JP3085289B2 - Moving picture coding method and moving picture coding apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding method for compressing digital moving picture data, and more particularly to a moving picture coding method and a moving picture coding apparatus using a motion compensation technique.

[0002]

2. Description of the Related Art In order to transmit a moving image over a data communication line, or to record and reproduce a moving image on a data recording medium, it is necessary to compress digital moving image data having an enormous amount of information. . Since a predetermined number of screens are sequentially sent per second for a moving image, a method of compressing data using data correlation within a screen as in the case of a still image and a method of compressing data between screens in the time axis direction are considered. The moving picture coding method is used in combination with the method of compressing data using the data correlation.

[0003] As in the case of a still image, methods for utilizing data correlation within a screen include a predictive coding method and a transform coding method. The predictive coding method is a method of predicting a data value of a pixel to be coded from data values of coded neighboring pixels and quantizing a prediction error which is a difference value from an actual data value. This utilizes data correlation with encoded neighboring pixels. The transform coding method is a method in which a screen is divided into blocks of a predetermined size, and image data is orthogonally transformed into frequency components in block units, and then the frequency components are quantized.
This is a method using data correlation of pixels in a block. The orthogonal transform includes a Hadamard transform and a discrete cosine transform.

As a method of compressing data using data correlation between screens in the time axis direction, there are an inter-picture predictive coding method, a change area coding method, and the like. The inter-screen prediction coding method is a method of predicting a data value of a pixel to be coded from a data value of a coded previous screen pixel and coding a prediction error thereof. This encodes difference data from the previous screen. Further, the change area encoding method is a method of encoding only image data of an area that has changed compared to the previous screen. This encodes only the change area from the previous screen.

[0005] As one of the inter-picture prediction coding methods, a motion compensation prediction coding method is widely used. This motion-compensated predictive coding method divides a screen into blocks of a predetermined size, and, for each block, a block having the closest data value pattern in the previous screen, that is, a block having the smallest difference value between blocks. In this method, a prediction error with the data value of the block is searched for, that is, a motion compensation prediction error is encoded. Of course, it is also necessary to encode a motion vector, which is a block shift amount. For example,
When detecting a motion vector by full search, if the detection range is ± 7 pixels for each of the horizontal and vertical directions, it is necessary to calculate the difference value between blocks a total of (2 × 7 + 1) 2 = 225 times. However, the amplitude of the difference data from the previous screen can be greatly reduced, and the data compression ratio can be improved. This motion-compensated predictive coding method can be efficiently applied to a case where the spatial positions of pixels are aligned between adjacent screens, that is, to a non-interlaced moving image. However, when the spatial positions of the pixels are shifted between adjacent screens, that is, for a moving image of interlaced scanning, some contrivance is required for efficiently performing motion compensation prediction.

A normal television signal is a moving image of interlaced scanning. A screen called a field is transmitted 60 frames per second, and the vertical pixel position is shifted by 0.5 lines between adjacent fields. Therefore, a screen called a frame composed of two fields is
It is also considered that 30 frames are transmitted per second. Therefore, when the motion compensation prediction from the previous field (inter-field motion compensation prediction) is used, the inter-field 0.
There is a problem that the motion compensation prediction error does not become zero even in a still region in a moving image due to the influence of the displacement of the pixel positions of the five lines. In addition, when motion compensation prediction (inter-frame motion compensation prediction) from a field two fields before is used, since the time interval is twice as long as that between fields, the speed of the object in the moving region in the moving image is high. There is a problem that sufficient motion compensation for the motion cannot be performed, and the motion compensation prediction error increases.

As a conventional motion-compensated predictive coding method for interlaced images, which solves these problems, for example, a motion-compensated inter-frame inter-field coding apparatus described in Japanese Patent Application Laid-Open No. 59-128881 is used. Are known. It detects both motion vector between the motion vector and the frame between the fields, using motion vectors of better suitable either in blocks, a method of performing motion compensation predictive coding.

[0008]

In THE INVENTION to be solved INVENTION The above prior art, to detect both motion vectors between fields and between frames, but by selecting one of the motion vectors is carried out motion compensation, motion compensation predictive error In realizing the reduction of
It was not enough. Therefore, the purpose of the present invention is to
To reduce motion compensation prediction error in images
is there.

[0009]

[Means for Solving the Problems] To achieve the above purpose
In addition, the present invention provides a method of
Field-by-block using motion compensation
Of the current field and the predicted signal of the current field.
In video coding to encode prediction error with image signal
The local recovery of the field one field before
Processing to generate a prediction signal from an encoded image signal, and encoding
Locally decoded image signal of the field two fields before
To generate a prediction signal from the
Locally decoded image signal of field before field and encoded
Of the locally decoded image signal of the field two fields before
A process for generating a prediction signal in which both signals are mixed using both.
The prediction signal is generated by selectively using either
Configuration.

[0010] Incidentally, at the same time as the examples of specific means, previous field and upon detecting a motion vector between the current field, depending on the motion amount, one field before, also
Detects a motion vector even between an interpolated field generated by a motion adaptive line interpolation process from a pixel value two fields before and a current field, and selects a motion compensation prediction error which is smaller in both motion vectors. Thus, motion compensation prediction coding is performed .

In the motion adaptive line interpolation process, first, the amount of motion at the position of the pixel to be interpolated in the field is detected. For example, the difference between the actually existing pixel on the line 0.5 line below and the field two fields before, that is, the frame difference is determined as the motion amount. The motion amount takes a value of 0 when there is almost no motion, 1 when the motion is larger than a certain level, and takes a value between 0 and 1 when the motion is intermediate. Next, a weighted average value of the first interpolation value for the still area and the second interpolation value for the moving area is output as a motion adaptive interpolation value of the pixel to be interpolated according to the amount of motion. The first interpolation value is for a still area, and is the pixel value of the immediately preceding field at a position corresponding to the pixel to be interpolated. The second interpolation value is for a moving area, and is a weighted average value of a plurality of pixel values in the same field existing in the vicinity of the pixel to be interpolated.

For example, two lines 0.5 lines above and below
It is the average of two pixel values.

In the motion vector detecting process, a first motion vector detecting process between a previous field and a current field, and a second motion vector detecting process between an interpolated field generated by motion adaptive line interpolation and a current field are performed. Processing is performed. By the first motion vector detection processing, the first inter-field motion compensation is realized as in the conventional example. However, because of the interlaced scanning, the vertical component of the motion vector has a value of ± 0.5, ± 1.5,....

In the stationary area, the motion amount is 0, and the pixel value of the interpolation field is actually the pixel value of two fields before. Motion compensation is realized. Also,
In the moving area, the motion amount is 1, and the pixel value of the interpolation field is actually the weighted average value of the neighboring pixel values in the previous field. A second inter-field motion compensation is realized. However, the vertical component of the motion vector takes a value such as 0, ± 1, ± 2,. When the motion amount takes a value between 0 and 1, a special motion compensation process is performed in which the above two motion compensations are mixed.

Thus, the moving area, the stationary area, and
Inter-field motion compensation and inter-frame motion
Compensation, and a prediction signal in which two fields are mixed
Can be adaptively switched. I
Therefore, to reduce the motion compensated prediction error, thereby improving the performance of motion compensated prediction.

