second difference data based on the quantization width,
A block quantizer 7 as a quantizing means for generating quantized data, and a block dequantizer as an inverse quantizing means for dequantizing the quantized data based on a quantization width to generate dequantized data. 10, an arithmetic unit 11 as an addition unit that adds the dequantized data and the representative value data to generate locally decoded data, and an arithmetic unit as a prediction unit that generates a new predicted image based on the locally decoded data. 12 and a motion compensation circuit 14 are provided.

According to a fourteenth aspect of the present invention, in the image decoding method, the encoded image data, the representative value data set for each predetermined block, and the quantization width are extracted from the transmission data to obtain the encoded image data. Inverse quantization is performed to generate first inverse quantized data, predetermined inverse transform processing is applied to the first inverse quantized data, inverse transform data is generated, and inverse transform data is inversely quantized based on a quantization width. The second inverse quantized data is generated, the second inverse quantized data is added to the representative value data, and the decoded image data is generated.

Further, in this image decoding method, the representative value data can be the difference between the representative value data of the image data in the adjacent blocks.

Further, according to this image decoding method, the quantization width of the image data in the block decoded immediately before and the quantization width of the image data in the block to be decoded are the same from the transmission data. If the quantization width of the image data in the block decoded immediately before and the quantization width of the image data in the block that is about to be decoded are the same and the comparison information of whether or not there is any more is extracted, The inverse transform data can be inversely quantized on the basis of the quantization width of the image data in the block that has been decoded into.

According to a seventeenth aspect of the present invention, in the image decoding method, the encoded image data, the representative value data set for each predetermined unit area, and the quantization width are extracted from the transmission data, and each predetermined block is extracted. , Inversely quantize the encoded image data to generate first inverse quantized data, perform predetermined inverse transform processing on the first inverse quantized data, generate inverse transformed data, Then, the inverse transformed data is inversely quantized based on the quantization width to generate the second inverse quantized data, and the second inverse quantized data and the representative value data are added to generate the decoded image data. Characterize.

Further, in this image decoding method, the unit area and the size of the block may have an integral multiple relationship.

An image decoding method according to a nineteenth aspect is an image decoding method for decoding an image based on a predicted image, and is a representative set from transmission data to encoded image data and predetermined blocks. Value data and quantization width are extracted, coded image data is inversely quantized based on the quantization width, inversely quantized data is generated, inversely quantized data and representative value data are added, and decoded difference data Produces
It is characterized in that the decoded difference data and the predicted image are added to generate the decoded image data.

Furthermore, this image decoding method calculates the difference between the decoded image data and the representative value data, quantizes the difference based on the quantization width, generates decoded quantized data, and decodes the quantized quantized data. The representative value data and the quantization width are stored in a storage unit such as the memory unit 106, and the decoded quantized data, the representative value data, and the quantization width are read from the memory unit 106 to generate a predicted image. be able to.

An image decoding apparatus according to a twenty-first aspect of the invention is an inverse VLC as an extracting means for extracting encoded image data, representative value data set for each predetermined block, and quantization width from transmission data. The circuit 21 and the first dequantized data are dequantized to generate the first dequantized data, and then the first dequantized data is subjected to a predetermined deconversion processing to generate deconversion data. A differential signal decoder 22 as an inverse quantizer, and a block inverse quantizer as a second inverse quantizer that inversely quantizes the inverse transform data based on the quantization width to generate second inverse quantized data. Chemicalizer 23
And the second dequantized data and the representative value data are added to each other, and an arithmetic unit 24 as an addition means for generating decoded image data is provided.

An image decoding apparatus according to a twenty-second aspect is an image decoding apparatus which decodes an image based on a predicted image, and is a representative set from transmission data to encoded image data and predetermined blocks. An inverse VLC circuit 21 as an extracting means for extracting the value data and the quantization width,
Decode the encoded image data based on the quantization width,
A block dequantizer 23 as dequantization means for dequantized data, an arithmetic unit 24 as first addition means for adding dequantized data and representative value data to generate decoding difference data, and decoding An arithmetic unit 25 as a second addition means for adding the difference data and the predicted image to generate the decoded image data is provided.

An image decoding apparatus according to a twenty-third aspect of the present invention comprises an inverse VLC circuit 21 as an extraction means for extracting coded image data and accuracy information regarding bit accuracy of the image data from transmission data, and the accuracy information. On the basis of the bit accuracy determination circuit 61 as a determination means for determining whether or not the encoded image data is decodable, if the encoded image data is undecodable, the encoded image data is decoded. The display control circuit 62 is provided as a notification means for notifying that the display is impossible.

According to a twenty-fourth aspect of the present invention, in an image decoding device, encoded image data, accuracy information regarding bit precision of image data, representative value data set for each predetermined block, and quantum data are transmitted. An inverse VLC circuit 21 as an extracting means for extracting the quantization width, and inverse quantization of the encoded image data to generate first inverse quantized data, and then a predetermined inverse conversion to the first inverse quantized data. A differential signal decoder 22 as first dequantization means for performing processing to generate detransformed data, and dequantization of the detransformed data based on a quantization width to generate second dequantized data. Block dequantizer 23 as second dequantization means
, The second dequantized data and the representative value data are added to generate the first decoded image data, and the intra-block representative value adder 41 as an adding means, and the first decoded image data are used as accuracy information. And a block quantizer 51 as a quantizing means for generating second decoded image data based on the quantization.

In the image recording medium according to the twenty-fifth aspect, the representative value data and the quantization width of the image data are determined for each predetermined block, the difference between the image data and the representative value data is calculated, and the difference is quantized. Quantize based on the width to generate first quantized data, perform a predetermined conversion process on the first quantized data, generate a transform coefficient, quantize the transform coefficient, and generate the second quantized data. Generate, representative value data, quantization width,
And a bitstream generated by encoding the second quantized data is recorded.

[0086]

In the image coding method according to claim 1 and the image coding apparatus according to claim 12, the representative value data and the quantization width of the image data are determined for each predetermined block, and the image data And the difference between the representative value data is calculated, and the difference is quantized based on the quantization width to generate first quantized data. Then, the first quantized data is subjected to a predetermined conversion process, the transform coefficient is quantized to generate second quantized data, and the representative value data, the quantized width, and the second quantized data are obtained. Encode. Therefore, the image
The bit precision can be converted and encoded.

In the image coding method according to the eighth aspect, the representative value data and the quantization width of the image data are determined for each predetermined unit area, and the difference between the image data and the representative value data is calculated, Quantize the difference based on the quantization width,
First quantized data is generated. Then, for each predetermined block, a predetermined conversion process is performed on the first quantized data, the conversion coefficient is quantized to generate second quantized data, and the representative value data, the quantization width, and the The quantized data of 2 is encoded. Therefore, since the representative value data and the quantization width can be changed for each predetermined unit area according to the complexity of the image, the deterioration of the image quality of the image can be prevented.

In the image coding method according to the tenth aspect and the image coding apparatus according to the thirteenth aspect, the difference between the input image and the predicted image is calculated for each predetermined block, and the first difference data is calculated. Is generated to determine the representative value data and the quantization width of the first difference data. And the first
The difference between the difference data and the representative value data is calculated, the second difference data is generated, the second difference data is quantized based on the quantization width, and the quantized data is generated. Further, the quantized data is inversely quantized based on the quantization width, the inverse quantized data is generated, the inverse quantized data and the representative value data are added, the locally decoded data is generated, and then the locally decoded data is generated. To generate a new predicted image. Therefore, the bit precision of the image can be converted and the image can be efficiently encoded.

According to the image decoding method of the fourteenth aspect and the image decoding apparatus of the twenty-first aspect, the transmission data, the encoded image data, and the representative value data set for each predetermined block, And the quantization width are extracted, and the encoded image data is inversely quantized to generate the first inverse quantized data. Then, a predetermined inverse transform process is performed on the first inverse quantized data, the inverse transformed data is inversely quantized based on the quantization width, and second inverse quantized data is generated. Data and representative value data,
Generates decoded image data. Therefore, the image signal whose bit precision has been converted can be decoded.

In the image decoding method according to the seventeenth aspect, the encoded image data, the representative value data set for each predetermined unit area, and the quantization width are extracted from the transmission data to obtain a predetermined block. Each time, the encoded image data is dequantized to generate the first dequantized data.
Then, the first inverse quantized data is subjected to predetermined inverse transform processing, and the inverse transformed data is inverse quantized for each predetermined unit area based on the quantization width to generate second inverse quantized data. Then, the second dequantized data and the representative value data are added to generate decoded image data. Therefore, it is possible to decode the encoded image by changing the representative value data and the quantization width for each predetermined unit area according to the complexity of the image.

According to the image decoding method of the nineteenth aspect and the image decoding apparatus of the twenty-second aspect, from the transmission data to the encoded image data, and the representative value data set for each predetermined block, And the quantization width are extracted, the encoded image data is inversely quantized based on the quantization width, the inverse quantized data is generated, the inverse quantized data and the representative value data are added, and the decoded difference data is obtained. To generate. Then, the decoded difference data and the predicted image are added to generate decoded image data. Therefore, it is possible to efficiently decode image signals that are encoded and have different bit precisions.

In the image decoding apparatus according to the twenty-third aspect, the encoded image data and the accuracy information regarding the bit accuracy of the image data are extracted from the transmission data, and encoded based on the accuracy information. It is determined whether the image data can be decoded. Then, if the encoded image data is undecodable, the fact that the encoded image data is undecodable is notified. Therefore, it is possible to inform the user that the device does not support the bit precision of the input data.

In the image decoding apparatus according to the twenty-fourth aspect, from the transmission data, the encoded image data, accuracy information regarding the bit accuracy of the image data, representative value data set for each predetermined block, and After the quantization width is extracted, the encoded image data is inversely quantized to generate the first inverse quantized data, the first inverse quantized data is subjected to a predetermined inverse transform process, and the inverse transformed data is obtained. To generate. Then, the inversely transformed data is inversely quantized based on the quantization width to generate second inversely quantized data, the second inversely quantized data and the representative value data are added, and the first decoded image data is obtained. To generate. Further, the first decoded image data is quantized based on the accuracy information to generate the second decoded image data. Therefore, an image signal with arbitrary bit precision can be obtained.

In the image recording medium according to the twenty-fifth aspect, the representative value data and the quantization width of the image data are determined for each predetermined block, the difference between the image data and the representative value data is calculated, and the difference is quantized. Quantization is performed based on the quantization width to generate first quantized data, the first quantized data is subjected to predetermined conversion processing, conversion coefficients are generated, the conversion coefficient is quantized, and second quantized data is generated. Is generated and the representative value data, the quantization width, and the second quantization data are encoded, and the bit stream generated by the encoding is recorded. Therefore, it is possible to record an image signal with arbitrary bit precision.

[0095]

1 is a block diagram showing the configuration of an embodiment of an image coding apparatus according to the present invention. In the figure, parts corresponding to those in FIG. 20 are designated by the same reference numerals. Here, description will be made assuming that image data having a bit precision higher than 8-bit precision, for example, 10-bit precision is input to the apparatus.

The field memory group 1 is adapted to store image data of 10-bit precision, and temporarily stores digital image data (image pixel data) divided into blocks of 16 × 16 pixels, for example. A block of 10-bit precision image data is stored in the storage unit 3 and is output to the arithmetic unit 3 based on the timing control signal generated by the field memory controller 16 in response to the input image synchronization signal.

