JP2004274411A - Picture encoder and picture decoder, and picture encoding method and picture decoding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce memory capacity used for prediction processing and reverse prediction processing. <P>SOLUTION: A motion vector prediction device 12 conducts INTER4V prediction processing by connecting the motion vector prediction device to a motion vector memory 13 storing a prediction motion vector and using motion vector information detected by a motion detection block 10. The motion vector memory 13 has a region storing a reference block portion deciding the prediction motion vector with respect to four brightness components of an 8X8 macro-block in accordance with standards. The motion vector prediction device 12 extracts the prediction motion vector used as reference and specified by MPEG4 with respect to the focused brightness component block from the motion vector memory 13 to calculate the prediction motion vector with respect to the focused brightness component block by using the extracted reference prediction motion vector. The calculated prediction motion vector is stored on a region corresponding to the motion vector memory 13 and is referred to in calculating the prediction motion vector of the following focused blocks. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image encoding device, an image decoding device, an image encoding method, and an image decoding method, and more particularly to an image encoding device that divides an entire image into predetermined blocks, performs prediction processing for each block, and encodes the image. The present invention relates to an image encoding method, an image decoding device that decodes an encoded image signal, and an image decoding method thereof.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, related organizations have been working on standardization of a video / audio coding system, and practical use of the standard has been promoted in accordance with established standards.
[0003]
Of these, MPEG4 has been established by the ISO / IEC Moving Image Compression Standards Group, MPEG (Moving Picture Experts Group). The MPEG4 standard is defined and standardized in the standard document "ISO / IEC14496 Information Technology-Generic Coding of Audio-Visual Object". However, in the MPEG4 standard described above, the standard itself specifies a decoding method, and there is no specification regarding a method of implementing encoding. Further, in the book "MPEG-4 Everything (edited by Michiru Yuichi, 1998, published by the Industrial Research Institute, Inc.)", although the principle of encoding is explained, there is no explanation on the mounting method.
[0004]
As described above, in the standard book and the reference book, the memory management method used for the prediction processing is not specified, so that the memory area can be freely implemented. Therefore, a method of securing a memory area for an image size is conceivable.
[0005]
However, in the image encoding device and the image decoding device that realize the prediction process and the inverse prediction process, the memory cannot be mounted without limitation. Further, since a device having a small circuit scale and low power consumption is desired, a reduction in memory capacity is required. For this reason, an image decoding apparatus that reduces the memory capacity used for prediction processing when decoding an encoded image signal encoded by Discrete Cosine Transform (DCT), quantization, and variable-length encoding, and An image decoding method has been proposed (for example, see Patent Document 1).
[0006]
In such an image decoding device, every time the prediction process for one line is completed, the reference value used for the prediction process of the next target macroblock is copied to the storage unit for storing the DC value of the entire image size. The area for storing the (DC) component and the alternating current (AC) component is reduced, and the memory capacity is reduced.
[0007]
[Patent Document 1]
JP-A-2002-118853 (pages 11 to 15, FIGS. 7 to 12)
[0008]
[Problems to be solved by the invention]
However, the conventional image encoding device and image decoding device have a problem that the memory capacity is still large even if the above method is performed.
[0009]
In MPEG4, in addition to the conventional motion compensation in units of macroblocks, motion compensation of a prediction signal is performed on each block of four luminance components (Y0, Y1, Y2, Y3) using a motion vector. can do. Hereinafter, the prediction processing and the inverse prediction processing performed by such motion compensation are referred to as INTER4V prediction processing and INTER4V inverse prediction processing in which four vectors (Vectors) are provided for the normal Inter.
[0010]
The memory configuration required for the INTER4V prediction process and the inverse prediction process will be described. FIG. 16 is a diagram showing a memory configuration used in the INTER4V prediction process. In the INTER4V prediction process, a motion vector of the target macroblock 100 to be predicted is prepared for four luminance components forming the macroblock. In the example of the figure, the motion vectors of the blocks B0 (101), B1 (102), B2 (103), and B3 (104) are stored in the memory for each luminance component. As the prediction motion vector of the macro block of interest 100, the median of three reference motion vectors included in the reference macro blocks 110, 120, and 130 adjacent to the macro block of interest 100 is obtained. The motion vector referred to at this time is defined according to the luminance component blocks (B0 (101), B1 (102), B2 (103), and B3 (104)) of the macroblock of interest 100 to be predicted. It depends on the luminance component block. Therefore, in order to calculate the predicted motion vector, it is necessary to hold the motion vectors of the adjacent reference macroblocks 110, 120, and 130.
[0011]
However, the MPEG4 standard does not specify a memory management method in the INTER4V prediction process. Further, in the conventional image decoding device described above, a method for reducing the memory capacity used in the DC / AC prediction processing is described, and the memory capacity used for INTER4V prediction is not mentioned. For this reason, it is conceivable that the necessary memory area is mounted for the number of macro blocks existing in the frame.
[0012]
Therefore, the memory area for storing the predicted motion vector used for the INTER4V prediction is implemented as the frame memory 200 for the number of macro blocks constituting the frame.
[0013]
In the INTER4V prediction process, a difference motion vector between the predicted motion vector calculated using the motion vector of the reference macroblock and the motion vector of the target block is calculated. Then, the differential motion vector is subjected to variable-length encoding to generate an encoded image signal.
[0014]
On the other hand, in the INTER4V inverse prediction process, the motion vector of the block of interest is restored from the predicted motion vector calculated using the motion vector of the same reference macroblock as in the INTERV4 prediction process and the variable-length decoded difference motion vector. That is, the INTER4V prediction process and the INTER4V inverse prediction process are the same for the macroblock to be referred to. Therefore, the memory area used for the INTER4V reverse prediction processing is also mounted in the same manner as the INTER4V prediction processing.