[0016]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram of a first embodiment of a moving picture coding apparatus for realizing data compression in a moving picture coding system to which a motion compensated prediction coding method according to the present invention is applied. FIG. 2 shows a configuration of a video coding system together with the video coding device shown in FIG.
FIG. 3 is a block diagram of a video decoding device that realizes data decompression.

In the moving picture coding apparatus shown in FIG. 1, 1 is an input terminal for image data, 2 is a scan conversion circuit, 3 is a data delay circuit, 4 is a data subtraction circuit, 5 is an orthogonal conversion circuit,
6 is a quantization circuit, 7 is a variable length coding circuit, 8 is a data multiplexing circuit, 9 is a buffer memory, 10 is an output terminal for coded data, 11 is an inverse quantization circuit, 12 is an inverse orthogonal transform circuit,
13 is a data addition circuit, 14a and 14b are field memories, 15 is a motion compensation circuit, 16 is a motion vector detection circuit, 17 is a motion vector encoding circuit, and 32 is a motion adaptive line interpolation circuit.

In the moving picture decoding apparatus shown in FIG. 2, reference numeral 18 denotes an input terminal for encoded data, 19 denotes a buffer memory, and 20 denotes a buffer memory.
Is a data separation circuit, 21 is a variable length decoding circuit, 22 is an inverse quantization circuit, 23 is an inverse orthogonal transformation circuit, 25 is a data addition circuit, 26 is a scan inverse transformation circuit, 27 is an image data output terminal, and 28a 28b is a field memory, 29 is a motion compensation circuit, 30 is a motion vector decoding circuit, and 33 is a motion adaptive line interpolation circuit.

First, the operation of the moving picture coding apparatus shown in FIG. 1 will be described. Original image data obtained by digitizing a video signal is input from an input terminal 1, and the scan conversion circuit 2 converts the order of the original image data from raster scan to block scan. This is shown in FIG. As shown in (a), a raster scan is performed in order from the left to the right pixel and then from the top to the bottom pixel according to the display scan.
Original image data is input from an input terminal 1. Then, since the subsequent encoding processing unit is a block of 8 × 8 pixels, the scan conversion circuit 2 rearranges the order of the original image data from the raster scan (a) to the block scan (b). Can be As shown in (b), the block scan means that the inside of a block is scanned in order from top to bottom pixels, and then from left to right, in order, and between blocks, from left to right, and from top to bottom. Are sequentially scanned.

The original image data, the order of which has been converted into a block scan, is delayed by a predetermined time in a data delay circuit 3 and then the motion compensated prediction image data is subtracted in a data subtraction circuit 4 to generate motion compensated prediction error data. Is done. The delay time in the data delay circuit 3 is for compensating a processing time of a motion vector detection circuit 16 described later.
The motion-compensated predicted image data is image data predicted by motion compensation from image data of a past field that has already been subjected to encoding processing, and is output from the motion compensation circuit 15.

For the motion compensation prediction error data, 8 × 8
The orthogonal transform circuit 5 performs a two-dimensional discrete cosine transform process using pixels as blocks, and transform coefficients after orthogonal transform corresponding to frequency components are quantized by a quantization circuit 6 at a predetermined quantization step width. . Then, the variable-length coding circuit 7 assigns codes to the quantized transform coefficients such that a short code is assigned to data having a high frequency of occurrence and a long code is assigned to data having a low frequency of occurrence. .

The coded image data is multiplexed by the data multiplexing circuit 8 based on the motion vector coded data of the additional information and the information of the quantization step width. Output as encoded data. Here, the coded motion vector data is information on a motion vector used for motion compensation, and is output from the motion vector coding circuit 17. Further, the buffer memory 9 determines the quantization step width according to the amount of data stored in the buffer memory 9, that is, the buffer memory occupancy, and outputs the information. If the buffer memory occupancy is high, the quantization step width is controlled to be large, and if the buffer memory occupancy is low, the quantization step width is controlled to be small.

The quantized transform coefficient output from the quantization circuit 6 is returned to a transform coefficient by the inverse quantization circuit 11, and then transformed into prediction error data by the inverse orthogonal cosine transform by the inverse orthogonal transform circuit 12. Will be returned. In the data addition circuit 13, the motion compensated prediction image data used for the subtraction in the data subtraction circuit 4 is added to the locally decoded prediction error data to generate locally decoded image data. The locally decoded image data is image data reproduced by data expansion in a moving image decoding device that is paired with the moving image encoding device, and is used to generate subsequent motion compensated prediction image data.

For the locally decoded image data, image data between lines is interpolated and generated by the motion adaptive line interpolation circuit 32, and the actual image data actually existing in the first field memory 14a is replaced with the second image data. Interpolated image data generated by interpolation is sequentially written to the field memory 14b. The locally decoded real image data and the interpolated image data of the current field are used to generate motion-compensated predicted image data necessary for encoding of the original image data of the next field.

According to the motion vector detected by the motion vector detecting circuit 16, the motion compensating circuit 15 stores the actual image data of the previous field after local decoding stored in the field memory 14a or the real image data stored in the field memory 14b. The interpolated image data of the previous field after the local decoding is shifted in units of blocks and output as motion compensated prediction image data. The motion vector detection circuit 16 compares the original image data of the current block of the current field with the locally decoded image data of the previous field held in the field memory 14a or 14b while shifting the image data on a pixel-by-pixel basis. The shift amount of the block in which the block difference value becomes small is detected and output as a motion vector. In this motion vector, the motion vector coding circuit 17 performs variable length coding on each component of the previous block.

Next, the operation of the moving picture decoding apparatus shown in FIG. 2 will be described. The encoded data input from the input terminal 18 is
After being temporarily stored in the buffer memory 19, the encoded image data, the encoded motion vector data as additional information, and the information of the quantization step width are separated by the data separation circuit 20. Image coded data, which is coded data of a transform coefficient after quantization, is first decoded by a variable length decoding circuit 21, returned to a transform coefficient by an inverse quantization circuit 22, and then decoded by an inverse orthogonal transform circuit 23. The data is returned to the prediction error data by the inverse discrete cosine transform.

Then, the motion-compensated predicted image data is added to the decoded prediction error data by the data addition circuit 25 to generate reproduced image data. The image data reproduced by decoding is converted in order from block scan to raster scan by the scan reverse conversion circuit 26,
It is output from the output terminal 27 as reproduced image data.

For the reproduced image data, the inter-line image data is interpolated and generated by the motion adaptive line interpolation circuit 33, and the actual image data actually existing in the first field memory 28a is replaced with the second image data. Interpolated image data generated by interpolation is sequentially written in the field memory 28b. The actual image data and the interpolated image data reproduced by decoding the current field are used to generate motion compensation predicted image data required when reproducing the reproduced image data of the next field by decoding.

Next, the configuration and operation of the motion adaptive line interpolation circuit 32 in the moving picture coding apparatus of FIG. 1 will be described in detail. FIG. 3 is a detailed block diagram of the motion adaptive line interpolation circuit 32, and FIG.