The motion prediction circuit 2 operates as described above.
A motion compensation mode (intra-picture prediction mode, forward prediction, backward prediction, or bidirectional prediction mode) is set for a block of image data as a reference image stored in the field memory group 1, and the set motion compensation mode is set. A motion vector between the corresponding predicted image and reference image is detected and output to the motion compensation circuit 14 and the VLC circuit 15. Further, the motion prediction circuit 2 supplies the motion compensation mode to the intra-block representative value calculation circuit 5 and the block quantization width calculation circuit 6 via the motion compensation circuit 14 and the VLC circuit 15 as well.

The motion prediction circuit 2 in FIG. 1 has the same configuration as that in FIG.
Therefore, since the motion prediction circuit 2 is configured to be able to input 8-bit image data, it is not possible to connect all the connection lines from the field memory group 1.

Therefore, the motion predicting circuit 2 is connected to the connection lines corresponding to the image data of 10-bit precision from the field memory group 1 except the lower 2-bit connection line.

That is, the motion predicting circuit 2 is adapted to receive 8-bit precision image data in which lower-order 2 bits of 10-bit precision image data are truncated.

A block of 10-bit precision image data as a reference image stored in the field memory group 1 is
When it is read from there and supplied to the arithmetic unit 3, a predicted image corresponding to the motion compensation mode set in the motion estimation circuit 2 is further supplied to the arithmetic unit 3 via the field memory group 13. Supplied by.

In the arithmetic unit 3, as described above,
A calculation process corresponding to the motion compensation mode is performed (when the motion compensation mode is the intra-picture prediction mode, the block of 10-bit precision image data from the field memory group 1 is output as it is, and the motion compensation mode is the image. In the case of the inter prediction mode (either forward prediction, backward prediction, or bidirectional prediction mode), the difference between the block of 10-bit precision image data from the field memory group 1 and the predicted image from the motion compensation circuit 14 (Prediction error) is calculated), and an arithmetic output with 10-bit precision is supplied to the arithmetic unit 4, the in-block representative value calculation circuit 5, and the block quantization width calculation circuit 5.

The in-block representative value calculation circuit (hereinafter, referred to as a representative value calculation circuit) 5 corresponds to the motion compensation mode supplied from the motion prediction circuit 2 via the motion compensation circuit 4 and corresponds to the motion compensation mode. The representative value data of the arithmetic output is calculated and output as described later. The block quantization width calculation circuit (hereinafter, referred to as a quantization width calculation circuit) 6 outputs the calculation output of the calculator 3 in response to the motion compensation mode supplied from the motion prediction circuit 2 via the motion compensation circuit 4. Is calculated by the block quantizer 7 as will be described later, and is output to the block quantizer 7 and the block dequantizer 10.

Here, in the present specification, the quantization width determined by the quantization width calculation circuit 6 and the quantization in the differential signal encoder 8, the differential signal decoder 9 and 22 (FIG. 4). The quantization widths are described as quantization widths Q and S to distinguish them.

The arithmetic output of the arithmetic unit 3 is subtracted from the representative value data output from the representative value calculation circuit 5 in the arithmetic unit 4, and is output to the block quantizer 7. In the block quantizer 7, the operation output of the operator 4 is the quantization width Q.
Is quantized by, that is, divided by the quantization width Q, and output to the differential signal encoder 8.

Although not shown, the differential signal encoder 8 is composed of a DCT circuit 111 and a quantizer 112 as in the case of FIG. 20, in which the quantized data from the block quantizer 7 is sent. , VLC circuit 15 is quantized with a quantization width S corresponding to the storage amount of a transmission buffer (not shown) provided in the subsequent stage, and DCT processing is performed.

Here, as described above, the DCT circuit 111 of the differential signal encoder 8 outputs 0 when 8-bit precision image data, that is, the motion compensation mode is the intra-picture prediction mode.
If the motion compensation mode is the inter-picture prediction mode (either forward prediction, backward prediction or bidirectional prediction mode),
8 bits plus sign bit -255 to 255 9
It is necessary to input image data that can be represented by bits.

That is, it is necessary to convert the bit precision of the block of the image data input to the apparatus from 10-bit precision to 8-bit precision and input it to the differential signal encoder 8.

Therefore, when the motion compensation mode is the intra-picture prediction mode, in the representative value calculation circuit 5, the arithmetic unit 3
Operation output (in this case, 1 from the field memory group 1)
The maximum value and the minimum value of a block of 0-bit precision image data) are detected, and for example, the minimum value is calculated as the representative value data of the block of 10-bit precision image data output from the calculator 3. It is output to the containers 4 and 11.

In this case, the representative value data of the block may not be the minimum value of the block. However, when the motion compensation mode is the intra-picture prediction mode, since the image data needs to be represented by 8 bits of 0 to 255, the case where the representative value data of the block is the minimum value of the block will be described later. You will be able to maximize the dynamic range of the block.

Then, in the arithmetic unit 4, the block of the image data of 10-bit precision output from the arithmetic unit 3 (each pixel data in the block) and the representative value data of the block output from the representative value calculation circuit 5, that is, The difference from the minimum value of the block is calculated and output to the block quantizer 7.

Therefore, in this case, the block quantizer 7 outputs a block of image data having a minimum value of 0 to the block quantizer 7.

That is, when the motion compensation mode is the intra-picture prediction mode, for example, as shown in FIG.
The maximum value or the minimum value of the block of 10-bit precision image data output from is 500 or 300, respectively.
If so, 0 (= 300−30) obtained by subtracting the minimum value 300 as the representative value data from the block of the image data
A block of image data having a value in the range of 0) to 200 (= 500-300) is output from the calculator 4 to the block quantizer 7, and, for example, as shown in FIG.
If the maximum value or the minimum value of the block of 10-bit precision image data output from the computing unit 3 is 1000 or 300, respectively, the minimum value 300 as the representative value data is subtracted from the block of the image data, 0 (= 300
A block of image data having a value in the range of −300) to 700 (= 1000−300) is output from the calculator 4 to the block quantizer 7.

At the same time, in this case, the quantization width calculation circuit 6 detects the maximum value and the minimum value of the arithmetic output of the arithmetic unit 3 (in this case, a block of 10-bit precision image data from the field memory group 1). , The dynamic range as the difference is calculated.

Here, in this specification, the dynamic range of a signal means the difference between the maximum value and the minimum value of the signal.

Then, in the quantization width calculation circuit 6, whether the dynamic range of the block of 10-bit precision image data from the field memory group 1 is within 8 bits which can be represented by the range of 0 to 255. Is determined and the dynamic range is determined to be within 8 bits, the quantization width Q is determined to be 1 and output to the block quantizer 7.

Therefore, when the motion compensation mode is the intra-picture prediction mode, if the dynamic range of the block of 10-bit precision image data output from the calculator 3 is within 8 bits (for example, FIG. 2A). ), The image data from the arithmetic unit 4 is output from the block quantizer 7 to the difference signal encoder 8 as it is.

That is, when the motion compensation mode is the intra-picture prediction mode, when the dynamic range of the block of the image data of 10-bit precision output from the arithmetic unit 3 is within 8 bits, it is output from the arithmetic unit 3. The bit precision of the block of image data is substantially 8 bits, and this block of image data of 8-bit precision is output to the differential signal encoder 8.

Further, when the motion compensation mode is the intra-picture prediction mode, in the quantization width calculation circuit 6, the dynamic range of the block of 10-bit precision image data from the field memory group 1 exceeds 8 bits (25
6 or more), the block quantizer 7
Dynamic range of quantized output of 8 bits (25
The quantization width Q is determined so that it falls within 5) and is output to the block quantizer 7.

That is, when the dynamic range of the block of 10-bit precision image data from the field memory group 1 is, for example, 256 or more and less than 512, the quantization width Q is determined to be 2. Further, when the dynamic range is, for example, 512 or more and less than 768, the quantization width Q is determined to be 3, and when the dynamic range is, for example, 768 or more and less than 1024, the quantization width Q is determined to be 4. .

Therefore, when the motion compensation mode is the intra-picture prediction mode, the dynamic range of the block of 10-bit precision image data from the field memory group 1 becomes
For example, as shown in FIG. 3A, 700 (= 1000-3
00), the block is quantized in the block quantizer 7 with a quantization width of 3 (divided by 3),
Dynamic range is within 8 bits (in this case, 233
(= 700/3, but round down the decimal point)).

As described above, when the motion compensation mode is the intra-picture prediction mode, the dynamic range of the block of the 10-bit precision image data from the arithmetic unit 4 is 8
When the number of bits exceeds (256 or more), the block quantizer 7 converts the block of 10-bit precision image data from the calculator 4 into a block of 8-bit precision image data, and the difference signal encoder It is output to 8.

On the other hand, when the motion compensation mode is the inter-picture prediction mode (either forward prediction, backward prediction or bidirectional prediction mode), in the representative value calculation circuit 5, the calculation output of the calculator 3 (in this case, the field The maximum value and the minimum value of the difference data between the block of 10-bit precision image data from the memory group 1 and the predicted image are detected, and, for example, the average value (= (maximum value + minimum value) / 2, where the decimal point The following (rounded down) is output to the arithmetic units 4 and 11 as the representative value data of the block of the 10-bit precision image data output from the arithmetic unit 3.

In this case, the representative value data of the block is not the average value of the maximum value and the minimum value of the block (pixels in the block) (hereinafter referred to as the average value of the block), but the minimum value of the block. It can be zero.
However, in the case where the motion compensation mode is the inter-picture prediction mode, the image data is set to -255 to 2
Since it has to be represented by 9 bits in the range of 55, when the representative value data of the block is the average value of the block, the dynamic range of the block can be maximized.

Then, in the arithmetic unit 4, the block of the 10-bit precision image data output from the arithmetic unit 3 (each pixel data in the block) and the representative value data of the block output from the representative value calculation circuit 5, that is, The difference from the average value of the block is calculated and output to the block quantizer 7.

Therefore, in this case, the block quantizer 7 outputs to the block quantizer 7 blocks of image data having the same absolute value of the maximum value and the minimum value.

That is, when the motion compensation mode is the inter-picture prediction mode, for example, as shown in FIG.
The maximum value or the minimum value of the block of 10-bit precision image data output from is -155 or 35, respectively.
If it is 5, the average value 100 (= (-155 + 355) /
2) is subtracted from -255 (= -155-100) to 2
A block of image data having a value in the range of 55 (= 355-100) is output from the arithmetic unit 4 to the block quantizer 7, and is output from the arithmetic unit 3 as shown in FIG. 3B, for example. The maximum or minimum value of a block of 10-bit image data is 1000 or -10, respectively.
If 00, the average value is 0 (= (-1000 + 1000)
/ 2) -1000 (= -1000-0) to 1000
A block of image data having a value in the range (= 1000-0) will be output to the block quantizer 7.

At the same time, in this case, in the quantization width calculation circuit 6, the calculation output of the calculator 3 (in this case, the difference data between the block of image data of 10-bit precision from the field memory group 1 and the predicted image) is output. The maximum value and the minimum value are detected, and the dynamic range as the difference between them is calculated.

In the quantization width calculation circuit 6, the dynamic range of the block of 10-bit precision image data from the field memory group 1 is -255 to 25.
It is determined whether or not it is within 9 bits including 1 code bit that can be represented by the range of 5, and when it is determined that the dynamic range is within 9 bits, the quantization width Q is determined to be 1. , To the block quantizer 7.

Therefore, when the motion compensation mode is the inter-picture prediction mode, if the dynamic range of the block of 10-bit precision image data (difference data) output from the computing unit 3 is within 9 bits (for example, as shown in FIG. Two
(A)), the block quantizer 7 outputs the image data from the calculator 4 to the difference signal encoder 8 as it is.