[0015]
However, in the conventional method of implementing the memory area for storing the prediction motion vector of the reference macroblock adjacent to the macroblock of interest for the number of macroblocks existing in the frame and performing the prediction processing and the inverse prediction processing, encoding and decoding are performed. As the number of frames to be processed increases, the memory capacity required for processing increases. If the memory capacity increases, the number of mounted memories must be increased, which causes a problem that power consumption increases and a chip size increases. As a result, there arises a problem that the cost of the image encoding device and the image decoding device increases.
[0016]
The present invention has been made in view of such a point, and an image encoding apparatus, an image decoding apparatus, an image encoding method, and an image decoding method capable of reducing a memory capacity used for prediction processing and inverse prediction processing. The purpose is to provide.
[0017]
[Means for Solving the Problems]
In the present invention, in order to solve the above problem, in an image encoding apparatus that divides the entire image into predetermined blocks, performs prediction processing for each of the blocks, and performs encoding, the image encoding apparatus defines a prediction block according to a target block to be predicted. A prediction unit that extracts a prediction value calculated for a reference block to be calculated as a reference value, and calculates a prediction value of the block of interest based on the extracted prediction value; and the prediction value referred to by the prediction unit. And a predicted value storage unit for storing the reference blocks required according to a prescribed rule.
[0018]
In the image coding apparatus having such a configuration, the prediction unit extracts, as a reference value used in the prediction processing, a previously calculated prediction value stored for a reference block defined according to a target block to be predicted. Then, a predicted value of the target block of the prediction target is calculated based on the extracted predicted value. For this reason, the prediction value storage means secures a storage area capable of storing the prediction values referred to in the prediction processing for the reference blocks required according to the regulations, and sets the target prediction value by the time the prediction processing is executed. Is stored.
[0019]
Further, in order to solve the above-described problem, an image decoding device that decodes an image encoded signal encoded by the image encoding device described above is provided. In this image decoding apparatus, the inverse prediction means extracts an inverse prediction value of a reference block defined according to a target block to be predicted based on the same rule as that of the image encoding apparatus, and generates a target based on the inverse prediction value. Of the block of interest is calculated. Therefore, as the predicted value storage means, a storage area capable of storing the inverse predicted value for the reference block required according to the regulation is secured, as in the image encoding device.
[0020]
In order to solve the above-mentioned problem, in an image encoding method in which an entire image is divided into predetermined blocks, and a prediction process is performed for each of the blocks to perform encoding, a reference block defined according to the target block is provided. Extracting, from the predicted value storage means stored for the reference block for which the predicted value calculated for is referred to in accordance with the regulation, the predicted value of the reference block to be processed in accordance with the regulation; Calculating a predicted value of the target block based on the predicted value of the reference block; and storing the calculated predicted value of the target block in a storage area of the predicted value storage unit corresponding to the target block. And performing an encoding process based on a difference between the value of the block of interest and the calculated predicted value. Picture coding method for, is provided.
[0021]
In the image coding method of such a procedure, the prediction value calculated for the reference block defined according to the target block is stored in the prediction storage unit used for the prediction process for the reference block required according to the definition. Is stored. In the prediction processing, a predicted value of a reference block referred to in accordance with a rule is extracted from predicted value storage means, and a predicted value of a target block is calculated based on the extracted predicted value. The calculated predicted value is stored in the storage area of the predicted value storage means corresponding to the block of interest. The predicted value calculated and stored in this way is referred to in the subsequent prediction process of the target block. Note that encoding can be performed by taking the difference between the predicted value of the target block calculated in this way and the original value of the target block.
[0022]
Further, in order to solve the above problem, an image decoding method for decoding an image coded signal coded by the image coding method described above is provided.
In this image decoding method, an inverse prediction value is extracted from a prediction value storage unit that stores an inverse prediction value of a reference block defined according to a target block to be inversely predicted based on the same rule as the image encoding method. , The inverse prediction value of the target block of interest is calculated based on the inverse prediction value. The calculated inverse prediction value is stored in the storage area of the prediction value storage means corresponding to the target block, and is referred to in the subsequent inverse prediction processing.
[0023]
BEST MODE FOR CARRYING OUT THE INVENTION
An embodiment of the present invention will be described with reference to the drawings. Hereinafter, as an embodiment, an image encoding device will be described, followed by an image decoding device, taking an example of INTER4V prediction processing and inverse prediction processing used for MPEG4 motion compensation.
[0024]
An image encoding device that implements the INTER4V prediction process will be described. FIG. 1 is a configuration diagram of an image encoding device according to an embodiment of the present invention.
An image coding apparatus according to the present invention includes a subtracter 1, a discrete cosine transform (DCT) block 2, a quantizer 3, a DC / AC predictor 4, a variable length coder 5, an inverse quantizer 6, and an inverse discrete Cosine transform (inverse DCT) block 7, adder 8, video memory 9, motion detection block 10, motion compensation block 11, motion vector predictor 12 which is prediction means for predicting a motion vector, and prediction referred to at the time of motion vector prediction. It has each processing unit of the motion vector memory 13 which is a predicted value storage means for storing a value.
[0025]
The subtractor 1 receives an external input signal (luminance signal, color difference signal in the example in the figure) as a first input, an output signal of the motion compensation block 11 as a second input signal, and an external input signal. Or the difference between the external input signal and the motion compensation block 11 is output to the DCT block 2. The DCT block 2 performs a discrete cosine transform on the signal input from the subtractor 1 and outputs the signal to the quantizer 3. The quantizer 3 quantizes the DCT coefficient obtained by the DCT transform, and outputs the quantized DCT coefficient to the DC / AC predictor 4 and the inverse quantizer 6. The DC / AC predictor 4 calculates a quantized value of a direct current (DC) component and an alternating current (AC) component based on the quantized signal, and supplies the quantized value to a first input of the variable length encoder 5. Here, the variable-length encoder 5 performs variable-length encoding on the quantized DCT coefficients and the quantization width. This is intra (Intra) coding, and the coded VOP (Video Object Plane) is called an I-VOP. In the I-VOP process, zero is supplied to the second input of the subtractor 1.