In the motion adaptive line interpolation circuit 32 shown in FIG. 3, 38 is a scan inverse conversion circuit, 39 is a 263 line delay circuit, 40 is a 262 line delay circuit, 41 is an edge detection circuit, 42 is a motion detection circuit, and 43 is a motion detection circuit. A mixing ratio generation circuit, 45 is a line delay circuit, 46 is a data addition circuit,
7 is a 1/2 multiplication circuit, 49 and 50 are data multiplication circuits, 5
1 is a data addition circuit, and 44, 48 and 52 are data delay circuits.

First, in the scan reverse conversion circuit 38, the input image data of the current field is rearranged in the same raster scan order as block scan to display scan.

The image data of the current field is 26
The image data output from the 3-line delay circuit 39 and output from the 263-line delay circuit 39 is further applied to a 262-line delay circuit 40. Since one field of the interlaced scan image data is composed of 262.5 lines, the 263 line delay circuit 39 outputs the image data at a position one field before the current field and 0.5 line above the current line. Is output. The 262 line delay circuit 40 outputs the image data at the same position as the current line just two fields before the current field.

The motion detecting circuit 42 calculates a difference between the image data of the current line of the current field and the image data at the same position as the current line just two fields before the current field, that is, a frame difference, and calculates the difference value. A motion amount having a value between 0 and 1 is generated. Predetermined first
When the frame difference is larger than the threshold value, the motion amount is 1 indicating the moving image mode. When the frame difference is smaller than the second predetermined threshold value, the motion amount is the still image mode. 0. Naturally, the first threshold value is larger than the second threshold value, and when the frame difference is a value between both threshold values, the motion amount is 0 or 1.
Take a value between and.

The edge detecting circuit 41 applies a high-pass filter in the space to the image data of the current field, and generates an edge amount having a value between 0 and 1 from the output value. The edge amount is a value corresponding to the strength of the edge in the screen, and a value of 0 is given to a flat portion where there is no edge, and a value of 1 is given to a steep portion where there is an edge. In the intermediate case, the value takes a value between 0 and 1. This edge amount is input to the motion detection circuit 42,
Two thresholds used in the motion detection circuit are variably controlled according to the edge amount. That is, when the edge amount is small, the two threshold values are small, and when the edge amount is large, the two threshold values are large.

This corresponds to weakening the sensitivity of motion detection in a portion having a strong edge.

In the mixing ratio generation circuit 43, the motion detection circuit 4
From the motion amount generated in step 2, a mixture ratio km for weighting and mixing two types of line interpolation values is generated. Mixing ratio k
m is a discrete value between 0 and 1, 0, 1/8, 2
/ 8,..., 1 When the movement amount is small, the mixture ratio takes a small value, and when the movement amount is large, the mixture ratio takes a large value. When the mixture ratio km is 0, the line interpolation value in the still image mode is used, and when the mixture ratio km is 1, the line interpolation value in the moving image mode is used.

The image data of the interpolation line located on the 0.5 line of the current line in the current field is generated by mixing the following two types of line interpolation values. The line interpolation value in the still image mode is the value one field before, and the output of the 263 line delay circuit 39, that is, the value of the image data at the same position as the interpolation line one field before the interpolation line is used. The line interpolation value in the moving image mode is an average value of the upper and lower lines, and is multiplied by 後 に after the image data of the current line and the image data of the previous line which is the output of the line delay circuit 45 are added by the data addition circuit. The circuit 47 converts the value into a value of 1/2 and uses the generated average value. Here, the data delay circuit 44
And 48 are for compensating the processing time in the edge detection circuit 41, the motion detection circuit 42, and the mixture ratio generation circuit 43.

The mixing ratio k which is the output of the mixing ratio generating circuit 43
In accordance with the value of m, first, the data multiplication circuit 49 adds (1) to the previous field value which is the line interpolation value in the still image mode.
−km), and the data multiplication circuit 50 multiplies the upper and lower line averages, which are line interpolation values in the moving image mode, by a value of km. And 49 and 50
The outputs of the two data multiplying circuits are added by the data adding circuit 51 and output as interpolated image data. On the other hand, the image data of the current line of the current field is output as real image data together with the interpolated image data after the data delay circuit 52 compensates for the processing time for generating the interpolated image data.

FIG. 4 shows the processing contents of the motion adaptive line interpolation circuit 32 described above, that is, the concept of a method of generating image data of an interpolation line. A mark indicates each line of the moving image to be interlaced, and a mark indicates an interpolation line generated by the motion adaptive line interpolation process. Motion detection is performed from both the current line image data and the 525 line previous image data to generate a mixture ratio km, and the 263 line previous image data, the current line image data, and the 1 line previous image data Are weighted and mixed according to the mixing ratio km.

The configuration and operation of the motion adaptive line interpolation circuit 33 in the video decoding device are exactly the same as those of the motion adaptive line interpolation circuit 32 in the video coding device described above.

Next, the configuration and operation of the motion compensation circuit 15 in the moving picture coding apparatus of FIG. 1 will be described in detail.
FIG. 8 is a detailed block diagram of the motion compensation circuit 15, and FIG.

In the motion compensation circuit 15 shown in FIG.
Reference numerals 3 and 55 denote extended block data generation circuits, 54 and 56 denote shift block output circuits, 57 denotes a motion vector vertical component determination circuit, and 58 denotes a data switching circuit.

First, the real image data of the previous field held in the field memory 14a of FIG. 1 is input to the extended block data generating circuit 53 of FIG. 3, and the image data of the extended block corresponding to the current block is output. You. Similarly, the interpolation image data by the motion adaptive line interpolation of the previous field held in the field memory 14b is
The image data is input to the extended block data generation circuit 55, and the image data of the extended block corresponding to the current block is output.
As shown in FIG. 7, the extended block is obtained by integrating nine neighboring blocks including the current block of 8 × 8 pixels.
It has a size of 24 × 24 pixels around the current block.

FIG. 9 shows an extended block data generation circuit 53.
FIG. 4 shows a more detailed block diagram of FIG.

In the extended block data generation circuit 53, reference numerals 59 and 60 denote block line delay circuits, 61 to 6
6 is a block delay circuit. The block line delay circuits 59 and 60 delay the input block data by one block line, that is, one horizontal line of a block of 8 × 8 pixels, that is, by eight lines. The input block data is delayed by one block of the 8 × 8 pixel block.

The block line delay circuit 59 and the block line delay circuit 60 are cascade-connected, and the block delay circuits 61 and 62, 63 and 64, and 65 and 66 are cascade-connected. First, the block data input to the extended block data generation circuit 53 of FIG.
Given to one. The block data output from the block line delay circuit 59 is passed to the block delay circuit 63, and the block data output from the block line delay circuit 60 is passed to the block delay circuit 65.

The inputs and outputs 9 of the block delay circuits 61 to 66 indicated by the symbols a to i in FIG.
The types of block data are put together and output as extended block data. Each block data corresponds to the block data of nine blocks indicated by the symbols a to i in FIG. The current block, which is the center of the extended block, is a block a, which is delayed by one block line and one block from the block i input to the motion compensation circuit 15. The configuration and operation of the extended block data generation circuit 55 in FIG. 8 are completely the same.