That is, when the motion compensation mode is the inter-picture prediction mode, when the dynamic range of the block of the image data of 10-bit precision output from the arithmetic unit 3 is within 9 bits, it is output from the arithmetic unit 3. The bit precision of the block of image data is substantially 8 bits, and this block of image data of 8-bit precision is output to the differential signal encoder 8.

Further, when the motion compensation mode is the inter-picture prediction mode, in the quantization width calculation circuit 6, the dynamic range of the block of 10-bit precision image data from the field memory group 1 exceeds 9 bits (- Two
Quantized so that the dynamic range of the quantized output of the block quantizer 7 is 9 bits or less (less than 511) (within the range of -255 to 255) when it is determined that the value is outside the range of 55 to 255). The width Q is determined and output to the block quantizer 7.

That is, when the dynamic range of the block of 10-bit precision image data from the field memory group 1 is, for example, 512 or more and less than 1024, the quantization width Q is determined to be 2. Further, when the dynamic range is, for example, 1024 or more and less than 1536, the quantization width Q is determined to be 3, and the dynamic range is
For example, when it is 1536 or more and less than 2048, the quantization width Q is determined to be 4.

Therefore, when the motion compensation mode is the inter-picture prediction mode, the dynamic range of the block of the 10-bit precision image data from the calculator 4 is, for example, as shown in FIG.
As shown in (b), 2000 (= 1000 − (− 100
0), the block of the difference data is quantized in the block quantizer 7 with a quantization width of 4 (divided by 4), and the dynamic range is within 9 bits (in this case, 500 (= 2000/4). However, the numbers after the decimal point are rounded down)).

As described above, when the motion compensation mode is the inter-picture prediction mode, the dynamic range of the block of the 10-bit precision image data from the arithmetic unit 4 is 9
When the number of bits exceeds (512 or more), the block quantizer 7 converts the block of 10-bit precision image data from the calculator 4 into the block of 8-bit precision image data, and the difference signal encoder 8 is output.

As described above, when a block of 10-bit precision image data is input to the device, the difference signal encoder 8 is operated so that the block of 8-bit precision image data is input. In the device 4 and the block quantizer 8, the image data is converted into a block of 10-bit precision image data and a block of 8-beat precision image data.

By the way, an image having a bit precision higher than that of 8 bit precision, for example, 10 bit precision is converted into, for example, 8 ×.
When divided into small blocks such as 8 pixels or 16 × 16 pixels, the dynamic range in each block is generally not large (when the motion compensation mode is the intra-picture prediction mode, the dynamic range of the block is within 8 bits). When the motion compensation mode is the inter-picture prediction mode, the block dynamic range is often within 9 bits). Further, when the motion compensation mode is the inter-picture prediction mode, the difference between the block of the image data and the predicted image is calculated by the arithmetic unit 3, so that the dynamic range of the difference block rarely exceeds 9 bits.

Therefore, by the subtraction processing of the representative value data from the block of the image data in the arithmetic unit 4, the image data of 10-bit precision is converted into the image data of 8-bit precision without damaging the information in most cases. be able to.

That is, even if the apparatus is configured without providing the quantization width calculation circuit 6 and the block quantizer 7,
It is possible to convert 10-bit precision image data into 8-bit precision image data without damaging the image.

Further, in this case, for example, the quantization width S in the differential signal encoder 8 is made small (fine), and the image can be encoded while suppressing the deterioration of the image quality.

When the dynamic range of a block of image data is large (when the motion compensation mode is the intra-picture prediction mode, the dynamic range of the block exceeds 8 bits, and when the motion compensation mode is the inter-picture prediction mode, When the dynamic range exceeds 9 bits), the image data is quantized by the block quantizer 7 as described above, and the resolution when the image data is decoded is slightly deteriorated.

However, since the luminance discrimination degree of human eyes is low in a portion having a large dynamic range such as a contour portion of an image, the deterioration (deterioration) of the resolution in the level direction due to the above-mentioned quantization is given to the viewer. It is thought that there will be little impact.

The image data converted from 10-bit precision to 8-bit precision is supplied to the differential signal encoder 8.
Then, in the differential signal encoder 8, quantized, D
It is CT processed and supplied to the VLC circuit 15.

The VLC circuit 15 includes a differential signal encoder 8
In addition to the block of the image data that has been quantized by the DCT process, the representative value data or the quantization width Q of the block is supplied from the representative value calculation circuit 5 or the quantization width calculation circuit 6, respectively, and the motion vector and The motion compensation mode is provided by the circuit in motion estimation. Further, the VLC circuit 15 is supplied with the quantization width S in the differential signal encoder 8. Then, the VLC circuit 15 has a block of image data whose bit precision is converted from 10-bit precision to 8-bit precision, representative value data of the block,
The quantization widths Q and S, the motion vector, and the motion compensation mode are variable-length coded and output via a transmission buffer (not shown).

In this case, in the VLC circuit 15,
The representative value data of the block and the quantization width Q are added to the header of each block of the encoded image data.

The representative value data of the block and the quantization width Q are not added to the header of the block as described above, but as a layer higher than the block,
For example, it can be added to the header of the macroblock layer or the picture layer together with the representative value data and the quantization width Q of other blocks belonging to the macroblock or the picture layer.

The bit stream output from the VLC circuit 15 is multiplexed with the coded audio signal, sync signal, etc., further, an error correction code is added, and after being subjected to a predetermined modulation, the laser beam is outputted. As a result, it is recorded on the master disk (not shown) as uneven pits. Then, a stamper (not shown) is formed by using this master disk, and by the stamper,
The duplicate disks 17 (for example, optical disks) will be mass-produced.

Quantized by the difference signal encoder 8,
The DCT-processed image data is supplied to the differential signal decoder 10 when it is I or P picture data.

Although not shown, the differential signal decoder 9 is composed of an inverse quantizer 113 and an inverse DCT circuit 114 as in the case of FIG. 20, in which the data from the differential signal encoder 8 is used. The (quantized DCT data) is inversely quantized with the same quantization width as the quantization width S in the difference signal encoder 8, and further inverse DCT processing is performed.

The image data output from the differential signal decoder 9 is input to the block dequantizer 10, where
Dequantization is performed with the same quantization width as the quantization width Q in the block quantizer 7 output from the quantization width calculation circuit 6. That is, in the block inverse quantizer 10, the image data output from the differential signal decoder 9 is multiplied by the same quantization width as the quantization width Q in the block quantizer 7, and output to the calculator 11. It

In the arithmetic unit 11, the representative value calculation circuit 5 is added to the image data output from the block dequantizer 10.
The representative value data, which is the same as the representative value data subtracted from the arithmetic output of the arithmetic unit 3 in the arithmetic unit 4, is added and output to the arithmetic unit 12.

As a result, the arithmetic unit 12 has the same (more accurately, as the block of image data of 10-bit precision before being converted into the block of image data of 8-bit precision by the arithmetic unit 4 and the block quantizer 7). , The same block of image data is supplied because the quantization error is included.

The arithmetic unit 12 is also supplied with the prediction image which has already been decoded and motion-compensated by the motion compensation circuit 14.
There, the predicted image and the image data from the arithmetic unit 11 are added and decoded into the original 10-bit precision image data (10-bit precision image data before being encoded). The decoded 10-bit precision image data is supplied to and stored in the field memory group 13.

The field memory group 13 is designed to be able to store 10-bit precision image data, and stores the decoded 10-bit precision image data from the arithmetic unit 12. The already-decoded 10-bit image data stored in the field memory group 13 is read based on the timing control signal generated by the field memory controller 16 in response to the input image synchronization signal, and the motion compensation circuit At 14, motion compensation is performed according to the motion vector from the motion prediction circuit 2, and the motion is supplied to the computing units 3 and 12.

That is, the predicted image is supplied to the arithmetic units 3 and 12.

As described above, since the 10-bit precision image data is converted into the 8-bit precision image data, the motion prediction circuit 2 for the 8-bit precision image data, the differential signal encoder 8, and An image encoding device using 9 or the like can encode an image with 10-bit precision with almost no deterioration in image quality.

Next, FIG. 4 is a block diagram showing the configuration of an embodiment of the image decoding apparatus of the present invention. In the figure, FIG.
The same reference numerals are given to the portions corresponding to the case of. This image decoding apparatus is adapted to be able to decode the image coded by the image coding apparatus shown in FIG.

That is, in this image decoding apparatus, first, the image data whose bit precision is converted from 10-bit precision to 8-bit precision is reproduced from the disk 17 (FIG. 1), and the reproduced bit stream is converted into the inverse VLC circuit 21. Is supplied to. The inverse VLC circuit 21 variable-length-decodes the bit stream, and, for example, representative value data of a block added to a header of each block of image data or a quantization width Q.
Is supplied to the calculator 24 or the block dequantizer 23, and the motion vector and the motion compensation mode are supplied to the motion compensation circuit 27.

Further, the inverse VLC circuit 21 outputs a timing pulse to the memory controller 28 each time the variable length decoding of the data corresponding to one image (one screen) is completed. In the memory controller 28, a timing control signal is supplied to the field memory group 26 in response to this timing pulse, whereby the timing of reading image data from the field memory group 26 is controlled. It is done like this.

Further, the inverse VLC circuit 21 sequentially supplies the block of the variable length decoded image data and the quantization width S to the differential signal decoder 22.

Although not shown, the differential signal decoder 22 is composed of an inverse quantizer 121 and an inverse DCT circuit 122 as in the case of FIG. 21, in which the inverse VL is used.
A block of image data (quantized DCT data) output from the C circuit 21 and having bit precision converted from 10-bit precision to 8-bit precision has a quantization width S output from the inverse VLC circuit 21. Inverse quantization is performed and inverse DCT processing is performed.

Since the bit precision of the image data input / output to / from the differential signal decoder 22 is converted to 8-bit precision, the differential signal decoder 22 shown in FIG.
In the same manner as in the case of, the 8-bit precision image data can be inversely quantized, and further the inverse DCT processing can be performed.

The block of image data output from the differential signal decoder 22 is input to the block inverse quantizer 23, where the block quantizer 7 (FIG. 1) output from the inverse VLC circuit 21. Quantization width Q when quantized
Is inversely quantized with the same quantization width as. That is, in the block dequantizer 23, the image data output from the differential signal decoder 22 is multiplied by the same quantization width as the quantization width Q when quantized by the block quantizer 7. .

As a result, the dynamic range of the block of the image data is converted into the same value (nearly the same) as before the coding by the image coding apparatus of FIG.

The block of image data inversely quantized by the block inverse quantizer 23 is supplied to the calculator 24. The arithmetic unit 24 adds the representative value data from the inverse VLC circuit 21 and the block of image data (each pixel data in the block) from the block inverse quantizer 23, and outputs the sum to the arithmetic unit 25.

As a result, as in the case of the arithmetic unit 12 of the image coding apparatus of FIG. 1, the arithmetic unit 25 is converted into a block of 8-bit precision image data by the arithmetic unit 4 and the block quantizer 7. The block of the image data which is the same (almost the same) as the block of the image data of 10-bit precision before the operation is supplied.

That is, in the block dequantizer 23 and the calculator 24, the bit precision of the image data converted from the 10-bit precision to the 8-bit precision by the image encoding device of FIG. 1 becomes 10-bit precision. It is converted and supplied to the calculator 25.

The arithmetic unit 25 also receives the already decoded I
Alternatively, a prediction image in which the image data of the P picture is motion-compensated by the motion compensation circuit 27 is supplied, in which the prediction image and the image data from the arithmetic unit 24 are added to obtain the original 1
The image data is decoded into 0-bit precision image data (10-bit precision image data before being encoded). This decrypted one
Image data with 0-bit precision is stored in the field memory group 26.
Are stored and stored in.