[0026]
On the other hand, by passing the output signal of the quantizer 3 through the inverse quantizer 6 and the inverse DCT block 7, a signal before input to the DCT block 2 is obtained. This signal is provided to a first input of adder 8. The original image signal is obtained by the adder 8 and stored in the video memory 9 as a reference image signal of the motion detection block 10 and the motion compensation block 11. The motion detection block 10 detects a motion portion of an image and generates motion vector information. The motion vector information is sent to the motion compensation block 11 and the motion vector predictor 12. The motion compensation block 11 generates a motion-compensated reference image signal using the motion vector information, and outputs the reference image signal to the subtractor 1 and the adder 8. The subtractor 1 calculates a difference (prediction error signal) between the image signal and the reference image signal. This difference signal is sent to the variable-length encoder 5 via the DCT block 2, the quantizer 3, and the DC / AC predictor 4.
[0027]
Further, the motion vector predictor 12 is connected to a motion vector memory 13 for storing a predicted motion vector, and performs an INTER4V prediction process using the motion vector information detected by the motion detection block 10. The motion vector memory 13 has an area for storing prediction operation vectors for four luminance components of an 8 × 8 macro block for reference blocks determined according to the standard. In the INTER4V prediction process performed by the motion vector predictor 12, for a target luminance component block, a reference predicted motion vector defined by MPEG4 is extracted from the motion vector memory 13, and the target reference component is extracted using the extracted reference predicted motion vector. A predicted motion vector for a luminance component block is calculated. The calculated predicted motion vector is stored in a corresponding area of the motion vector memory 13 and is referred to when calculating a predicted motion vector of a target block thereafter. As described above, the calculated predicted motion vectors are sequentially stored in the motion vector memory 13 and are set as the next reference values, so that the storage capacity of the motion vector memory 13 is changed to a capacity required for prediction, for example, one horizontal column macro block. Minutes. Details of the processing will be described later.
[0028]
Finally, a difference motion vector between the target luminance component block and the predicted motion vector is calculated, and output to the variable-length encoder 5. The variable-length encoder 5 performs variable-length coding on the quantized DCT coefficients of the difference signal together with the motion vector and the quantization width. This is called inter-VOP coding or inter coding.
[0029]
As described above, in the image coding apparatus according to the present invention, the predicted motion vector storage amount of the motion vector memory 13 used in the motion vector predictor 12 stores the conventional predicted motion vector of the entire image. It is possible to make much smaller than the storage amount in the case.
[0030]
Next, the image decoding apparatus will be described. FIG. 2 is a configuration diagram of the image decoding device according to the embodiment of the present invention.
The image decoding apparatus according to the present invention includes a variable length decoder 20, a DC / AC inverse predictor 21, an inverse quantizer 22, an inverse discrete cosine transform (inverse DCT) block 23, an adder 24, a video memory 25, a motion compensation The processing unit includes a block 26, a motion vector inverse predictor 27 as inverse prediction means for inversely predicting a motion vector, and a motion vector memory 28 for storing an inverse prediction value referred to at the time of motion vector inverse prediction.
[0031]
The variable length decoder 20 receives the video bit stream encoded by the image encoding device shown in FIG. 1 as input and performs variable length decoding. The signal decoded by the variable length decoder 20 is output to the DC / AC inverse predictor 21 and the motion vector inverse predictor 27.
[0032]
The DC / AC inverse predictor 21 inversely predicts a quantization value for a direct current (DC) component and an alternating current (AC) component of the signal decoded by the variable length decoder 20, and outputs the result to the inverse quantizer 22. The inverse quantizer 22 performs an inverse quantization process on the calculated quantization value, and outputs the result to an inverse DCT block 23 that performs an inverse discrete cosine transform. The luminance signal and the color difference signal decoded via the inverse DCT block 23 and the adder 24 are stored in the video memory 25 as a reference image signal.
[0033]
The motion vector inverse predictor 27 is connected to a motion vector memory 28 that stores the inversely predicted inverse predicted motion vector, performs INTER4V inverse prediction on the variable-length decoded signal, and restores the motion vector. Like the motion vector memory 13 of the image encoding apparatus, the motion vector memory 28 stores prediction motion vectors for four luminance components of an 8 × 8 macro block for reference blocks determined according to the standard. In the INTER4V inverse prediction process by the motion vector inverse predictor 27, an inverse prediction motion vector to be defined by MPEG4 as a reference is extracted from the motion vector memory 28 for the target luminance component block, and the extracted reference inverse prediction motion vector is extracted. To calculate the inverse prediction motion vector for the target luminance component block. The calculated inverse prediction motion vector is stored in the corresponding area of the motion vector memory 28, and is referred to when calculating the inverse prediction motion vector of the target block thereafter. In this way, the calculated inversely predicted motion vectors are sequentially stored in the motion vector memory 28 and used as the next reference value, so that the storage capacity of the motion vector memory 28 can be reduced to the capacity required for prediction, for example, one horizontal column macro. It becomes possible to make it for a block. Details of the processing will be described later.
[0034]
Then, the motion vector of the luminance component block of interest is decoded based on the differential motion vector and the inverse prediction motion vector subjected to the variable length decoding, and output to the motion compensation block 26. The motion compensation block 26 acquires an inversely predicted macroblock based on the motion vector information and outputs the macroblock to the adder 24. This signal is added to the difference signal by the adder 24 to become the original image signal.
[0035]
As in the case of the image encoding apparatus shown in FIG. 1, the motion vector memory 28 used in the motion vector de-prediction unit 27 is smaller than the storage amount when the conventional de-prediction motion vector of the entire image is stored. Thus, the memory capacity can be much smaller.