[0049] Now, the shift block output circuits 54 of FIG. 8
Then, the image data of the shift block, which is a part of the extended block, is extracted from the image data of the extended block centered on the current block in the actual image data of the previous field given from the extended block data generating circuit 53. Similarly, in the shift block output circuit 56, given from the extension block data generation circuits 5 5 from the extended block image data centered on the current block in the interpolation image data of the previous field, is the shift block at a portion thereof Image data is extracted.

As shown in FIG. 7, the shift block is a block of 8 × 8 pixels which forms a part of the extended block.
This is extracted from a position shifted from the center current block position by a predetermined shift vector. As described later, the range of the value of the shift vector is ± 7 pixels in both the horizontal direction and the vertical direction, and the shift block always exists inside the extended block.

The range of the motion vector value is ± 7 pixels in one pixel unit in the horizontal direction, and ± 7 pixels in 0.5 pixel units in the vertical direction. Then, based on the motion vector input to the motion compensation circuit 15 of FIG. 1, the motion vector vertical component determination circuit 57 of FIG.
, And a shift block selection signal applied to the data switching circuit 58 are generated. Since the shift vector is generated by truncating the decimal part of the vertical component of the motion vector, the range of the value is ± 7 pixels in one pixel unit in both the horizontal and vertical directions. Further, the shift block selection signal is generated such that when the decimal part of the vertical component of the motion vector is 0, it is “0”, and when the decimal part is 0.5, it is “1”.

Either the shift block in the real image data of the previous field which is the output of the shift block output circuit 54 or the shift block in the interpolation image data of the previous field which is the output of the shift block output circuit 56 is a data switching circuit 58. And output as motion compensated prediction image data. The data switching circuit 58 selects the shift block of the interpolation image data when the shift block selection signal supplied from the vertical component determination circuit 57 is “0”, and selects the shift block of the real image data when the shift block selection signal is “1”.

The motion vector detecting circuit 16 shown in FIG.
In this case, the processing takes one block line and one block of processing time, and is equal to the delay time of the block a treated as the current block with respect to the block i input to the motion compensation circuit 15. Therefore, in the shift block output circuits 54 and 56, a shift vector with the correct timing is given to the extended block centering on the current block a, and a correct shift block is extracted.

FIG. 5 shows the processing contents of the motion compensation circuit 15 described above and the concept of the motion compensation method.
A bold black frame indicates the current block of the current field, and a white square indicates a pixel in the block. Further, a thin black frame is a motion compensation prediction block of the previous field generated by the motion adaptive line interpolation circuit 32, and a black mark indicates a pixel of actual image data in the block, and a cross indicates a pixel of interpolated image data in the block. ing. However, here, for simplification of the drawing, the block size is shown as eight horizontal pixels × four vertical pixels.

When the vertical component of the motion vector is an integer, that is, the fractional part is 0, the pixel of the interpolated image data indicated by the cross mark of the motion compensation prediction block is replaced by the triangle mark of the current block as shown in FIG. It has a shape corresponding to the pixel indicated by. On the other hand, when the vertical component of the motion vector is the sum of an integer and 0.5, that is, the fractional part is 0.5, the real image indicated by the black circle of the motion compensation prediction block as shown in FIG. The pixel of the data has a shape corresponding to the pixel indicated by the triangle in the current block. Then, in the data subtraction circuit 4, the motion adaptive line interpolation circuit 3 indicated by a thin black frame is converted from the image data of the current block of the current field indicated by a thick black frame.
2, the image data of the motion compensation prediction block of the previous field generated by step 2 is subtracted, and the motion compensation processing is performed.

The configuration and operation of the motion compensation circuit 29 in the video decoding device are exactly the same as those in the motion compensation circuit 15 in the video coding device described above.

Finally, the motion vector detection circuit 16 shown in FIG.
The configuration and operation of will be described in detail. FIG. 10 is a detailed block diagram of the motion vector detection circuit 16.

In the motion vector detecting circuit 16 shown in FIG. 10, 67 is a data delay circuit, 68 and 69 are extended block data generating circuits, 70a to 70z, and 73a to 73a.
73z is a shift block output circuit, 71a to 71z and 74a to 74z are block difference value generation circuits, and 75 is a minimum value detection / motion vector generation circuit.

First, in the extended block data generation circuit 68, the image data of the extended block is generated from the actual image data of the previous field held in the field memory 14a of FIG. Similarly, in the extended block data generating circuit 69, the image data of the extended block is generated from the interpolated image data by the motion adaptive line interpolation of the previous field held in the field memory 14b. The configuration and operation of the extended block data generation circuits 68 and 69 are the same as those of the extended block data generation circuit 53 in the motion compensation circuit 15 already described with reference to FIGS.

Shift block output circuits 70a to 70z
Are the number of all shift vectors that can exist within the motion vector detection range, and each shift block output circuit 70
Different shift vectors are given to all of a to 70z. The motion vector detection range is ± 7 pixels in both the horizontal direction and the vertical direction, and the shift vector can exist in a unit of one pixel in both the horizontal direction and the vertical direction. Therefore, the total number of the shift block output circuits 70a to 70z is (( 2 × 7 + 1) 2 =
There are 225.

Thus, the shift block output circuit 70
From a to 70z, image data of all types of shift blocks that can be extracted from the extended block is output. The shift block data output circuit 70
The configuration and operation of a to z are the same as those of the motion compensation circuit 15 described above.
Is the same as that of the shift block output circuit 54 in FIG.

The block difference value generating circuits 71a to 71z are also 225 like the shift block output circuits 70a to 70z.
There are pieces. The block difference value generation circuit 71a outputs the image data of the first shift block in the previous field, which is the output of the shift block output circuit 70a, and the current image of the current block in the current field, which is delayed by the data delay circuit 67 by a predetermined time. The data is input. The data delay circuit 67 includes extended block data generation circuits 68 and 6
9 is to compensate for the processing time, and to realize a data delay of only one block line and one block. Then, after the difference of the image data between the pixels at the corresponding positions is obtained for the two blocks, 8 × 8 = 6
The four difference values are summed to generate a block difference value. The same applies to the operation of the block difference value generation circuits 71b to 71z.

Next, shift block output circuits 73a to 73a
There are also 225 3z, and each shift block output circuit 73a
７３73z are all given different shift vectors. Therefore, the shift block output circuits 73a to 73z
From this, image data of all types of shift blocks that can be extracted from the extended block of the interpolated image data in the previous field is output. The configuration and operation of the shift block output circuits 73a to 73z are the same as those of the shift block output circuit 54 in the motion compensation circuit 15 described above.

There are also 225 block difference value generation circuits 74a to 74z, and each of the shift block output circuits 73a to 73z.
The image data of each shift block in the previous field, which is the output of z, and the current image data of the current block in the current field delayed by the data delay circuit 67 by a predetermined time are input. Then, each block difference value generation circuit 74a
In the case of .about.74z, for the two blocks, after the difference of the image data is obtained between the pixels at the corresponding positions, the 64 difference values are summed to generate a block difference value.