The field memory group 26 is adapted to be able to store 10-bit precision image data, and stores the decoded 10-bit precision image data from the calculator 25. Of the already decoded 10-bit image data stored in the field memory group 26, I
Alternatively, the image data of the P picture is read out based on the timing control signal generated by the field memory controller 28 corresponding to the synchronization signal of the input image, and is converted into the motion vector from the motion prediction circuit 2 in the motion compensation circuit 14. Correspondingly motion-compensated and supplied to the calculator 25.

That is, the predicted image is supplied to the calculator 25.

The image data stored in the field memory group 26 is output to the output terminal based on the timing control signal from the field memory controller 28. Then, the image data output to the output terminal is subjected to predetermined processing such as D / A conversion processing, and is supplied to a display (not shown) or the like for display.

As described above, by the image decoding device using the differential signal decoder 22 for 8-bit image data, the image is encoded by the image encoding device of FIG.
It is possible to decode an image with 10-bit precision that is higher than 8-bit precision.

Note that in the image coding apparatus shown in FIG. 1 (the image decoding apparatus shown in FIG. 4), not only an image having a bit precision of 10 bits but also a bit precision of 11 bits or 11 bits, for example.
An image such as bit precision can be encoded (decoded).

Next, FIG. 5 is a block diagram showing the configuration of the second embodiment of the image coding apparatus of the present invention. This image encoding device is provided with a block quantization width buffer 33 and an intra-block representative value buffer (hereinafter referred to as a buffer) 34, and also has a computing unit 4 or 11
Instead of being provided with an in-block representative value differencer (hereinafter referred to as a representative value differencer) 31 or an in-block representative value adder (hereinafter referred to as a representative value adder) 32, respectively, It is configured similarly to the image coding apparatus in FIG.

This image encoding device is capable of encoding an image in which the bit precision of the image and the parameter block size described later change in arbitrary area units.

That is, in this image coding apparatus, from a processing circuit (not shown), the bit precision (code of the bit precision) of the image that changes in an arbitrary area unit such as one screen, one GOP, or one sequence, The parameter block size (parameter block size code) (in FIG. 5, shown as block size) is supplied.

The bit precision of the image and the parameter block size are independent of each other and can be changed not in the same area unit but in different area units. That is, the bit precision of the image or the parameter block size can be changed, for example, for each screen or 1 GOP.

Here, the parameter block size is
It means the size of a unit area (hereinafter referred to as a parameter block) of an image in which beat accuracy is converted. Therefore,
Parameter block size is, for example, 16x16 or 8x
If it is set to 8, the representative value data or the quantization width Q is calculated in the representative value calculation circuit 5 or the quantization width calculation circuit 6 for each of the 16 × 16 pixel block (parameter block) and the 8 × 8 pixel block (parameter block). Each will be decided.

The parameter block size and the block size of a processing unit for coding an image are set to have an integral multiple relationship. The block size of the processing unit to be encoded is, for example, 8 × 8.
In the case of pixels, for example, as shown in FIG. 6A, the parameter block size is four times as large as the size of a block (a block shown by a solid line in FIG. 6A) of a processing unit for encoding an image. When set to, in the representative value calculation circuit 5 or the quantization width calculation circuit 6, in the four blocks that refine the parameter block (four blocks surrounded by a dotted line in FIG. 6A) The representative value data or the quantization width Q is determined based on the maximum value and the minimum value of the pixel, and the representative value differentiator 31 and the representative value adder 32, or the block quantizer 7 and the block dequantizer 8 are provided. Each is supplied.

The representative value differentiator 31 corresponds to the arithmetic unit 4 in FIG. 1. In the representative value differentiator 31 and the block quantizer 7, the block from the arithmetic unit 3 (input to the device). The bit precision of the (image) is converted to the bit precision of the differential signal encoder 8 (8-bit precision in this embodiment) in parameter block units.

That is, in the case of FIG. 6A, in the representative value differentiator 31 and the block quantizer 7, it is determined from a block of 16 × 16 pixels consisting of four blocks of 8 × 8 pixels as parameter blocks. Based on the representative value data and the quantization width Q, the four 8 ×
The bit precision of all blocks of 8 pixels is sequentially converted to 8-bit precision in the same manner as in the image encoding device of FIG. 1, and encoded by the differential signal encoder 8.

The representative value data or the quantization width Q determined for each parameter block is supplied to the buffer 34 or 33 and stored therein. In the representative value data stored in the buffer 34 or 33, the parameter block (in this case, all four blocks of 8 × 8 pixels) is encoded by the differential signal encoder 8 and supplied to the VLC circuit 15. , To the VLC circuit 15.
Then, in the VLC circuit 15, the bit precision and the parameter block size input to the device, the buffer 3
The quantization width Q or the representative value data from 3 or 34, the motion vector and the motion compensation mode from the motion prediction circuit 2 are variable-length coded, output to the disk 17 via a transmission buffer (not shown), and recorded. It

In this case, in the VLC circuit 15, the representative value data of the parameter block and the quantization are added to the 16 × 16 pixel block consisting of four 8 × 8 pixel blocks as parameter blocks, that is, the header of each macroblock. A width Q is added.

For example, in an image that does not change significantly,
A change in the dynamic range of a plurality of adjacent blocks is small, and in such an image, after converting the bit precision of the plurality of adjacent blocks with the same representative value data and the quantization width Q as described above, Block (8 x
The code amount can be reduced by adding the representative value data and the quantization width Q to the header of each parameter block including a plurality of adjacent blocks instead of the header of each block of 8 pixels). .

Further, when the size of the block of the processing unit to be encoded is 8 × 8 pixels, for example, FIG.
As shown in (b), when the size four times the parameter block size is set to the size of the block of the processing unit for encoding the image (the block indicated by the solid line in FIG. 6B) , The representative value calculation circuit 5 or the quantization width calculation circuit 6 has four parameter blocks (see FIG.
In (b), the representative value data or the quantization width Q is sequentially determined for each of the four blocks indicated by the dotted line, and the representative value differencer 31 and the representative value adder 32, or the block quantizer 7 and the block dequantizer It is supplied to each rectifier 8.

Then, in the representative value differentiator 31 and the block quantizer 7, the representative value data and the quantization width Q determined for each of the four 4 × 4 pixel blocks as parameter blocks The bit precision of a block of 8 × 8 pixels constitutes the block 4
For each 4 × 4 pixel block (parameter block), 8 blocks are generated in the same manner as in the image coding apparatus of FIG.
Each is converted into bit precision and encoded by the differential signal encoder 8.

Also, the four representative value data or the quantization width Q determined for each parameter block are stored in the buffer 3
4 and 33 are sequentially supplied and stored respectively. In the representative value data or the quantization width Q stored in the buffer 34 or 33, a block of 8 × 8 pixels (block of a processing unit for encoding an image) composed of four parameter blocks is a difference signal encoder. Coded with 8, V
When supplied to the LC circuit 15, it is read out to the VLC circuit 15. Then, in the VLC circuit 15, the bit precision and the parameter block size input to the device, the quantization width Q or the representative value data from the buffer 33 or 34, and the motion vector and the motion compensation mode from the motion prediction circuit 2 are variable length. The data is encoded and output to the disk 17 via a transmission buffer (not shown) for recording.

In this case, in the VLC circuit 15, the representative value data and the quantization width Q of the four parameter blocks are added to the header of each block of 8 × 8 pixels, which is composed of four 4 × 4 pixel blocks as parameter blocks. Is added.

For example, in an image that changes drastically, the dynamic range of blocks is large, and in such an image, for each region block (parameter block) smaller than the block of the processing unit for encoding the image as described above, By determining the representative value data and the quantization width Q and converting the bit precision, it is possible to prevent deterioration of the resolution of the image quality.

On the other hand, the output of the differential signal encoder 8 is VL
The signal is supplied not only to the C circuit 15 but also to the differential signal decoder 10, and the same processing as in the case of FIG. 1 is performed.

However, the representative value adder 32 corresponds to the arithmetic unit 11 in FIG. 1, and in the representative value adder 31 and the block dequantizer 10, the block from the differential signal decoder 9 is changed. The bit precision of is converted into the bit precision of the image input to the device in units of parameter blocks.

In this image coding apparatus, the parameter block size and the bit precision of the image are set to 1
When changing in units of screen, 1 GOP, or 1 sequence area, information indicating that is added to the header of the picture layer, the header of the GOP layer, or the sequence header, respectively.

Further, in this image encoding device, when the bit precision input to the device is the same as the bit precision of the differential signal encoder 8 (in this embodiment, 8-bit precision).
The representative value calculation circuit 5 outputs 0 as representative value data, and the quantization width calculation circuit 6 outputs 1. Therefore, in this case, the representative value differencer 31,
The block quantizer 7, the block dequantizer 10, and the representative value adder 32 are substantially inoperative.

Next, FIG. 7 is a block diagram showing the configuration of the second embodiment of the image decoding apparatus of the present invention. In the figure,
The same reference numerals are given to the portions corresponding to the case of. This image decoding apparatus is adapted to be able to decode the image coded by the image coding apparatus shown in FIG. 5 (or the image coding apparatus shown in FIG. 1).

That is, in this image decoding apparatus, the bit stream recorded therein is reproduced from the disk 17 (FIG. 1), and the reproduced bit stream is supplied to the inverse VLC circuit 21. The inverse VLC circuit 21 performs variable length decoding on the bit stream, and separates the bit precision, the parameter block size, the representative value data, the quantization width Q, the motion vector, and the motion compensation mode.

Then, the inverse VLC circuit 21 outputs the representative value data or the quantization width Q to the buffer 42 or 43 for storage, and outputs the parameter block size to the block inverse quantizer 23 and the representative value adder 41. To do.

Further, the inverse VLC circuit 21 sequentially supplies the variable length decoded image data block and the quantization width S to the differential signal decoder 22.

In the differential signal decoder 22, the block of the image data from the inverse VLC circuit 21 is inversely quantized with the quantization width S, and further inverse DCT processing is performed to obtain the block inverse quantizer 23 and the representative value adder. It is sequentially output to 41.

The representative value adder 41 corresponds to the arithmetic unit 24 of the image decoding apparatus shown in FIG. 4, and in the representative value adder 41 and the block dequantizer 23, the differential signal decoder 22 is used. The bit precision of the data from (the bit precision of the differential signal decoder 22, in the present embodiment, 8 bit precision) is the bit precision of the original image in units of parameter blocks (coded by the image coding apparatus of FIG. 5). Bit precision of the image before being converted).

Here, the size of the block of the processing unit for decoding the image in this image decoding device is the same as the block of the processing unit for coding the image in the image coding device of FIG. That is, this image decoding apparatus is designed to decode an image in block units of 8 × 8 pixels.

Therefore, as in the case of the image coding apparatus of FIG. 5, the block size of the processing unit for decoding an image and the parameter block size output from the inverse VLC circuit 21 have an integral multiple relationship. It is designed to be set as follows.

For example, as shown in FIG. 6A, when the parameter block size is set, in the block dequantizer 23 and the representative value adder 41, four 8 × 8 pixels as parameter block blocks are set. The bit precision of all blocks is sequentially converted to the bit precision of the original image by the same representative value data or the quantization width Q stored in the buffer 42 or 43, respectively.

Further, for example, as shown in FIG. 6B, when the parameter block size is set, in the block dequantizer 23 and the representative value adder 41,
The representative value data or the quantization width Q of the four 4 × 4 pixel blocks is output from the inverse VLC circuit 21, and the buffer 4
After being stored in 2 or 43 respectively, conversion is performed to the bit precision of the data from the differential signal decoder 22.