[0036]
Here, the macroblock referred to by the motion vector predictor 12 and the motion vector inverse predictor 27 will be described. Note that the blocks to be referred to are the same in both cases, and hence the description will be made in the case of the motion vector prediction processing. FIG. 3 is a diagram showing macroblocks referred to in the motion vector prediction processing.
[0037]
In the INTER4V prediction process of predicting the motion vector of the block of interest, a predicted motion vector of a motion vector for a block (B0, B1, B2, B3) corresponding to four luminance components forming one macro block 100 is calculated. As the predicted motion vector of the target block of interest, the median value is obtained by referring to the motion vectors (V0, V1, V2) of the macroblocks (left, upper, upper right) adjacent to the block including the motion vector. The motion vector referred to at this time differs depending on the target motion vector of interest. As shown in the drawing, when the target motion vector (101) of the block 0 (B0) is the target, the motion vector V0 (111) of the macroblock adjacent to the left and the motion vector V1 (112) of the macroblock adjacent to the upper left. ) And the motion vector V2 (113) of the macroblock adjacent to the upper right. When the target motion vector (102) of the block 1 (B1) is the target, the motion vector V0 (114), the motion vector V1 (115), and the motion vector V2 (116) are referred to. Similarly, in the target motion vector 103 of the block 2 (B2), the motion vector V0 (117), the motion vector V1 (118), and the motion vector V2 (119) are referred to, and in the target motion vector 104 of the block 3 (B3), , The motion vector V0 (120), the motion vector V1 (121), and the motion vector V2 (122).
[0038]
In the first embodiment of the present invention, focusing on the relationship between the target block described above and a reference block defined according to the target block, the motion vectors referred to by the motion vector predictor 12 and the motion vector inverse predictor 27 are described. The memory capacity of the vector memories 13 and 28 is set to be equal to one horizontal row of the macro block.
[0039]
FIG. 4 is a configuration diagram of a motion vector storage area according to the first embodiment of this invention.
The motion vector memory 13a stores four areas for storing predicted motion vectors corresponding to four luminance components forming the macro block 100, a block B0 (101), a block B1 (102), a block B2 (103), and a block B3. (104) includes a memory area capable of storing a predicted motion vector for one horizontal column of a macroblock. That is, assuming that the number of macroblocks for one horizontal column is N, the motion vector memory 13a is configured of a memory area for storing N predicted motion vectors for each block. As shown in the figure, the horizontal column memory of the block B0 having the memory area of B0 [0], B0 [1],..., B0 [n],. 131, a horizontal column memory 132 of the block B1 having a memory area of B1 [0], B1 [1],..., B1 [n],. B2 [0], B3 [1],..., B2 [n], the horizontal column memory 133 of the block B2 having the memory area of B2 [N-1], and B3 [0], B3 [1 ,..., B3 [n] and B3 [N−1]. Here, n indicates any one block.
[0040]
Next, the memory management of the motion vector memory 13a having such a configuration will be described. FIG. 5 is a diagram showing a method for managing a motion vector memory according to the first embodiment of this invention. (A) shows the current macro block of interest and its reference macro block, and (B) shows the next macro block of interest and its reference macro block. As described above, an image is divided into predetermined macroblocks (hereinafter, MBs). In the example shown in the figure, the horizontal rows of MBs are 0, 1,..., N,.
[0041]
When the predicted motion vector processing is performed on the current attention MB [n] 301 (the n-th MB in the horizontal column m + 1) shown in FIG. 3A, the storage area of the motion vector memory 13 provided in one horizontal column includes Predicted motion vectors for one horizontal column MB are stored in order from MB [n] in the horizontal column m. In the area corresponding to the block MB [n-1], the predicted motion vector of the block MB [n-1] of the horizontal column m + 1, which is the target block in the previous process, is stored. In the example of the figure, the predicted motion vectors of the reference MB except for the MB of interest [n] are stored in the order of the reference MB [n-1] 302, the reference MB [n] 303, and the reference MB [n + 1] 304. At the start of the prediction process, the memory pointer of the motion vector memory 13 points to n.
[0042]
To calculate the predicted motion vector of the MB [n] 301 of interest, the predicted motion of the reference MB [n-1] 302, the reference MB [n] 303, and the reference MB [n + 1] 304 adjacent to the MB [n] of interest 301 The vector is chosen as the reference prediction motion vector. In this way, a specified predicted motion vector existing in the reference MB is extracted, and the median value is calculated. The calculated median value is stored in the storage area of the block [n] indicated by the pointer as the predicted motion vector of the MB [n] of interest 301.
[0043]
This is the state of FIG. The memory pointer of the motion vector memory 13 advances by one, and n points to the next target MB [n] 305. The previous attention MB [n] 301 becomes the reference MB [n-1] 301. Further, the MB 304 is the reference MB [n + 1] in (A), but becomes the reference MB [n] because the pointer advances by one. Similarly, the reference MB [n + 1] points to the next MB 306.
[0044]
As described above, the motion vector predictor 12 calculates the predicted motion vector with reference to the storage area (motion vector memory 13a) of the predicted motion vector provided for one horizontal column of the macroblock, and calculates the calculated predicted motion vector. The memory capacity of the motion vector memory 13a can be reduced by storing it in the motion vector memory 13a and referencing it in the next block of interest.
[0045]
Next, the INTER4V prediction processing performed by the motion vector predictor 12 will be described. FIG. 6 and FIG. 7 are diagrams illustrating the operation of the motion vector prediction process in the image encoding device according to the first embodiment of the present invention. 6 (A) shows the operation of the motion vector prediction process of block B0, FIG. 6 (B) shows the operation of block B1, FIG. 7 (C) shows the operation of block B2, and FIG. 7 (D) shows the operation of the motion vector prediction process of block B3. It is assumed that the memory structure of the motion vector memory 13a used is that shown in FIG.