That is, the extended block data generation circuit 6
8. The shift block output circuits 70a to 70z and the block difference value generation circuits 71a to 71z generate block difference values between all the shift blocks that may exist in the actual image data of the previous field and the current block in the current field. .

The extended block data generation circuit 69
The shift block output circuits 73a to 73z and the block difference value generation circuits 74a to 74z generate block difference values between all possible shift blocks in the interpolation image data of the previous field and the current block in the current field.

The minimum value detection / motion vector generation circuit 75 compares all the generated block difference values to detect the minimum value, and generates a motion vector from the number of the block difference value having the minimum value. Is output. That is, when the block difference value having the minimum value is one of the outputs of the block difference value generation circuits 74a to 74z,
The shift vectors set in the corresponding shift block output circuits 73a to 73z are used as they are as the motion vectors. If the block difference value having the minimum value is an output of any of the block difference value generation circuits 71a to 71z, the corresponding shift block output circuit 70
The value of the vertical component of the shift vector set to a to 70z is reduced by 0.5 and used as a motion vector.

FIG. 11 is a block diagram of a second embodiment of a moving picture coding apparatus for realizing data compression in a moving picture coding system to which the motion compensated moving picture coding method according to the present invention is applied. FIG. 12 is a block diagram of a moving picture decoding apparatus which constitutes a moving picture coding system together with the moving picture coding apparatus shown in FIG. 11, and realizes data decompression.

In the moving picture coding apparatus shown in FIG. 11, 1 is an image data input terminal, 2 is a scan conversion circuit, 3 is a data delay circuit, 4 is a data subtraction circuit, 5 is an orthogonal transformation circuit, and 6 is a quantization circuit. , 7 is a variable length coding circuit, 8 is a data multiplexing circuit, 9 is a buffer memory, 10 is an output terminal for coded data, 11 is an inverse quantization circuit, 12 is an inverse orthogonal transform circuit, 12 is a data addition circuit, 14a And 14b are field memories, 15 is a motion compensation circuit, 16 is a motion vector detection circuit, 17 is a motion vector encoding circuit, 32 and 34 are motion adaptive line interpolation circuits, and 35a and 35b are field memories.

In the moving picture decoding apparatus shown in FIG.
Is an input terminal for encoded data, 19 is a buffer memory, 2
0 is a data separation circuit, 21 is a variable length decoding circuit, 22 is an inverse quantization circuit, 23 is an inverse orthogonal transform circuit, 25 is a data addition circuit, 27 is an image data output terminal, and 28a and 28b.
Is a field memory, 29 is a motion compensation circuit, 30 is a motion vector decoding circuit, 33 is a motion adaptive line interpolation circuit,
37 is a double speed conversion circuit.

First, the operation of the moving picture coding apparatus shown in FIG. 11 will be described. The original image data input from the input terminal 1 is
After the order is changed from raster scan to block scan by the scan conversion circuit 2 and delayed by a predetermined time by the data delay circuit 3, the motion compensation prediction image data is subtracted by the data subtraction circuit 4 to generate motion compensation prediction error data. Is done. The motion compensation predicted image data is output from the motion compensation circuit 15.

Then, the orthogonal transformation circuit 5, the quantization circuit 6,
The variable-length coding circuit 7 performs coding on the motion-compensated prediction error data by discrete cosine transform to generate coded image data. The coded image data is multiplexed with the motion vector coded data of the additional information and the information of the quantization step width by the data multiplexing circuit 8, temporarily stored in the buffer memory 9, and then output from the output terminal 10 as coded data. Is done. Here, the motion vector encoded data is stored in the motion vector encoding circuit 1
7 is output.

The quantized transform coefficient output from the quantization circuit 6 is locally decoded by the inverse quantization circuit 11 and the inverse orthogonal transform circuit 12, and is returned to motion compensation prediction error data. Then, the data addition circuit 13 adds the motion-compensated predicted image data to generate locally decoded image data. For this locally decoded image data, image data between lines is interpolated and generated by the motion adaptive line interpolation circuit 32, and actual image data actually existing in the first field memory is
Interpolated image data generated by interpolation is sequentially written to the second field memory.

The motion compensating circuit 15 converts the real image data or the interpolated image data of the previous field, which has been stored in the field memories 14a and 14b, after the local decoding or the interpolated image data into blocks according to the motion vector detected by the motion vector detecting circuit 16. It is shifted as a unit and output as motion compensated prediction image data. The above encoding processing is exactly the same as that of the moving picture encoding apparatus of FIG. 1 described above.

However, the processing of motion vector detection is the same as that of FIG.
Is different from the case of the moving picture coding apparatus of FIG. As for the image data of the current field, after the inter-line image data is interpolated and generated by the motion adaptive line interpolation circuit 34, the actual image data is written into the field memory 35a and the interpolated image data is sequentially written into the field memory 35b. The motion vector detection circuit 16 compares the original image data of the current block of the current field with the original image data of the previous field held in the field memory 35a or 35b while shifting it by a pixel unit. The shift amount of the smaller block is detected and output as a motion vector.

In the moving picture coding apparatus of FIG. 1 described above, the motion vector of the current block in the current field is detected from the past field of the reproduced picture which has been locally decoded after coding. In the moving picture coding apparatus, a motion vector of a current block in a current field is detected from a past field of an original picture. As described above, since the motion vector is detected using the original image in which the encoding distortion is not mixed, more accurate detection of the motion vector becomes possible, and the encoding distortion is not conspicuous in the video decoding device. A reproduced image is obtained, and the image quality can be improved.

The motion adaptive line interpolation circuits 32 and 34
The detailed configuration and operation are exactly the same as those of the motion adaptive line interpolation circuit 32 in the moving picture coding apparatus of FIG. 1 described above. The detailed configurations and operations of the motion compensation circuit 15 and the motion vector detection circuit 16 are described in FIG.
Is exactly the same as the case of the motion compensation circuit 15 and the motion vector detection circuit 16 in the moving picture coding apparatus of FIG.

Next, the operation of the moving picture decoding apparatus shown in FIG. 12 will be described. The encoded data input from the input terminal 18 is temporarily stored in the buffer memory 19, and then the image separation data, the motion vector encoded data of the additional information, and the information of the quantization step width are separated by the data separation circuit 20. Is done. The encoded image data is decoded by the variable length decoding circuit 21, the inverse quantization circuit 22, and the inverse orthogonal transform circuit 23, and is returned to the motion compensation prediction error data. And
The data addition circuit 25 adds the motion-compensated predicted image data to the decoded motion-compensated predicted error data to generate reproduced image data. With respect to the reproduced image data, the image data between lines is interpolated by the motion corresponding line interpolation circuit 33 to generate the actual image data and the interpolated image data, and at the same time, the order of the image data is changed from the block scan to the raster scan. Is converted. The double-speed conversion circuit 37 multiplexes the actual image data and the interpolated image data on a line-by-line basis, so that one field has twice the number of lines as usual and double-speed reproduced image data of non-interlace scanning is used. Generated and output.