That is, in this case, the block inverse quantizer 23
In the representative value adder 41, the bit precision of the block of 8 × 8 pixels, which is the processing unit for decoding the image, is 4
Each block (parameter block) of × 4 pixels is converted into the bit precision of the original image based on the representative value data and the quantization width Q.

The data whose bit precision has been converted by the block dequantizer 23 and the representative value adder 41 is input to the calculator 24, and is decoded in the same manner as in the image decoding apparatus shown in FIG.

In this image decoding device, the inverse VLC is used.
When the bit precision separated by the circuit 21 is the same as the bit precision of the differential signal encoder 22 (in the present embodiment,
(8-bit precision), the inverse VLC circuit 21 outputs 0 as the representative value data and 1 as the quantization width Q. Therefore, in this case, the block dequantizer 23 and the representative value adder 41
Would essentially not work.

Next, FIG. 8 is a block diagram showing the configuration of the third embodiment of the image decoding apparatus of the present invention. This image decoding apparatus is configured similarly to the image decoding apparatus shown in FIG. 7, except that a block quantizer 51 is provided between the representative value adder 41 and the arithmetic unit 24.

Therefore, the image decoding apparatus can decode the image coded by the image coding apparatus of FIG. 5 (or FIG. 1) or the like.

Furthermore, in this image decoding device,
The block quantizer 51 configured similarly to the block quantizer 7 of the image coding apparatus shown in FIG.
Based on the bit precision of the original image, which is supplied from 1,
The bit precision of the image data output from the representative value adder 41 is converted into a predetermined bit precision (for example, 8-bit precision).

That is, when the bit precision of the image data output from the representative value adder 41 is, for example, 10 bit precision (when the bit precision of the original image is 10 bit precision), the inverse VLC circuit 21. From, the code indicating the 10-bit precision is supplied to the block quantizer 51.

Here, in order to convert a 10-bit precision image into, for example, an 8-bit precision image, the 10-bit precision image (all pixel data of the image) is converted into 2 of 2 (= 10-
8) Divide by the power. To convert a 10-bit precision image into, for example, a 9-bit precision image, the 10-bit precision image (all the pixel data of the image) is converted into 2 of 1.
It may be divided by the (= 10−9) th power.

Therefore, when an image is to be decoded with 8-bit precision, for example, when a code indicating 10-bit precision is output from the inverse VLC circuit 21, the representative value adder 41 in the block quantizer 51 is used. The output 10-bit precision image data is divided by 2 ² and converted into 8-bit precision image data.

Therefore, the image can be decoded with the bit precision desired by the user.

Further, in this case, the field memory group 2
The image data of 8-bit precision output from the block quantizer 51 is supplied to 6 via the calculator 24.

Therefore, the field memory group 26 is
As in the case of the image decoding apparatus shown in FIG. 4, a memory that can store highly versatile 8-bit precision image data having a smaller number of bits than 10-bit precision image data can be provided. Cost reduction can be achieved.

Next, FIG. 9 is a block diagram showing the configuration of the fourth embodiment of the image decoding apparatus of the present invention. This image decoding device includes a bit determination circuit 61 and a display control circuit 6.
2 is provided, and instead of the inverse VLC circuit 124, an inverse VLC circuit 21 having the same configuration as in the image decoding apparatus is provided in FIG.
The image decoding device shown in FIG.

In this image decoding device, the inverse VLC is used.
In the circuit 21, the bit precision of the original image is separated from the bit stream and supplied to the bit precision determination circuit 61.

In the bit precision determination circuit 61, first, the bit precision supplied from the inverse VLC circuit 21 is the bit precision in the differential signal decoder 22 (the bit precision which the differential signal decoder 22 can process). (In this embodiment, 8-bit precision) is determined.

In the bit precision determination circuit 61, the inverse VL
When it is determined that the bit precision supplied from the C circuit 21 is equal to or less than 8 bit precision, this image decoding apparatus performs the same processing as in FIG. 20 to decode the image of 8 bit precision. It

In the bit precision judgment circuit 61,
When it is determined that the bit precision supplied from the inverse VLC circuit 21 is larger than the 8-bit precision, that is, when the bit precision of the original image corresponding to the bit stream supplied to the device is higher than the 8-bit precision, Since this image decoding device cannot perform the decoding process, the differential signal decoder 22 and the field memory group 26 are supplied with a control signal for stopping their operations.

When the differential signal decoder 22 and the field memory group 26 receive the control signal from the bit precision determination circuit 61, the operation thereof is stopped.

At the same time, the bit precision judgment circuit 61 controls the display control circuit 62 to output a message (for example, "decoding not possible") incapable of decoding the bit stream input to the device. The signal is output.

When the display control circuit 62 receives the control signal from the bit precision determination circuit 61, the message "decoding not possible" is output to a display (not shown) and displayed.

Therefore, when a bit stream corresponding to an image with a bit precision that cannot be decoded is input to the apparatus, the message "decoding not possible" is displayed.
The user can be notified that data that cannot be decrypted has been input to the device.

Next, FIG. 10 is a block diagram showing the configuration of the third embodiment of the image coding apparatus of the present invention. In the figure, parts corresponding to those in FIG. 1 are designated by the same reference numerals.

As shown in FIG. 11A, the separation circuit 72 is configured in the same way as the quantization width calculation circuit 6, the representative value calculation circuit 5, the calculator 4 or the block quantizer 7 of FIG. Quantization width calculation circuit 91, representative value calculation circuit 92,
It is composed of an arithmetic unit 93 or a block quantizer 94.

In the separation circuit 72, the bit precision is
For example, when a block of 10-bit precision image data is input, the quantization width calculation circuit 91 or the representative value calculation circuit 9
At 2, the quantization width Q or the representative value data of the block is calculated and output.

Then, in the separation circuit 72, the arithmetic unit 9 corresponding to the arithmetic unit 4 and the block quantizer 7 of FIG.
3 and the block quantizer 94 obtains the difference between the image data of 10-bit precision and the representative value data and quantizes the image data with the quantization width Q to obtain the same bit precision as that of the difference signal encoder 8. The image data is converted into bit precision image data and output.

The block of the quantization width Q, the representative value data, or the image data from the separation circuit 72 (FIG. 10) is the quantization width memory group 73a, the representative value memory group 73b, or the field memory group 73c of the memory section 73. Is output to and stored in each.

Here, when the bit precision of the block of the input image data is 10-bit precision and it is converted into the block of image data of 8-bit precision as described above, the quantization width Q is as shown in FIG. , As in
It has a value of either 2, 3, or 4.

Therefore, in this case, the quantization width Q can be represented by 2 bits, and the separation circuit 72 supplies the quantization width Q of 2 bits to the quantization width memory group 73a of the memory section 73. .

Therefore, in this case, the quantization width memory group 73
As a, a memory composed of a 2-bit width memory can be used.

Further, since the representative value data is, for example, the minimum value or the average value of the minimum value and the maximum value of the block of the input image data (pixel data in the block), the input image data Bit precision of block is 1
With 0-bit precision, it can be represented by at least 10 bits.

Therefore, in this case, as the representative value memory group 73b of the memory section 73, it is necessary to use a memory composed of a memory of 10-bit width.

Further, since the bit precision of the image data output from the separation circuit 72 is converted to the 8-bit precision as described above, the field memory group 73c of the memory section 73 has versatility. A high-bit width 8-bit memory can be used.

The memory unit 73 including the quantization width memory group 73a, the representative value memory group 73b, and the field memory group 73b corresponds to the field memory group 1 in FIG.

When a block of 10-bit precision image data is input to the device, the field memory group 1 of FIG. 1 has a memory capable of storing 10-bit precision image data as described above, that is, 10-bit precision. Had to use wide memory.

However, in the image coding apparatus of FIG. 10, the separation circuit 72 converts a block of 10-bit precision image data into 8-bit precision image data and outputs it to the field memory group 73c. For the 73c, an 8-bit wide memory having high versatility as described above can be used.

Therefore, the cost of the device can be reduced.

In the image coding apparatus shown in FIG. 10, the field memory group 73c for storing image data is used.
Other than the quantization width memory group 73a or the representative value memory group 7
3b is required, but the quantization width Q stored in each is
Alternatively, the representative value data is data (parameter) output from the separation circuit 72 one by one for each block, and therefore, the quantization width memory group 73a and the representative value memory group 73 are included.
The number of memories forming b is much smaller than the number of memories forming the field memory group 73 for storing pixel data, and the quantization width memory group 73a and the representative value memory group 73b are provided. The cost of the device is not affected.

The blocks of quantization width Q, representative value data, or 8-bit precision image data stored in the quantization width memory group 73a, representative value memory group 73b, or field memory group 73c, respectively, are stored in the memory controller 81.
On the basis of the timing control signal generated corresponding to the synchronizing signal of the input image, the signals are sequentially read in the order of encoding and supplied to the synthesizing circuit 74.

As shown in FIG. 11B, the synthesizing circuit 74 has a block dequantizer 95 or a calculator 96 corresponding to the block dequantizer 10 or the calculator 11 of the image coding apparatus of FIG. Composed of.

Block inverse quantizer 95 and calculator 96
In the above, the 8-bit precision image data from the field memory group 73c is dequantized with the quantization width Q from the quantization width memory group 73a, and the representative value data from the representative value memory group 73b is added. Thus, the image data is converted into image data of 10-bit precision similar to the image data before being input to the separation circuit 72 and is output.

The 10-bit precision image data output from the synthesizing circuit 74 is input to the separating circuit 75 configured similarly to the separating circuit 72 via the arithmetic unit 3. In the separation circuit 75, as in the case of the separation circuit 72, 1
A block of 0-bit precision image data is converted into a block of 8-bit precision image data, a quantization width Q, and representative value data, and is output.

The 8-bit precision image data output from the separation circuit 75 is input to the differential signal encoder 8 designed as a circuit for processing 8-bit precision image data.
There, it is encoded (DCT processed and quantized).

The block of image data encoded by the differential signal encoder 8 is output from the quantization circuit Q and the representative value data output from the separation circuit 75, and the motion prediction circuit 2 via the motion compensation circuit 71. It is input to the encoder 76 together with the motion vector and the motion compensation mode that have been selected, where it is variable-length coded and output via a transmission buffer (not shown).

The bit stream output from the encoder 76 is the same as in the case of FIG.
(Not shown in FIG. 10) or the like.

In the encoder 76, the representative value data of the block and the quantization width Q are added to the header of each block of the encoded image data.

The block of image data encoded by the differential signal encoder 8 is also supplied to the differential signal decoder 9 where it is decoded (dequantized and inverse DCT processed) and separated. The quantization width Q and the representative value data output from the circuit 75 are input to a combining circuit 77 configured similarly to the combining circuit 74.

In the synthesizing circuit 74, the image data (8-bit precision image data) from the differential signal decoder 9 is converted into a 10-bit precision image based on the quantization width Q and the representative value data from the separating circuit 75. Converted to data, arithmetic unit 1
2 is output.

The arithmetic unit 12 is also supplied with a prediction image which has already been decoded and motion-compensated by the motion compensation circuit 71, which will be described later, in which the prediction image and the image data from the synthesizing circuit 77 are added, The original image data of 10-bit precision (image data of 10-bit precision before encoding) is decoded to be the same (almost the same) image data. This decoded 10-bit precision image data is supplied to a separation circuit 78 configured similarly to the separation circuit 72, and is converted again into 8-bit precision image data, quantization width Q, and representative value data. Is output.

The 8-bit precision image data, the quantization width Q, or the representative value data from the separation circuit 78 is stored in the memory unit 7.
9 field memory group 79c, representative value memory group 79
b, or supplied to and stored in the quantization width memory group 79a.