[0046]
Since the processing in the motion vector predictor 12 is executed in the order of (A), (B), (C), and (D), the processing will be described in the order of execution.
(A) In the motion prediction process 12-1 of the block B0, B1 [n-1] on the left of the block B0, B2 [n] on the upper right, and B2 [n + 1] on the upper right of the block B0 are selected as the reference motion vectors. Therefore, these predicted motion vectors are read from the horizontal column memory 132 of the block B1 and the horizontal column memory 133 of the block B2, and the median value is calculated. The calculated predicted motion vector of the block B0 is written into the horizontal column memory 131 of the block B0 at the position of B0 [n].
[0047]
In the subsequent (B) motion prediction processing 12-2 of the block B1, as a reference prediction motion vector, B0 [n] on the left of the block B1 and calculated in the processing of (A), and B3 [n] above , B2 [n + 1] at the upper right are selected, and these predicted motion vectors are read from the horizontal column memory 131 of the block B0, the horizontal column memory 134 of the block B3, and the horizontal column memory 133 of the block B2. Then, the median is calculated as the predicted motion vector, and the calculated predicted motion vector is written to the position of B1 [n] in the horizontal column memory 132 of the block B1.
[0048]
In the next (C) motion prediction processing 12-3 of the block B2, as a reference prediction motion vector, B3 [n-1] on the left of the block B2, B0 [n] on the upper right, and B0 [n] on the upper right and The predicted motion vector of B1 [n] calculated by the processing is read from the horizontal column memory 134 of the block B3, the horizontal column memory 131 of the block B0, and the horizontal column memory 132 of the block B1. These median values are calculated and written as the predicted motion vector of the block B2 at the position of B2 [n] in the horizontal column memory 133 of the block B2.
[0049]
In the motion prediction processing 12-4 of the last (D) block B3, B2 [n] on the left side, B0 [n] on the upper left, and B1 [n on the upper left, which have been calculated in the previous processing, are used as reference prediction motion vectors. ] Are read from the horizontal column memory 133 of the block B2, the horizontal column memory 131 of the block B0, and the horizontal column memory 132 of the block B1. These medians are calculated and written as the predicted motion vector of the block B3 at the position of B3 [n] in the horizontal column memory 134 of the block B3.
[0050]
As described above, according to the image encoding device and the image decoding device of the first embodiment of the present invention, the memory capacity used in the INTER4V prediction processing and the inverse prediction processing is set to one horizontal row of a macroblock. And the memory capacity can be significantly reduced. As a result, a device with a small circuit size and low power consumption can be provided.
[0051]
Next, processing procedures of the INTER4V prediction processing and the inverse prediction processing used in the image encoding method and the image decoding method according to the first embodiment of the present invention will be described. FIGS. 8 and 9 are flowcharts showing the processing procedure of the motion vector prediction method in the image encoding device according to the first embodiment of the present invention. FIG. 8 shows a processing procedure for blocks B0 and B1, and FIG. 9 shows a processing procedure for blocks B2 and B3.
[0052]
The processing is started when each motion vector is detected.
[Step S01] The process of block B0 is started. As the reference vector of the block of interest B0, predicted motion vectors of three reference blocks B1 [n-1], B2 [n], and B2 [n + 1] are read from the horizontal column memory of each block.
[Step S02] The motion prediction processing of the block of interest B0 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B0 is obtained.
[Step S03] In the subsequent processing, the predicted motion vector of the block B0 serving as a reference vector is written to B0 [n] of the horizontal column memory of the block B0.
[Step S04] The process of block B1 is started. As the reference vector of the block of interest B1, predicted motion vectors of three reference blocks B0 [n], B3 [n], and B2 [n + 1] are read from the horizontal column memory of each block.
[Step S05] The motion prediction processing of the block of interest B1 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B1 is obtained.
[Step S06] The predicted motion vector of the block B1 serving as a reference vector in the subsequent processing is written to B1 [n] of the horizontal column memory of the block B1.
[0053]
The processing proceeds to the processing in FIG. 9 via the point A.
[Step S07] The process of the block B2 is started. As the reference vector of the block of interest B2, predicted motion vectors of three reference blocks B3 [n-1], B0 [n], and B1 [n] are read from the horizontal column memory of each block.
[Step S08] The motion prediction processing of the block of interest B2 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B2 is obtained.
[Step S09] The predicted motion vector of the block B2 serving as a reference vector in the subsequent processing is written to B2 [n] of the horizontal column memory of the block B2.
[Step S10] The process of the block B3 is started. As the reference vector of the block of interest B3, predicted motion vectors of three reference blocks B2 [n], B0 [n], and B1 [n] are read from the horizontal column memory of each block.
[Step S11] The motion prediction processing of the block of interest B3 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B3 is obtained.
[Step S12] In the subsequent processing, the predicted motion vector of the block B3 serving as a reference vector is written to B3 [n] of the horizontal column memory of the block B3.
[0054]
Note that the case of the INTER4V inverse prediction is the same as the case of the INTER4V prediction process, except that the motion prediction process of the above procedure is the motion inverse prediction process, and thus the description is omitted.
[0055]
As described above, according to the image encoding method and the image decoding method of the first embodiment of the present invention, the memory capacity used in the INTER4V prediction processing and the inverse prediction processing is set to one horizontal row of a macroblock. be able to.
[0056]
Here, the relationship between the block of interest and the reference block will be examined in more detail. FIG. 10 is a diagram illustrating a relationship between a target block and a reference block according to the embodiment of the present invention.