Further, the actual image data and the interpolated image data output from the motion adaptive line interpolation circuit 33 are sequentially written to the field memories 28a and 28b, respectively.
Then, the motion compensation circuit 29 converts the decoded real image data or interpolation image data of the previous field held in the field memories 28a and 28b into blocks according to the motion vector decoded by the motion vector decoding circuit 30. It is shifted as a unit and output as motion compensated prediction image data.

In the moving picture decoding apparatus shown in FIG. 2, the output of the data adding circuit 25 is output as it is as reproduced picture data after passing through the scan inverse conversion circuit 26. In the multiplexing device, the output of the data addition circuit 25 is doubled by the motion adaptive line interpolation circuit 33, and is output as a double-speed reproduced image of non-interlace scanning by the double-speed conversion circuit 37.
As described above, since the double-speed reproduced image obtained by performing the non-interlaced scanning motion adaptive line interpolation is output, a reproduced image with smooth motion and no flicker can be obtained, and the image quality can be improved.

The detailed configuration and operation of the motion adaptive line interpolation circuit 33 and the motion compensation circuit 29 are the same as those of the motion adaptive line interpolation circuit 32 in the moving picture coding apparatus shown in FIG.
This is exactly the same as the case of the motion compensation circuit 15.

As described above, two embodiments have been described with respect to the moving picture coding apparatus and the moving picture decoding apparatus which constitute the moving picture coding system to which the motion compensated moving picture coding method according to the present invention is applied. Of course, it is possible to combine the first embodiment of the moving picture coding apparatus with the second embodiment of the moving picture decoding apparatus, and conversely, the second embodiment of the moving picture coding apparatus. It is also possible to combine the example with the first embodiment of the video decoding device.

FIG. 13 is a block diagram of a third embodiment of a moving picture coding apparatus for realizing data compression in a moving picture coding system to which the motion compensated moving picture coding method according to the present invention is applied. . FIG. 14 is a block diagram of a moving picture decoding apparatus which constitutes a moving picture coding system together with the moving picture coding apparatus shown in FIG. 13, and realizes data decompression.

In the moving picture coding apparatus shown in FIG. 13, 1 is an image data input terminal, 2 is a scan conversion circuit, 3 is a data delay circuit, 4 is a data subtraction circuit, 5 is an orthogonal transformation circuit, and 6 is a quantization circuit. , 7 is a variable length coding circuit, 8 is a data multiplexing circuit, 9 is a buffer memory, 10 is an output terminal of coded data, 11 is an inverse quantization circuit, 12 is an inverse orthogonal transform circuit, 13 is a data addition circuit, 14a And 14b are field memories, 15 is a motion compensation circuit, 16 is a motion vector detection circuit, 76 is a motion adaptive line interpolation circuit, 77 is a data selection circuit, 78 is a coding mode determination circuit, and 79 is an additional information coding circuit. .

In the moving picture decoding apparatus shown in FIG.
Is an input terminal for encoded data, 19 is a buffer memory, 2
0 is a data separation circuit, 21 is a variable length decoding circuit, 22 is an inverse quantization circuit, 23 is an inverse orthogonal transform circuit, 25 is a data addition circuit, 27 is an image data output terminal, and 28a and 28b.
Is a field memory, 29 is a motion compensation circuit, 80 is a motion adaptive line interpolation circuit, 81 and 86 are data selection circuits, 8
2 is an output image determination circuit, 83 is an input terminal of a reproduction mode,
84 is a decoding circuit for additional information.

First, the operation of the moving picture coding apparatus shown in FIG. 13 will be described. This moving image encoding apparatus is used to record encoded data after data compression on a large-capacity data recording medium such as an optical disk. In the corresponding moving picture decoding apparatus of FIG. 14, in order to enable reproduction from a scene in the middle of a recorded moving picture, a field for intra-field encoding is provided periodically every predetermined number of fields.

The order of the original image data input from the input terminal 1 is changed from raster scan to block scan by the scan conversion circuit 2, delayed by the data delay circuit 3 for a predetermined time, and then predicted by the data subtraction circuit 4. The image data is subtracted to generate prediction error image data. As the prediction image data, the motion compensation prediction image data output from the motion interpolation circuit 15 and the fixed value “0” for intra-field encoding are switched by the data selection circuit 77 and output.

Then, the orthogonal transformation circuit 5, the quantization circuit 6,
The variable length coding circuit 7 performs coding on the prediction error data by discrete cosine transform to generate coded image data. The image coded data is multiplexed with the motion vector coded data of the additional information, the coding mode coded data, and the information of the quantization step width by the data multiplexing circuit 8, and is temporarily stored in the buffer memory 9 before being output. The data is output from the terminal 10 as encoded data. Here, the motion vector encoded data and the encoding mode encoded data are output from the additional information encoding circuit 79.

The transform coefficient after quantization, which is the output of the quantization circuit 6, is locally decoded by the inverse quantization circuit 11 and the inverse orthogonal transform circuit 12, and is returned to the prediction error image data. And
The data addition circuit 13 adds the predicted image data to generate locally decoded image data. For this locally decoded image data, image data between lines is interpolated and generated by a motion adaptive line interpolation circuit 76, and actual image data actually present in the first field memory is stored in the second field memory. Are sequentially written with interpolated image data generated by interpolation.

The motion compensating circuit 15 converts the real image data or the interpolated image data of the previous field, which has been stored in the field memories 14a and 14b, after the local decoding, into the block according to the motion vector detected by the motion vector detecting circuit 16. It is shifted as a unit and output as motion compensated prediction image data. The motion vector detection circuit 16 compares the image data of the current block of the current field with the image data held in the field memory 14a or 14b while shifting the image data on a pixel-by-pixel basis. The amount is detected and output as a motion vector.

However, unlike the case of the moving picture coding apparatus of FIG. 1, a coding mode instruction signal is output from the coding mode determining circuit 78 so that intra-field coding is performed periodically every predetermined number of fields. You. That is, the coding mode instruction signal is "1" indicating the intra-field coding mode during two fields, and "0" indicating the motion compensation inter-field coding mode during the subsequent 28 fields. Thereby, when the encoding mode instruction signal is the intra-field encoding mode of “1”, the data selection circuit 77 selects the fixed value “0”, and the motion compensation of the encoding mode instruction signal of “0”. In the case of the inter-field coding mode, the motion-compensated predicted image data output from the motion compensation circuit 15 is selected and output as predicted image data. The encoding mode includes the additional information encoding circuit 79 together with the motion vector.
And output to the data multiplexing circuit 8.

The encoding mode instruction signal is also supplied to the motion adaptive line interpolation circuit 76. In the motion adaptive line interpolation circuit 76, in the case of the motion compensation inter-field coding mode in which the encoding mode instruction signal is "0", normal motion adaptive line interpolation is performed, but the encoding mode instruction signal is "1". In the case of the intra-field encoding mode, the interpolation using the average value of the upper and lower lines in the complete moving image mode is forcibly fixed. This is so that a field for intra-field coding and a field to be subsequently coded between motion compensation fields can be coded and decoded regardless of past fields.