Quantization width memory group 79a of the memory unit 79,
Representative value memory group 79b or field memory group 79
c corresponds to the field memory group 13 in FIG. 1, and is configured similarly to the quantization width memory group 73a of the memory unit 73, the representative value memory group 73b, or the field memory group 73c.

Therefore, the field memory group 79c can be constituted by a memory of 8-bit width having high versatility, and as described above, the memory of the quantization width memory group 79a and the representative value memory group 79b can be constituted. Since the number is small, the cost of the device can be reduced.

On the other hand, in the motion predicting circuit 2, the motion vector of the image data output from the synthesizing circuit 74 is detected, and the motion compensation mode of the image data is determined and supplied to the motion compensating circuit 71.

The motion compensation circuit 71 moves the read address of the memory unit 79 from the position corresponding to the position of the block currently output by the motion prediction circuit 2 based on the motion compensation mode from the motion prediction circuit 2. The data is read out after being shifted by the amount corresponding to the vector and supplied to the synthesizing circuit 80.

The timing of reading data from the memory section 79 is controlled based on the timing control signal generated by the memory controller 81 in response to the synchronizing signal of the input image.

Here, in this image coding apparatus,
Unlike the image coding apparatus in FIG. 1, the field memory group 1
The memory unit 79 corresponding to 3 (FIG. 1) does not store the image data itself obtained by decoding the 10-bit precision image data input to the device, but the 10-bit precision image input to the device. An image whose precision is converted to 8-bit precision is stored together with the quantization width Q and representative value data for each block.

Therefore, in the motion compensation circuit 71,
Corresponding to the motion vector from the motion prediction circuit 2, the quantization width memory group 79 a and the representative value memory group 79 of the memory unit 79.
b, or the address is output to the field memory group 79c, the block of 8-bit precision image data stored in the field memory group 79c is read, and the quantization width memory group 7 corresponding to the block is read.
The quantization width Q or the representative value data stored in the group 9a or the representative value memory group 79b is read and supplied to the synthesizing circuit 80 configured similarly to the synthesizing circuit 74.

In the synthesizing circuit 80, the quantization width Q read out from the quantization width memory group 79a or the representative value memory group 79b by the motion compensation circuit 71, respectively.
Alternatively, based on the representative value data, the field memory group 7
A block of 8-bit precision image data read from 9c is converted into 10-bit precision image data to generate a predicted image.

Therefore, in this image encoding device, when an image is divided into blocks of 8 × 8 pixels and encoded, the motion vector detected by the motion prediction circuit 2 is, for example, As shown in FIG.
4, 4), that is, the region (block) of the already-decoded image that is most similar to the block of the image data to be coded is, for example, as shown in FIG. × 8 pixel blocks A, B, C,
When 4 × 4 pixels are applied to each D, in the motion compensation circuit 71 and the synthesis circuit 80, according to the following equation,
The 8-bit precision image data in each of the blocks A, B, C, and D stored in the field memory group 79c of the memory unit 79 is converted into 10-bit precision image data, and the prediction image (each pixel data in the block of the prediction image is converted. P (i, j) (where i, j = 0, 1, 2, ...,
At 7, i, j indicate the position of the pixel)).

That is, the block block of 8 × 8 pixels stored in the quantization width memory group 79c (block =
The quantization width Q of (A, B, C, D) is set to Q (block), and the block block stored in the representative value memory group 79b is stored.
The representative value data of ck is R (block), and each pixel data in the block block of 8 × 8 pixels stored in the field memory group 79c is p (bl).
ock, i, j), in the motion compensation circuit 71 and the synthesis circuit 80, the expression P (i, j) = R (A) + Q (A) × p (A, i, j) (however, 0 ≦ i <4 and 0 ≦ j <4) P (i, j) = R (B) + Q (B) × p (B, i, j) (where 4 ≦ i <8 and 0 ≦ j <4 ) P (i, j) = R (C) + Q (C) × p (C, i, j) (where 0 ≦ i <4 and 4 ≦ j <8) P (i, j) = R ( D) + Q (D) × p (D, i, j) (where 4 ≦ i <8 and 4 ≦ j <8), each pixel data P in the block of the prediction image
(I, j), that is, the predicted image is generated.

This predicted image is supplied to the arithmetic unit 12,
Therefore, as described above, the original image data is decoded and is supplied to the arithmetic unit 3 to generate the difference data.

Thereafter, the above-mentioned processing is repeated to encode the image.

Next, FIG. 13 is a block diagram showing the configuration of the fifth embodiment of the image decoding apparatus of the present invention. In the figure, portions corresponding to those in FIG. 4 are designated by the same reference numerals. Further, in the figure, the numbers attached to the lines connecting the blocks indicate the number of bits of data transmitted by the lines.

The decoder 101 is the disk 17 (FIG. 1).
The reproduced bit stream reproduced from is subjected to variable length decoding, and bit precision, parameter block size, representative value data, quantization width Q, motion vector, and motion compensation mode are separated therefrom.

Then, the decoder 101 outputs the representative value data or the quantization width Q to the synthesizing circuit 102 configured similarly to the synthesizing circuit 74 of FIG. The quantization width S is sequentially supplied to the differential signal decoder 22.

In the differential signal decoder 22, the decoder 10
The block of image data from 1 is inversely quantized by the inverse quantizer 121 with the quantization width S, and the inverse DCT circuit 122
Inverse DCT processing is performed by and is sequentially output to the synthesis circuit 102.

In the synthesis circuit 102, the synthesis circuit 7 of FIG.
As in the case of 4, the 8-bit precision image data output from the differential signal decoder 22 is the decoder 101
Dequantization is performed with the quantization width Q output by the output and the representative value data output by the decoder 101 is added, and the same as the original image data (image data before encoding) is obtained. It is converted into prediction error data of 10-bit precision image data and output.

The prediction error data of 10-bit precision output from the synthesizing circuit 102 is supplied to the calculator 25, where it is added to the prediction image from the synthesizing circuit 104, which will be described later, to obtain image data of 10-bit precision. Be decrypted. This decoded image data is the separation circuit 7 of FIG.
It is input to the separation circuit 103 configured in the same manner as 2. In the separation circuit 103, a block of 10-bit precision image data is converted into a block of 8-bit precision image data, a quantization width Q, and representative value data in the same manner as in the separation circuit 72 of FIG. Is output.

The 8-bit precision image data, the quantization width Q, or the representative value data from the separation circuit 103 is stored in the field memory group 106c, the representative value memory group 106b, or the quantization width memory group 106a of the memory unit 106, respectively. Supplied and stored.

Quantization width memory group 106 of memory unit 106
a, the representative value memory group 106b, or the field memory group 106c corresponds to the field memory group 26 of FIG. 4, and the quantization width memory group 7 of the memory unit 73 of FIG.
3a, the representative value memory group 73b, or the field memory group 73c, respectively.

Therefore, the field memory group 106c is
It can be configured by an 8-bit wide memory having high versatility, and as described above, the quantization width memory group 106
Since the number of memories constituting the a and the representative value memory group 106b is small, the cost of the device can be reduced.

On the other hand, the motion compensation mode and the motion vector are supplied from the decoder 101 to the motion compensation circuit 108 having the same structure as the motion compensation circuit 71 of FIG.

The motion compensation circuit 108 moves the read address of the memory unit 106 from the position corresponding to the position of the block currently output by the decoder 101, based on the motion compensation mode from the decoder 101. The data is read out by shifting by the amount corresponding to the vector, and the synthesis circuit 1
Supply to 04.

Here, the inverse VLC circuit 21 outputs the timing pulse to the memory controller 107 at the timing of variable length decoding one image (one screen), and the memory controller 107 outputs this timing pulse. The memory unit 1 outputs the timing control signal corresponding to the pulse.
It is designed to supply to 06. The timing of reading data from the memory unit 106 is controlled based on the timing control signal from the memory controller 107.

Here, in this image decoding apparatus,
Unlike the image decoding apparatus in FIG. 4, the field memory group 2
6 in the memory unit 106 corresponding to 6 (FIG. 4)
The image data itself with 0-bit precision is not stored, but as described above, the separation circuit 103 stores 10 bits.
An image in which the precision of the bit precision image is converted to 8 bit precision is stored together with the quantization width Q and representative value data for each block.

Therefore, in the motion compensation circuit 108, the quantization width memory group 106a, the representative value memory group 106b, or the field memory group 106c of the memory unit 106 is respectively corresponding to the motion vector from the decoder 101. The address is output and the field memory group 106
A block of 8-bit precision image data stored in c is read, and the quantization width Q or representative value data stored in the quantization width memory group 106a or the representative value memory group 106b corresponding to the block is read. It is read out and supplied to the combining circuit 104 configured similarly to the combining circuit 74 of FIG.

Then, in the synthesizing circuit 104, based on the quantization width Q or the representative value data read from the quantization width memory group 106a or the representative value memory group 106b by the motion compensation circuit 108, the field memory group 106c. A block of 8-bit precision image data read from is converted into 10-bit precision image data to generate a predicted image.

This predicted image is supplied to the calculator 25,
Therefore, as described above, the decoded image data is generated by adding the output of the synthesizing circuit 102.

Thereafter, the above-described processing is repeated to decode the image.

The quantization width Q, representative value data, or 8-bit precision image data stored in the quantization width memory group 106a, the representative value memory group 106b, or the field memory group 106c of the memory unit 106 is Based on the timing control signal from the memory controller 107, it is read out to the synthesizing circuit 105 configured similarly to the synthesizing circuit 74 of FIG. 10, where it is converted into image data of 10-bit precision and output.

The image data output from the synthesizing circuit 105 is subjected to predetermined processing such as D / A conversion processing,
It is supplied and displayed on a display (not shown).

Note that the image decoding apparatus shown in FIG. 13 (see FIG.
0), the bit precision is 1
It is possible to decode (encode) not only an image with 0-bit precision but also an image with 9-bit precision or 11-bit precision, for example.

Next, the format of the bit stream output from the image coding apparatus of the present invention will be described with reference to FIG.
It will be described with reference to FIGS. 14 to FIG.
CCITT SG XV Working Party XV / 1 Experts group on AT
M Video Coding, INTERNATIONAL ORGANIZATION FOR STA
NDARDISATION ISO-IEC / JTC1 / SC29 / WG11 CODED REPRESEN
TATION OF PICTURE AND AUDIO INFORMATION, ISO-IEC / J
TC1 / SC29 / WG11 N0328,25-Nov-92 Test Model 3, Draft
The bitstream format described in Revision 1 (hereinafter referred to as the standard format) is L1
The format is shown by adding the portions indicated by L6 to L6.

First, the format shown in FIG. 14 is a format for indicating whether or not the encoded image is a normal 8-bit precision image, and the flag extension_bit_input (in the figure, is added to the sequence header of the standard format. , L1) is added.

If the encoded image is a normal 8-bit precision image, for example, 0 out of 0 and 1 is set in this flag extension_bit_input, and the encoded image is a normal 8-bit precision image. If the image is not a bit-precision image, for example, 1 of 0 and 1 is set.

Next, the format shown in FIG. 15 is a format in which the bit precision of the image or the parameter block size is made variable for each sequence, and a portion indicated by L2 is added to the sequence header of the standard format.

Of the portion indicated by L2, the code bit_accura
cy is, for example, a 3-bit code indicating the bit precision of the image. When the bit precision of the image in the sequence is, for example, 8, 6, 7, 9, 10, 11, ..., The code bit_accuracy is As a 3-bit code, for example, 000,001,010,100,100,101, ...
.. are set respectively.