[0057]
Assuming that the prediction target is the target MB [n] 301, the reference MBs [n−1] 302, the reference MB [n] 303, and the reference MB [n + 1] 304 are selected as the referenced MBs. When this is expressed in units of luminance component blocks, B1 [of the reference MB [n-1] 302 is calculated in order to calculate a predicted motion vector for the 8 × 8 block B0, B1, B2, B3 of the MB [n] 301 of interest. n-1] and B3 [n-1], B2 [n] and B3 [n] of reference MB [n] 303, B2 [n + 1] of reference MB [n + 1] 304, and B0 of target MB [n] 301 , B1, and B2 are used.
[0058]
From this, the luminance component block referred to for one MB is B0 in the block B0 and B1 [n-1] in the block B1, and each requires a memory area for storing a predicted motion vector for one block. You can see that. Further, since the block B2 refers to B2 and B2 [n + 1], and the block B3 refers to B3 [n-1] and B3 [n], a memory area for storing a predicted motion vector of one horizontal column memory is required. You can see that there is.
[0059]
From the above, in the second embodiment of the present invention, the memory configuration used for the INTER4V prediction processing and the inverse prediction processing is divided into two memories for storing one block of predicted motion vectors and one horizontal column of blocks. There are two memories for storing the predicted motion vector.
[0060]
FIG. 11 is a configuration diagram of a motion vector storage area according to the second embodiment of this invention. The same components as those in FIG. 4 are denoted by the same reference numerals, and description thereof is omitted.
The motion vector memory 13b according to the second embodiment has a memory area for storing a predicted motion vector for each of the four blocks B0 (101), B1 (102), B2 (103), and B3 (104) constituting the MB 100. Is provided. Specifically, for the block B0, the motion vector memory 13b includes a one-block memory 135 for B0 that stores one block of predicted motion vectors. Similarly, the block B1 includes one block memory 136 of B1 for storing the predicted motion vector of one block. On the other hand, the blocks B2 and B3 include a memory for one horizontal column, a horizontal column memory 133 for B2, and a horizontal column memory 134 for B3, as in the first embodiment.
[0061]
As described above, by securing only one block memory for the blocks B0 and B1, the memory capacity of the first embodiment can be further reduced to about half.
[0062]
Next, the INTER4V prediction processing by the motion vector predictor 12 using the motion vector storage area having such a configuration will be described.
FIG. 12 and FIG. 13 are diagrams illustrating the operation of the motion vector prediction process in the image encoding device according to the second embodiment of the present invention. 12 (A) shows the operation of the block B0, FIG. 12 (B) shows the operation of the block B1, FIG. 13 (C) shows the operation of the block B2, and FIG. 13 (D) shows the operation of the motion vector prediction process of the block B3. The processing is performed in the order of (A), (B), (C), and (D).
[0063]
(A) In the motion prediction process 12-5 of the block B0, B1 [n-1] on the left side of the block B0, B2 [n] on the upper side, and B2 [n + 1] on the upper right of the block B0 are selected as reference prediction motion vectors. Therefore, these predicted motion vectors are stored in the one block memory 136 of the block B1 (before the B1 processing of the block of interest, the one block memory 136 of the block B1 stores the predicted motion vector of the previous target MB [n-1]. Stored) and from the horizontal column memory 133 of the block B2 to calculate the median value. The calculated predicted motion vector of the block B0 is written to the one-block memory 135 of the block B0.
[0064]
In the following (B) motion prediction processing 12-6 of the block B1, as a reference prediction motion vector, B3 calculated on the processing B of FIG. B2 [n + 1] is selected, and these predicted motion vectors are read from one block memory 135 of block B0, horizontal column memory 134 of block B3, and horizontal column memory 133 of block B2. Then, the median is calculated as the predicted motion vector, and the calculated predicted motion vector is written to the one-block memory 136 of the block B1.
[0065]
In the next (C) motion prediction processing 12-7 of the block B2, the reference prediction motion vector is calculated by the processing of (B) in B3 [n-1] on the left of the block B2, on B0 on the upper right, and on the upper right. The predicted motion vector of B1 is read from the horizontal column memory 134 of block B3, the one block memory 135 of block B0, and the one block memory 136 of block B1. These median values are calculated and written as the predicted motion vector of the block B2 at the position of B2 [n] in the horizontal column memory 133 of the block B2.
[0066]
In the motion prediction processing 12-8 of the last (D) block B3, the prediction motion vectors of B2 [n] on the left side, B0 on the upper left, and B1 on the upper left calculated in the processing so far are used as reference prediction motion vectors. , The horizontal column memory 133 of the block B2, the one block memory 135 of the block B0, and the one block memory 136 of the block B1. These medians are calculated and written as the predicted motion vector of the block B3 at the position of B3 [n] in the horizontal column memory 134 of the block B3.
[0067]
In the INTER4V inverse prediction processing, the calculation of the predicted value is only the calculation of the inverse predicted value, and the operation of the referenced block is the same.
As described above, according to the image encoding device and the image decoding device of the second embodiment of the present invention, the memories used in the INTER4V prediction processing and the inverse prediction processing include two block memories and two horizontal columns. A memory can be configured, and the memory capacity can be further reduced as compared with the first embodiment. As a result, a device with a small circuit size and low power consumption can be provided.
[0068]
Next, the processing procedure of the INTER4V prediction processing and the inverse prediction processing used in the image encoding method and the image decoding method according to the second embodiment of the present invention will be described. FIGS. 14 and 15 are flowcharts showing processing procedures of a motion vector prediction method in the image encoding device according to the second embodiment of the present invention. FIG. 14 shows a processing procedure for blocks B0 and B1, and FIG. 15 shows a processing procedure for blocks B2 and B3.
[0069]
The processing is started when each motion vector is detected.
[Step S21] The process in block B0 is started. Predicted motion vectors of three reference blocks B1 [n-1], B2 [n], and B2 [n + 1] are read from the block memory and horizontal column memory of each block as reference vectors of the block of interest B0.
[Step S22] The motion prediction processing of the block of interest B0 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B0 is obtained.