FIG. 15 is a detailed block diagram of the motion adaptive line interpolation circuit 76. 38 is a scan inverse conversion circuit, 3
9 is a 263 line delay circuit, 40 is a 262 line delay circuit, 41 is an edge detection circuit, 42 is a motion detection circuit, 43
Is a mixing ratio generation circuit, 45 is a line delay circuit, 46 is a data addition circuit, 47 is a 乗 算 multiplication circuit, 49 and 50 are data multiplication circuits, 51 is a data addition circuit, 44, 48 and 52 are data delay circuits, 85 is a data selection circuit.

The configuration and operation of the motion adaptive line interpolation circuit 76 are almost the same as those of the motion adaptive line interpolation circuit 32 shown in FIG. The difference is that there is a data selection circuit 85 that switches the mixing ratio according to the coding mode, and performs motion adaptive line interpolation in the case of the motion compensation inter-field coding mode in which the coding mode instruction signal is “0”. Therefore, the mixture ratio which is the output of the mixture ratio output circuit 43 is selected. However, in the case of the intra-field encoding mode in which the encoding mode instruction signal is “1”, the fixed value “1” is used to perform the interpolation in the complete moving image mode. "Is selected. The operation of the other parts is the same as in the case of the above-described motion adaptive line interpolation circuit 32, and a description thereof will be omitted.

In the moving picture coding apparatus shown in FIG. 1 described above, only the motion compensation inter-field coding mode for performing motion compensation prediction from the previous field is used. In the moving picture coding apparatus shown in FIG. In addition to the compensated inter-field coding mode, there is an intra-field coding mode periodically every 2 fields for every 30 fields. In the case of intra-field encoding, encoding is performed regardless of past fields.
By starting decoding from encoded data of an arbitrary intra-coded field, reproduction of a moving image from that field can be realized.

The detailed configuration and operation of the motion compensating circuit 15 and the motion vector detecting circuit 16 are exactly the same as those of the motion compensating circuit 15 and the motion vector detecting circuit 16 in the moving picture coding apparatus shown in FIG. It is.

Next, the operation of the moving picture decoding apparatus shown in FIG. 14 will be described. This moving picture decoding apparatus converts encoded data recorded on a large-capacity data recording medium such as an optical disc into
Used for reproduction by decoding. Reproduction can be performed from an arbitrary field of intra-field encoding periodically provided for a predetermined number of fields, that is, from a scene in the middle of a moving image. In addition to the normal playback mode, the camera has a slow playback mode for decoding and playing back encoded data at half the normal speed.

The coded data input from the input terminal 18 is temporarily stored in the buffer memory 19, and then, in the data separation circuit 20, the image coded data, the motion vector coded data of the additional information, and the coding mode coding Data and quantization step width information are separated. The encoded image data is decoded by the variable length decoding circuit 21, the inverse quantization circuit 22, and the inverse orthogonal transform circuit 23, and is returned to the prediction error data.

Then, the predicted image data is added to the decoded prediction error data by the data addition circuit 25 to generate reproduced image data. With respect to the reproduced image data, the image data between lines is interpolated by the motion adaptive line interpolation circuit 80 to generate actual image data and interpolated image data. The actual image data and the interpolated image data output from the motion adaptive line interpolation circuit 80 are sequentially written to the field memories 28a and 28b, respectively.

According to the motion vector decoded by the additional information decoding circuit 84, the motion compensation circuit 29 converts the decoded real image data or interpolation image data of the previous field held in the field memories 28a and 28b into The data is output as motion-compensated predicted image data shifted in units of blocks. Then, according to the coding mode decoded by the additional information decoding circuit 84, the fixed value “0” is set in the case of the intra-field coding mode, and the fixed value “0” is set in the case of the motion compensation inter-field coding mode. The motion-compensated predicted image data that is output is selected by the data selection circuit 86 and output as predicted image data.

When performing normal reproduction, the slow reproduction instruction signal input from the input terminal 83 is "0", and the output image determination circuit 82 always outputs "0". Therefore, the actual image data output from the motion adaptive line interpolation circuit 80, that is, the output of the data addition circuit 25 is converted from block scan to raster scan and output as reproduced image data.

On the other hand, when performing slow playback at 1/2 speed, the slow playback instruction signal input from the input terminal 83 is "1", and the encoded data is input at 1/2 speed. The data is input from the terminal 18 and decoded. In this case, since the original image data of one field is decoded in the time of two fields, the actual image data and the interpolated image data for one field, which are the outputs of the motion adaptive line interpolation circuit 80, are output to the data selection circuit. At 81, the data is alternately switched for each field and output as reproduced image data.

That is, in the slow playback mode in which the slow playback instruction signal is “1”, the output of the output image determination circuit 82 is the output of the real image data if the originally coded field is an odd field. Is first set to "0" in order to select the first, and then set to "1" in order to output the interpolated image data later. Conversely, when the originally coded field is an even field, the output of the output image determination circuit 82 becomes "1" first to select an interpolation field first, and the real image data is later output. Next, it becomes "0" for output.

As described above, in the slow reproduction mode of 1/2 speed, instead of reproducing the field twice after decoding the coded data, the number of double lines obtained by the motion adaptive line interpolation is used. Since the real image data and the interpolated image data of the field having are reproduced once each, a slow reproduction image with smooth motion and high resolution can be obtained, and the image quality can be improved.

The detailed configuration and operation of the motion adaptive line interpolation circuit 80 are exactly the same as those of the motion adaptive line interpolation circuit 76 in the moving picture coding apparatus shown in FIG. The detailed configuration and operation of the motion compensation circuit 28 are as follows.
This is exactly the same as the case of the motion compensation circuit 15 in the moving picture coding apparatus of FIG. 1 described above.

As described above, the third embodiment has been described with respect to the moving picture coding apparatus and the moving picture decoding apparatus constituting the moving picture coding system to which the motion compensation moving picture coding method according to the present invention is applied. As a matter of course, the moving picture coding apparatus may use a method of detecting a motion vector from the previous field and the current field of the original picture as in the second embodiment. Also in the moving picture decoding apparatus, as in the second embodiment, a double-speed playback image of non-interlaced scanning may be output with twice the number of lines in one field.

The above-described embodiments of the moving picture coding apparatus and the moving picture decoding apparatus perform motion compensation and motion vector detection using a reference picture having twice the number of lines generated by motion adaptive line interpolation. Is realized by hardware processing, but may be realized by software processing by a digital signal processor (DSP) or the like. At this time, one process can be realized by parallel processing of a plurality of DSPs, and one D
A plurality of processes can be realized by serial processing of SPs.

As described above, motion compensation is performed by software processing,
When realizing motion detection, it is not necessary to keep the amount of processing in each block constant as in the case of hardware processing, and it is necessary to reduce excess processing as much as possible and perform other types of processing with that amount Is desirable. Therefore, it is preferable to reduce unnecessary processing by detecting unnecessary motion vector detection processing without deteriorating the performance of motion compensation. For example, the real image data and the interpolated image data of the previous field are generated by motion adaptive line interpolation, and a motion vector in which the block difference value between the blocks is the smallest between them and the current field of the original image is detected. Although compensation is performed, the motion vector detection range for detecting a motion vector can be variably controlled based on the amount of motion of each pixel detected when performing motion adaptive line interpolation.