Further, in the portion indicated by L2, the code bl
ock_size is, for example, a 2-bit code indicating the parameter block size, and the parameter block size in the sequence is, for example, 16 × 16, 8 × 8, 4 × 4,
Or if it is 32 × 32, the code block_size is
For example, 00, 01, 10, or 11 as a 2-bit code is set.

[0291] Note that from the if of the portion indicated by L2,
Part is the code bit_accuracy and block_size in {} when the flag extension_bit_input is true (1)
Means that is valid.

The format shown in FIG. 16 or 17 is a format in which the bit precision of an image or the parameter block size is made variable for each GOP or picture (one screen), and is the format of the standard format GOP layer or picture layer. To the header, portions similar to those shown in FIG. 15 except for the portion of the flag extension_bit_input of L2 shown by L3 or L4 are added.

When the bit precision or parameter block size of the image changes in different units as described above, the code indicating the bit precision or parameter block size of the image is only in the header corresponding to the changing unit. Described. That is, for example, when the bit precision of the image changes for each sequence and the parameter block size changes for each GOP, only the code bit_accuracy is described in {} of the if statement in the sequence header (FIG. 15). GOP layer header (Fig. 1
In 6), only the code block_size is described in {} of the if statement.

Next, the format shown in FIG. 18 is a format for transmitting the representative value data and the quantization width Q of the block belonging to the picture layer for each picture (one screen). A portion indicated by L5 is added to the header.

Of the portion indicated by L5, the code block_base
Is set to the representative value data of each block forming the picture.

Here, the number of bits of the representative value data of one block (the same number of bits as the number of bits of the image bit precision (the number of bits of the bit precision represented by the code bit_accuracy)) is set to A, and the pixels of one screen are displayed. If the number of blocks per screen that can be obtained from the number and the block size (block size represented by the code block_size) is n, the code block_base is set to n representative value data of A bits. Therefore, the code block_base is defined as a code of A × n bits in the header of the picture layer.

Furthermore, in the portion indicated by L5, the code bl
In ock_q_scale, the quantization width Q of each block forming the picture is set.

Here, if the quantization width Q of one block is represented by, for example, 3 bits (2 bits may be used as in the above-described embodiment, or 4 bits may be used), the code block_q_scale Since n quantization widths of 3 bits will be set in, the code block_q_scale in the header of the picture layer.
Is defined as a 3 × n bit code.

Next, the format shown in FIG. 19 is a format for transmitting, for each macroblock (16 × 16 pixel block), the representative value data of the blocks constituting the macroblock and the quantization width Q. A portion indicated by L6 is added to the header of the macro block layer of the standard format.

In this case, when the number of blocks forming the macroblock is m (block size (block size represented by the code block_size) is 16 × 16, 8 × 8, or 4 × 4, respectively, 1, 4, or 16), m pieces of A-bit representative value data are set in the code block_base. Therefore, in the header of the macroblock layer, the code block_base is set.
Is defined as an A × m bit code.

Further, in this case, since m pieces of 3-bit quantization widths are set in the code block_q_scale, the code block_q_scale is defined as a 3 × m-bit code in the header of the macroblock layer. Has been done.

The bit stream format is not limited to the above format, but the flag extension_bit_in
Only when changing the format, bit precision or parameter block size without put, for example, GOP
It is also possible to adopt a format in which information on bit precision or parameter block size is added to the layer header or the picture layer header.

As described above, when the image is encoded, the representative value data and the quantization width of the image data are determined, and the bit precision of the image is converted based on the representative value data and the quantization width Q. Therefore, it is possible to encode an image with higher bit precision than 8-bit precision using conventional hardware designed on the assumption that an image with 8-bit precision is input to and output from the device, for example. it can.

Also, in the case of decoding an image,
Since the image data, the representative value data, and the quantization width are extracted from the transmission data and the bit precision of the image is converted based on the representative value data and the quantization width Q, the encoding is performed as described above. An image with higher bit precision than 8-bit can be decoded using conventional hardware designed on the assumption that an image with 8-bit precision is input to and output from the device.

In the present embodiment, the 8-bit precision motion prediction circuit 2 detects the 8-bit precision motion vector by discarding the lower 2 bits of the 10-bit precision image motion vector. However, for example, the lower 2 bits of the motion vector of the image with 10-bit precision are rounded down to 4
The input 8-bit precision motion vector can be detected.

Furthermore, the motion prediction circuit 2 for 8-bit precision
Alternatively, a motion prediction circuit for 10-bit precision can be provided. In this case, since the accuracy of the motion vector is high, the image quality can be improved.

Further, in the image coding apparatus of this embodiment, the representative value data itself of the block is coded together with the image data, but for example, the difference between the representative value data of the adjacent blocks is calculated, and this difference is calculated. Information can be variable length coded together with image data as representative value data.

In this case, the image decoding device is configured to add the representative value data of the already decoded adjacent block to the difference information of the variable length decoded block to obtain the representative value data of the block. .

One screen is, for example, 8 × 8 pixels or 16 × 16.
When divided into small blocks such as pixel blocks, the difference between the representative value data of adjacent blocks is
It is smaller than the representative value data, and its variance is also smaller. Therefore, as described above, the code amount can be reduced by performing the variable length coding on the difference between the representative value data of the adjacent blocks.

Furthermore, in this case, if the variable length coding or variable length decoding method in the image coding apparatus or the image decoding apparatus is the same as that in the transmission of the DC component of MPEG1, the representative value data By performing variable-length coding or variable-length decoding on the difference, a new variable-length code table or a new variable-length decoding table is not required, and efficient processing can be performed.

In the image coding apparatus of this embodiment, the quantization width Q for each block is coded together with the image data. For example, the quantization width Q of the block coded immediately before, Quantization width information indicating whether or not the quantization width Q ′ of the currently encoded block is the same is encoded, and quantization is performed only when the quantization width Q and the quantization width Q ′ are not the same. The width Q'may be encoded.

In this case, the image decoding device refers to the quantization width information, and until the new quantization width Q ′ is transmitted, the quantization width Q is the same as the quantization width Q of the block decoded immediately before. When the decoding process is performed using the quantization width and a new quantization width is transmitted, the decoding process is performed using the quantization width Q ′.

One screen is, for example, 8 × 8 pixels or 16 × 16.
When divided into small blocks such as a block of pixels, the dynamic range of adjacent blocks hardly changes, and the quantization width Q also hardly changes. Therefore, as described above, the quantization width information indicating whether the quantization width Q of the immediately previous encoded block and the quantization width Q ′ of the currently encoded block are the same is encoded. Furthermore, the code amount can be reduced by coding the quantization width Q ′ only when the quantization width Q and the quantization width Q ′ are not the same.

Furthermore, in the present embodiment, the bit precision of the input image is set to 10-bit precision, and the image of this 10-bit precision is input to the differential signal encoder 8, the differential signal decoder 9,
The case of converting into an image of 8-bit precision as the bit precision that can be processed in Nos. 22 and 22 has been described, but the present invention is not limited to this. That is, the bit precision of an image can be converted into arbitrary bit precision, and can be encoded or decoded.

[0315]

According to the image coding method of the first aspect and the image coding apparatus of the twelfth aspect, the representative value data and the quantization width of the image data are determined for each predetermined block. , The difference between the image data and the representative value data is calculated, and the difference is quantized based on the quantization width to generate first quantized data. Then, the first quantized data is subjected to a predetermined conversion process, the conversion coefficient is quantized, and the second quantized data is quantized.
Quantized data is generated, and the representative value data, the quantization width, and the second quantized data are encoded. Therefore, an image can be encoded by converting its bit precision.

According to the image coding method of the eighth aspect, the representative value data and the quantization width of the image data are determined for each predetermined unit area, and the difference between the image data and the representative value data is calculated. , The difference is quantized based on the quantization width,
First quantized data is generated. Then, for each predetermined block, a predetermined conversion process is performed on the first quantized data, the conversion coefficient is quantized to generate second quantized data, and the representative value data, the quantization width, and the The quantized data of 2 is encoded. Therefore, since the representative value data and the quantization width can be changed for each predetermined unit area according to the complexity of the image, the deterioration of the image quality of the image can be prevented.

According to the image coding method of the tenth aspect and the image coding apparatus of the thirteenth aspect, the difference between the input image and the predicted image is calculated for each predetermined block,
First difference data is generated and representative value data and quantization width of the first difference data are determined. Then, the difference between the first difference data and the representative value data is calculated, the second difference data is generated, and the second difference data is quantized based on the quantization width to generate the quantized data. Further, the quantized data is inversely quantized based on the quantization width, the inverse quantized data is generated, the inverse quantized data and the representative value data are added, the locally decoded data is generated, and then the locally decoded data is generated. To generate a new predicted image. Therefore, the image
The bit precision can be converted for efficient coding.

According to the image decoding method of the fourteenth aspect and the image decoding apparatus of the twenty-first aspect, from the transmission data, the encoded image data and the representative value data set for each predetermined block are transmitted. , And the quantization width are extracted, and the encoded image data is inversely quantized to generate the first inverse quantized data. Then, a predetermined inverse transform process is performed on the first inverse quantized data, the inverse transformed data is inversely quantized based on the quantization width, and second inverse quantized data is generated. Data and representative value data,
Generates decoded image data. Therefore, the image signal whose bit precision has been converted can be decoded.

According to the image decoding method of the seventeenth aspect, the encoded image data, the representative value data set for each predetermined unit area, and the quantization width are extracted from the transmission data to obtain a predetermined value. For each block, the encoded image data is dequantized to generate first dequantized data. Then, the first inverse quantized data is subjected to predetermined inverse transform processing, and the inverse transformed data is inverse quantized for each predetermined unit area based on the quantization width to generate second inverse quantized data. Then, the second dequantized data and the representative value data are added to generate decoded image data. Therefore, it is possible to decode the encoded image by changing the representative value data and the quantization width for each predetermined unit area according to the complexity of the image.

According to the image decoding method of the nineteenth aspect and the image decoding apparatus of the twenty-second aspect, from the transmission data to the encoded image data, and the representative value data set for each predetermined block. , And the quantization width are extracted, the coded image data is inversely quantized based on the quantization width, the inverse quantized data is generated, the inverse quantized data and the representative value data are added, and the decoded difference data is obtained. To generate. Then, the decoded difference data and the predicted image are added to generate decoded image data. Therefore, it is possible to efficiently decode image signals that are encoded and have different bit precisions.

According to the image decoding apparatus of the twenty-third aspect, the coded image data and the accuracy information regarding the bit accuracy of the image data are extracted from the transmission data and coded based on the accuracy information. It is determined whether the image data can be decoded. Then, if the encoded image data is undecodable, the fact that the encoded image data is undecodable is notified. Therefore, it is possible to inform the user that the device does not support the bit precision of the input data.

According to the image decoding apparatus of the twenty-fourth aspect, from the transmission data, the encoded image data, the accuracy information regarding the bit accuracy of the image data, and the representative value data set for each predetermined block, And the quantization width are extracted, the encoded image data is dequantized, and the first
After the inverse quantized data is generated, the first inverse quantized data is subjected to a predetermined inverse transform process to generate the inverse transformed data.
Then, the inversely transformed data is inversely quantized based on the quantization width to generate second inversely quantized data, the second inversely quantized data and the representative value data are added, and the first decoded image data is obtained. To generate. Further, the first decoded image data is quantized based on the accuracy information to generate the second decoded image data. Therefore, an image signal with arbitrary bit precision can be obtained.