[Step S23] The predicted motion vector of the block B0 serving as a reference vector in the subsequent processing is written to one block memory B0 of the block B0.
[Step S24] The process of block B1 is started. As the reference vector of the block of interest B1, predicted motion vectors of three reference blocks B0, B3 [n] and B2 [n + 1] are read from the block memory and the horizontal column memory of each block.
[Step S25] The motion prediction processing of the block of interest B1 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B1 is obtained.
[Step S26] In the subsequent processing, the predicted motion vector of the block B1 serving as a reference vector is written to one block memory B1 of the block B1.
[0070]
The processing proceeds to the processing in FIG. 15 via the point B.
[Step S27] The process of block B2 is started. As the reference vector of the block of interest B2, prediction motion vectors of three reference blocks B3 [n-1], B0, and B1 are read from the block memory and the horizontal column memory of each block.
[Step S28] The motion prediction processing of the block of interest B2 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B2 is obtained.
[Step S29] The predicted motion vector of the block B2 serving as a reference vector in the subsequent processing is written to B2 [n] of the horizontal column memory of the block B2.
[Step S30] The process of block B3 is started. As the reference vector of the block of interest B3, predicted motion vectors of three reference blocks B2 [n], B0, and B1 are read from the block memory and the horizontal column memory of each block.
[Step S31] The motion prediction processing of the block of interest B3 is performed, the median of the read reference vectors is calculated, and the predicted motion vector of the block B3 is obtained.
[Step S32] The predicted motion vector of the block B3 serving as a reference vector in the subsequent processing is written to B3 [n] of the horizontal column memory of the block B3.
[0071]
Note that the case of the INTER4V inverse prediction is the same as the case of the INTER4V prediction process, except that the motion prediction process in the above procedure is the motion inverse prediction process, and thus the description is omitted.
[0072]
As described above, according to the image encoding method and the image decoding method of the second embodiment of the present invention, the memory capacity used in the INTER4V prediction process and the inverse prediction process is further reduced as compared with the first embodiment. can do.
[0073]
In the above description, MPEG4 INTER4V prediction and inverse prediction have been described as examples, but the present invention is not limited to this, and generates the next prediction value by referring to the prediction value calculated in the previous processing. It can be applied in image signal processing.
[0074]
Note that the above processing functions can be realized by a computer. In that case, a program is provided that describes the processing contents of the functions that the image encoding device and the image decoding device should have. By executing the program on a computer, the above processing functions are realized on the computer. The program describing the processing content can be recorded on a computer-readable recording medium. The computer that executes the program stores, for example, the program recorded on the portable recording medium or the program transferred from the server computer in its own storage device. Then, the computer reads the program from its own storage device and executes processing according to the program.
[0075]
【The invention's effect】
As described above, the image coding apparatus according to the present invention performs prediction using the prediction value calculated for the reference block defined according to the target block to be predicted. At this time, an area capable of storing a required prediction value for the reference block is secured as a storage unit for holding the predicted value to be referenced, based on a rule for determining the reference block. As described above, since it is sufficient that the predicted value of the reference block required for the prediction based on the rule is secured, the storage area required for the prediction process can be reduced. Further, since the storage area can be reduced, it is possible to realize an image encoding device with a small circuit size and low power consumption.
[0076]
Further, the image decoding apparatus of the present invention performs inverse prediction on the coded image signal generated by the above-described image coding apparatus, using the reverse prediction value of the reference block defined in the same manner. For this reason, the storage area of the reference block including the inverse prediction value to be referred to is configured in the same manner as the image encoding device. As a result, it is possible to reduce the storage area required for the inverse prediction processing, and to realize an image decoding device with a small circuit size and low power consumption.
[0077]
Further, in the image encoding method of the present invention, the prediction value of the target block is calculated using the prediction value of the reference block defined according to the target block of the prediction target, and the calculated prediction value corresponds to the target block. Store in the predicted value storage area. This is referred to in the target block of the subsequent processing. As described above, the storage area for the reference block required in the prediction processing is secured, and the prediction processing is performed using the storage area. The calculated prediction value is stored in this storage area and referred to in the subsequent target block. . This makes it possible to reduce the storage area required for the prediction processing.
[0078]
Further, in the image decoding method of the present invention, the inverse prediction value of the block of interest is calculated using the inverse prediction value of the reference block defined in the same manner as the image encoded signal generated by the image encoding method, The calculated predicted value is stored in the inverse predicted value storage area corresponding to the block of interest. This is referred to in the target block of the subsequent processing. As described above, a storage area for the reference block required in the inverse prediction processing is secured, and the prediction processing is performed using the reserved area. The calculated inverse prediction value is stored in this storage area, refer. This makes it possible to reduce the storage area required for the inverse prediction processing.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of an image encoding device according to an embodiment of the present invention.
FIG. 2 is a configuration diagram of an image decoding device according to an embodiment of the present invention.
FIG. 3 is a diagram illustrating a motion vector referred to in an INTER4V prediction process.
FIG. 4 is a configuration diagram of a motion vector storage area according to the first embodiment of this invention.
FIG. 5 is a diagram illustrating a method of managing a motion vector memory according to the first embodiment of this invention.
FIG. 6 is a diagram illustrating an operation of a motion vector prediction process (block 0, block 1) in the image encoding device according to the first embodiment of the present invention.
FIG. 7 is a diagram illustrating an operation of a motion vector prediction process (block 2, block 3) in the image encoding device according to the first embodiment of the present invention.
FIG. 8 is a flowchart showing a processing procedure of a motion vector prediction method (block 0, block 1) in the image encoding device according to the first embodiment of the present invention.
FIG. 9 is a flowchart illustrating a processing procedure of a motion vector prediction method (block 2, block 3) in the image encoding device according to the first embodiment of the present invention.