That is, the motion vector detection range for detecting the motion vector for each block of the current field is determined from the motion amount of the position of each pixel to be interpolated in the previous field detected at the time of motion adaptive line interpolation. I do. Here, in the block of the previous field in the block and the same spatial position of the current field to be detected motion vector, if the sum of the motion amount of each pixel is larger than a predetermined threshold, the normal first Conversely, when the detection range is used, and when the sum of the motion amounts of the pixels is smaller than a predetermined threshold value, the second detection range smaller than the first detection range is used. The first detection range is a horizontal ± 7 pixel (1 pixel unit) and a vertical ± 7 pixel (0.5 pixel unit), and the second detection range is a horizontal ± 3 pixel (1 pixel unit) and a vertical ± 7 pixel (1 pixel unit). There are three pixels (0.5 pixel unit).

When the sum of the motion amounts of the respective pixels of the block in the previous field is small, the motion vector of the block in the current field is estimated to be small, so that the motion vector detection range is variably controlled as described above. In addition, the useless motion vector detection processing can be reduced and the processing amount can be reduced without lowering the performance of the motion compensation.

The embodiments of the moving picture coding apparatus and the moving picture decoding apparatus described above use the discrete cosine method using the data correlation in the field instead of the motion compensation prediction coding method using the data correlation between the fields. Although a combination of transform coding methods by transform is used, other orthogonal transform such as Adar transform may of course be used. Further, instead of the transform coding method, another method such as a predictive coding method or vector quantization may be used.

As the processing of motion compensation, motion vector detection, and motion adaptive line interpolation, various processing methods are conceivable other than those described in the operation of the circuit of the embodiment, and there is no problem in any method. Of course, other values may be used for the block size, the motion vector detection range, and the like. As in the third embodiment, the coding mode can be changed on a field basis, but the coding mode can be further finely changed on a block basis.

[0113]

According to the present invention, the immediately preceding 2
Since the motion compensation processing is performed by generating the prediction signal of the current field from the video signal of the field, the performance of motion compensation prediction can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram of a first embodiment of a moving picture coding apparatus using a motion compensated prediction coding method according to the present invention.

FIG. 2 is a block diagram of a first embodiment of a video decoding device that is paired with the video encoding device of FIG. 1;

FIG. 3 is a detailed block diagram of a motion adaptive line interpolation circuit in the first embodiment of the video encoding device.

FIG. 4 is a conceptual diagram showing processing contents of a motion adaptive line interpolation circuit.

FIG. 5 is a conceptual diagram showing processing contents of a motion compensation circuit.

FIG. 6 is a conceptual diagram showing processing contents of a scan conversion circuit in the first embodiment of the moving picture encoding device.

FIG. 7 is a conceptual diagram showing processing contents of an extended block generation circuit.

FIG. 8 is a detailed block diagram of a motion compensation circuit in the first embodiment of the video encoding device.

FIG. 9 is a more detailed block diagram of an extended block generation circuit in the motion compensation circuit.

FIG. 10 is a detailed block diagram of a motion vector detection circuit in the first embodiment of the video encoding device.

FIG. 11 is a block diagram of a second embodiment of the moving picture coding apparatus using the motion compensation prediction coding method according to the present invention.

FIG. 12 is a block diagram of a second embodiment of a video decoding device that is paired with the video encoding device of FIG. 11;

FIG. 13 is a block diagram of a third embodiment of the moving picture coding apparatus using the motion compensated prediction coding method according to the present invention.

FIG. 14 is a block diagram of a third embodiment of the moving picture decoding device that is paired with the moving picture encoding device of FIG. 13;

FIG. 15 is a detailed block diagram of a motion adaptive line interpolation circuit in a third embodiment of the video encoding device.

[Explanation of symbols]

2 scan conversion circuit, 4 data subtraction circuit, 5 orthogonal transformation circuit, 6 quantization circuit, 7 variable length encoding circuit, 8
... Data multiplexing circuit, 9/19 ... Buffer memory, 21 ...
Variable length decoding circuit, 11 ・ 22 ... inverse quantization circuit, 12 ・
23: inverse orthogonal transform circuit, 13/25: data adder circuit,
26: scan inverse conversion circuit, 15/29: motion compensation circuit, 16: motion vector detection circuit, 32/33/34
76, 80: Motion adaptive line interpolation circuit.

Continuation of the front page (56) References JP-A-56-117487 (JP, A) Proceedings of the 1979 IEEJ General Conference, Volume 5, p. 48 1990 IEICE Spring National Convention Lecture Paper Volume 7, p. 31 (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N ^7/ 24-7/68

Claims (4)

(57) [Claims] 1. A method for generating a prediction signal in units of blocks for a field of a moving image signal scanned by interlacing using motion compensation, and encoding a prediction error between the prediction signal and a moving image signal of a current field. A first process of generating a prediction signal from a locally decoded image signal of a field one field before the encoded field, and a local decoding of a field two fields before the encoded field. A second process of generating a prediction signal from the image signal, and using both the encoded local decoded image signal of the field one field before and the encoded local decoded image signal of the field two fields before A third process for generating a prediction signal in which both signals are mixed, a first process in a moving region, a second process in a still region,
A moving image encoding method, wherein the prediction signal is generated by selectively using a third process in an intermediate region between the region and the moving region . 2. A prediction signal generation circuit for generating a prediction signal in units of blocks using motion compensation with respect to a field of a moving image signal which has been interlaced and scanned, and A prediction encoding circuit that encodes a prediction error, wherein the prediction signal generation circuit generates a prediction signal from an encoded local decoding image signal of a field one field before. a first processing for a second process of generating a prediction signal from the local decoded image signal two fields prior to the field of coded, and the local decoded image signal of one field before the field coded a third process of generating a prediction signal both signals are mixed with both the local decoded image signal two fields prior to the field encoded, and in moving areas 1 process, the second process in the still region, still
A third processing by selectively using the intermediate area of the region and the dynamic region, the moving picture coding apparatus and generates the prediction signal. 3. A prediction signal is generated in units of blocks for a field of an interlaced scanned video signal using motion compensation, and a prediction error between the prediction signal and a video signal of a current field is encoded. A first generation process of generating a first block image signal from a locally decoded image signal of a field one field before the encoded field, The second processing is performed by a mixing process using both the locally decoded image signal of the field and the locally decoded image signal of the field two fields before the encoded field.
And a selection process of switching the first block image signal and the second block image signal on a block basis and outputting the prediction signal. Moving image encoding method. 4. A prediction signal generation circuit for generating a prediction signal in units of blocks using a motion compensation for a field of a video signal interlaced and scanned, and A prediction encoding circuit that encodes a prediction error, wherein the prediction signal generation circuit is configured to perform a first block from a locally decoded image signal of an encoded field one field before. A first generation process of generating an image signal; and a mixing process using both the encoded local decoded image signal of the previous field and the encoded local decoded image signal of the previous two fields Second by
And a selection process of switching the first block image signal and the second block image signal on a block basis and outputting the prediction signal. Moving image encoding apparatus.