According to the image recording medium of the twenty-fifth aspect, the representative value data and the quantization width of the image data are determined for each predetermined block, the difference between the image data and the representative value data is calculated, and the difference is calculated. Quantization is performed based on the quantization width, first quantized data is generated, predetermined conversion processing is performed on the first quantized data, conversion coefficients are generated, the conversion coefficients are quantized, and second quantization is performed. A bitstream generated by generating data and encoding the representative value data, the quantization width, and the second quantization data is recorded. Therefore, it is possible to record an image signal with arbitrary bit precision.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of an image encoding device of the present invention.

FIG. 2 is a diagram illustrating a method of determining representative value data and a quantization width Q.

FIG. 3 is a diagram illustrating a method of determining representative value data and a quantization width Q.

FIG. 4 is a block diagram showing the configuration of an embodiment of an image decoding apparatus of the present invention.

FIG. 5 is a block diagram showing the configuration of a second embodiment of the image encoding device according to the present invention.

FIG. 6 is a diagram illustrating a parameter block.

FIG. 7 is a block diagram showing the configuration of a second embodiment of the image decoding apparatus of the present invention.

FIG. 8 is a block diagram showing a configuration of a third embodiment of the image decoding apparatus of the present invention.

FIG. 9 is a block diagram showing the configuration of a fourth embodiment of the image decoding apparatus of the present invention.

FIG. 10 is a block diagram showing the configuration of a third embodiment of the image encoding device of the present invention.

11 is a block diagram showing more details of the separation circuit 72 and the combination circuit 74 of the embodiment of FIG.

12 is a diagram for explaining the operation of the motion compensation circuit 71 in the embodiment of FIG.

FIG. 13 is a block diagram showing the configuration of a fifth embodiment of the image decoding apparatus of the present invention.

FIG. 14 is a diagram illustrating a format of a bitstream as encoded image data.

FIG. 15 is a diagram showing a format of a bitstream as encoded image data.

FIG. 16 is a diagram showing a format of a bitstream as encoded image data.

FIG. 17 is a diagram illustrating a format of a bitstream as encoded image data.

FIG. 18 is a diagram showing a format of a bitstream as encoded image data.

FIG. 19 is a diagram showing a format of a bitstream as encoded image data.

FIG. 20 is a diagram showing a configuration of an example of a conventional image encoding device.

FIG. 21 is a diagram showing a configuration of an example of a conventional image decoding device.

[Explanation of symbols]

 1 Field Memory Group 2 Motion Prediction Circuit 3, 4 Calculator 5 In-Block Representative Value Calculation Circuit 6 Block Quantization Width Calculation Circuit 7 Block Quantizer 8 Differential Signal Encoder 9 Differential Signal Decoder 10 Block Inverse Quantizer 11, 12 arithmetic unit 13 field memory group 13 14 motion compensation circuit 15 VLC circuit 16 field memory controller 21 inverse VLC circuit 22 differential signal decoder 23 block inverse quantizer 24, 25 arithmetic unit 26 field memory group 27 motion compensation circuit 28 memory controller 33, 34 buffer 42, 43 buffer 51 block quantizer 61 bit precision determination circuit 62 display control circuit 71 motion compensation circuit 72 separation circuit 73 memory unit 74 combination circuit 75 separation circuit 77 combination circuit 78 separation circuit 79 memory unit 80 Synthesis circuit 9 1 block quantization width calculation circuit 92 representative value calculation circuit in block 93 arithmetic unit 94 block quantizer 95 block inverse quantizer 96 arithmetic unit 102 synthesis circuit 103 separation circuit 104, 105 synthesis circuit 106 memory unit 108 motion compensation circuit

Claims (25)

[Claims] 1. A representative value data of image data and a quantization width are determined for each predetermined block, a difference between the image data and the representative value data is calculated, and the difference is quantized based on the quantization width. To generate first quantized data, subject the first quantized data to a predetermined conversion process, generate transform coefficients, quantize the transform coefficients, and generate second quantized data. An image coding method, wherein the representative value data, the quantization width, and the second quantization data are coded. 2. The image encoding method according to claim 1, wherein a difference between representative value data of image data in adjacent blocks is encoded as the representative value data. 3. The comparison information for further determining whether or not the quantization width of the image data in the immediately preceding encoded block and the quantization width of the image data in the currently encoded block are the same, Only when the quantization width of the image data in the immediately preceding encoded block and the quantization width of the image data in the currently encoded block are different, the quantization width of the image data in the currently encoded block is set. The image coding method according to claim 1, wherein the image coding is performed. 4. The image coding method according to claim 1, wherein the representative value data and the quantization width are added to the header of the block. 5. The image coding method according to claim 1, wherein the representative value data and the quantization width are added to a header of a layer higher than the block. 6. The quantization width is determined based on size information regarding the size of the block or precision information regarding bit precision of image data in the block, and the size information and the precision information are further encoded. The image encoding method according to claim 1, wherein the image encoding method is the image encoding method. 7. The image encoding method according to claim 1, wherein the representative value data is determined based on size information regarding the size of the block, and the size information is further encoded. 8. A representative value data and a quantization width of image data are determined for each predetermined unit area, a difference between the image data and the representative value data is calculated, and the difference is based on the quantization width. Quantize to generate first quantized data, perform predetermined transform processing on the first quantized data for each predetermined block, generate transform coefficients, quantize the transform coefficients, and An image coding method, wherein quantized data is generated, and the representative value data, the quantization width, and the second quantized data are coded. 9. The image encoding method according to claim 8, wherein the unit area and the size of the block have an integral multiple relationship. 10. A difference between an input image and a predicted image is calculated for each predetermined block, first difference data is generated, representative value data and a quantization width of the first difference data are determined, and Calculating a difference between the first difference data and the representative value data, generating second difference data, quantizing the second difference data based on the quantization width, and generating quantized data, Dequantize the quantized data based on the quantization width to generate dequantized data, add the dequantized data and the representative value data, generate locally decoded data, and based on the locally decoded data. And an image encoding method for generating a new predicted image. 11. A difference between the locally decoded data and the representative value data is calculated, the difference is quantized based on the quantization width to generate difference quantized data, the difference quantized data, the representative value data. , And a quantization width are stored in a storage unit, the difference quantization data, the representative value data, and the quantization width are read from the storage unit, and the new predicted image is generated. The described image coding method. 12. A representative value determining unit that determines representative value data of image data for each predetermined block, and a quantization width determining unit that determines a quantization width of the image data for each predetermined block, A difference calculating means for calculating a difference between the image data and the representative value data; a first quantizing means for quantizing the difference based on the quantization width to generate first quantized data; Second quantizing means for generating a second quantized data by subjecting the first quantized data to a predetermined conversion process, and quantizing the quantized data; and the representative value data, the quantization width, and the second quantized data. An image encoding device comprising: an encoding unit that encodes the image. 13. A first difference calculating means for calculating a difference between an input image and a predicted image for each predetermined block to generate first difference data, and the first difference for each predetermined block. Representative value determining means for determining representative value data of data, quantization width determining means for determining the quantization width of the first difference data for each of the predetermined blocks, the first difference data and the representative Second difference calculating means for calculating a difference between the value data and generating second difference data; and quantizing means for quantizing the second difference data based on the quantization width to generate quantized data. Dequantizing the quantized data on the basis of the quantization width to generate dequantized data; and adding the dequantized data and the representative value data to generate locally decoded data. Addition means to Based on the local decoded data, the image encoding device characterized by comprising a prediction means for generating a new predicted image. 14. From the transmission data, the encoded image data,
And representative value data set for each predetermined block,
And a quantization width are extracted, the coded image data is inversely quantized, first inverse quantized data is generated, and the first inverse quantized data is subjected to predetermined inverse transform processing,
Inverse transform data is generated, the inverse transform data is inversely quantized based on the quantization width, second inverse quantized data is generated, and the second inverse quantized data and the representative value data are added. An image decoding method characterized by generating decoded image data. 15. The image decoding method according to claim 14, wherein the representative value data is a difference between representative value data of image data in adjacent blocks. 16. A comparison is made as to whether or not the quantization width of image data in a block decoded immediately before and the quantization width of image data in a block to be decoded are the same from the transmission data. If the quantization width of the image data in the block decoded immediately before is further extracted from the information and the quantization width of the image data in the block currently being decoded is the same, the block decoded immediately before 15. The image decoding method according to claim 14, wherein the inverse transform data is inversely quantized based on the quantization width of the image data in. 17. From the transmission data, the encoded image data,
And representative value data set for each predetermined unit area,
And a quantization width are extracted, the coded image data is dequantized for each predetermined block, first dequantized data is generated, and a predetermined inverse transform process is performed on the first dequantized data. Giving,
Inverse-transformed data is generated, the inverse-transformed data is inverse-quantized based on the quantization width for each predetermined unit area, second inverse-quantized data is generated, and the second inverse-quantized data is generated. An image decoding method comprising adding the representative value data to generate decoded image data. 18. The unit area and the size of the block have an integral multiple relationship.
7. The image decoding method according to 7. 19. An image decoding method for decoding an image based on a predicted image, wherein encoded image data, representative value data set for each predetermined block, and quantization width are extracted from transmission data. Decoding the encoded image data based on the quantization width to generate dequantized data, adding the dequantized data and the representative value data, generating decoding difference data, the decoding An image decoding method comprising adding differential data and the predicted image to generate decoded image data. 20. A difference between the decoded image data and the representative value data is calculated, the difference is quantized based on the quantization width to generate decoded quantized data, the decoded quantized data, the representative value. The data and the quantization width are stored in a storage unit, the decoded quantized data, the representative value data, and the quantization width are read from the storage unit, and the predicted image is generated. Image decoding method. 21. From transmission data, encoded image data,
And representative value data set for each predetermined block,
And extraction means for extracting a quantization width, dequantizing the encoded image data to generate first dequantized data, and then subjecting the first dequantized data to a predetermined inverse transform process, First inverse quantization means for generating inverse transformed data; second inverse quantization means for inversely quantizing the inverse transformed data based on the quantization width to generate second inverse quantized data; An image decoding apparatus comprising: an addition unit that adds the second dequantized data and the representative value data to generate decoded image data. 22. An image decoding apparatus for decoding an image based on a predicted image extracts encoded image data, representative value data set for each predetermined block, and quantization width from transmission data. Extraction means, dequantization means for dequantizing the coded image data based on the quantization width to generate dequantized data, and adding the dequantized data and the representative value data for decoding An image decoding method comprising: a first addition unit that generates difference data; and a second addition unit that adds the decoded difference data and the predicted image to generate decoded image data. 23. Extraction means for extracting encoded image data and precision information about bit precision of the image data from transmission data, and whether the encoded image data can be decoded based on the precision information. An image decoding device comprising: a determination unit that determines whether or not the encoded image data is undecodable; and a notification unit that notifies that the encoded image data is undecodable. Device. 24. From transmission data, encoded image data, accuracy information regarding bit accuracy of the image data,
And representative value data set for each predetermined block,
And extraction means for extracting a quantization width, and inversely quantizing the coded image data to generate first inverse quantized data, and then subjecting the first inverse quantized data to predetermined inverse transform processing. The first to apply and generate inverse transformed data
Dequantizing means for dequantizing the inverse transformed data based on the quantization width to generate second dequantized data, and the second dequantized data. And an addition unit that adds the representative value data to generate first decoded image data, and a quantum that quantizes the first decoded image data based on the accuracy information to generate second decoded image data. An image decoding apparatus comprising: an image decoding unit. 25. Representative value data and quantization width of image data are determined for each predetermined block, a difference between the image data and the representative value data is calculated, and the difference is quantized based on the quantization width. To generate first quantized data, subject the first quantized data to a predetermined conversion process, generate transform coefficients, quantize the transform coefficients, and generate second quantized data. An image recording medium, characterized in that a bit stream generated by encoding the representative value data, the quantization width, and the second quantization data is recorded.