FIG. 10 is a diagram illustrating a relationship between a target block and a reference block according to the embodiment of the present invention.
FIG. 11 is a configuration diagram of a motion vector storage area according to the second embodiment of this invention;
FIG. 12 is a diagram illustrating an operation of a motion vector prediction process (block 0, block 1) in the image encoding device according to the second embodiment of the present invention.
FIG. 13 is a diagram illustrating an operation of a motion vector prediction process (block 2, block 3) in the image encoding device according to the second embodiment of the present invention.
FIG. 14 is a flowchart illustrating a processing procedure of a motion vector prediction method (block 0, block 1) in the image encoding device according to the second embodiment of the present invention.
FIG. 15 is a flowchart illustrating a processing procedure of a motion vector prediction method (block 2, block 3) in the image encoding device according to the second embodiment of the present invention.
FIG. 16 is a diagram showing a memory configuration used in INTER4V prediction processing.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Subtractor, 2 ... Discrete cosine transform (DCT) block, 3 ... Quantizer, 4 ... DC / AC predictor, 5 ... Variable length encoder, 6 ... -Inverse quantizer, 7: inverse discrete cosine transform (inverse DCT) block, 8: adder, 9: video memory, 10: motion detection block, 11: motion compensation block, 12: motion vector predictor, 13: motion vector memory, 20: variable length decoder, 21: DC / AC inverse predictor, 22: inverse quantizer, 23 ... Inverse discrete cosine transform (inverse DCT) block, 24: adder, 25: video memory, 26: motion compensation block, 27: motion vector inverse predictor, 28: motion vector memory

Claims (8)

In an image encoding apparatus that divides the entire image into predetermined blocks and performs prediction processing for each of the blocks to perform encoding,
A prediction unit that extracts a prediction value calculated for a reference block defined according to a target block to be predicted as a reference value, and calculates a prediction value of the target block based on the extracted prediction value;
A prediction value storage unit that stores the prediction value referred to in the prediction unit for the reference block that is required according to a rule,
An image encoding device comprising: The prediction unit is included in a unit that performs motion compensation using motion vectors corresponding to four luminance component blocks constituting the block, and includes a prediction motion vector of a reference luminance component block defined according to the luminance component block. To generate a predicted motion vector of the target luminance component block,
2. The image coding apparatus according to claim 1, wherein the predicted value storage unit has an area for storing a predicted motion vector for each luminance component block for one horizontal column for each luminance component block. The prediction unit is included in a unit that performs motion compensation using motion vectors corresponding to four luminance component blocks constituting the block, and includes a prediction motion vector of a reference luminance component block defined according to the luminance component block. To generate a predicted motion vector of the target luminance component block,
The prediction value storage unit includes an area that stores the predicted motion vector of the one luminance component block for the luminance component block in which the prediction unit refers only to the target block and one adjacent block; 2. An area for storing the predicted motion vector of the luminance component block for one horizontal column with respect to the luminance component block to which a plurality of adjacent blocks are referred to. Image coding device. In an image decoding device that is divided into predetermined blocks and decodes an encoded image signal that has been subjected to prediction processing for each block and encoded,
An inverse prediction value calculated for a reference block defined according to the target block to be inverse predicted is extracted as a reference value, and an inverse prediction value of the target block is calculated based on the extracted inverse prediction value. Inverse prediction means;
An inverse prediction value storage unit that stores the inverse prediction value referred to in the inverse prediction unit for the reference block required according to a rule;
An image decoding device comprising: The inverse prediction means is included in a means for inversely transforming a signal motion-compensated using motion vectors corresponding to four luminance component blocks constituting the block, and includes a reference luminance defined according to the luminance component block. Using the inverse prediction motion vector of the component block to generate an inverse prediction motion vector of the luminance component block of interest,
The image decoding apparatus according to claim 4, wherein the inverse prediction value storage unit has an area for storing an inverse prediction motion vector for each luminance component block for one horizontal column for each luminance component block. The inverse prediction means is included in a means for inversely transforming a signal motion-compensated using motion vectors corresponding to four luminance component blocks constituting the block, and includes a reference luminance defined according to the luminance component block. Using the inverse prediction motion vector of the component block to generate an inverse prediction motion vector of the luminance component block of interest,
The inverse prediction value storage means includes an area for storing the inverse prediction motion vector of the one luminance component block for the luminance component block in which only the target block and one adjacent block are referred to in the inverse prediction means. And an area for storing one inverse column of the inverse prediction motion vector of the luminance component block for the luminance component block to which the plurality of adjacent blocks are referred. Item 5. The image decoding device according to Item 4. In an image encoding method in which an entire image is divided into predetermined blocks, and a prediction process is performed for each of the blocks to perform encoding,
The predicted value calculated for the reference block defined according to the target block is calculated from the predicted value storage unit stored for the reference block referred to according to the definition. Extracting a predicted value;
Calculating a predicted value for the target block based on the extracted predicted value of the reference block;
Storing the calculated predicted value of the target block in a storage area of the predicted value storage unit corresponding to the target block;
And performing an encoding process based on a difference between the value of the block of interest and the calculated predicted value. In an image decoding method for decoding an encoded image signal divided into predetermined blocks and subjected to prediction processing for each of the blocks,
When decoding the encoded image signal in which the difference between the predicted value of the target block and the value of the target block calculated from the predicted value of the reference block defined according to the target block is encoded,
The inverse prediction value calculated for the reference block defined according to the block of interest is referred to according to the regulation. Extracting a predicted value;
Calculating an inverse prediction value of the block of interest based on the extracted inverse prediction value of the reference block;
Storing the calculated inverse prediction value of the target block in a storage area of the prediction value storage unit corresponding to the target block;
And performing a decoding process based on a difference between a value of the block of interest and a predicted value obtained from the encoded image signal and an inverse prediction value of the block of interest.

Images