JPH10327416A - Dynamic image coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a dynamic image coder where limit of a retrieval range of a motion vector is reduced by detecting a motion vector over a border of division image signals so as to minimize the increase in a memory access speed of a reference image memory. SOLUTION: The proposed coder is provided with an image division section 12 that divides an input dynamic image signal into a plurality of division areas arranged in an image pattern in a direction so as to generate division image signals 111a-111d for each division area, coding sections 13a-13d that apply motion compensation prediction coding to each of the division image signals 111a-111d to provide an output of coded data 118a-118d, and a multiplexer section 14 that multiplexes the coded data 118a-118d. Then the coding sections 13a-13d store, as a reference image signals, the local decoding image signal corresponding to the division area of the coding object and the division area adjacent in the image so as to detect a motion vector between the division image signal in the division area of the coding object and the reference image signal.

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 apparatus used for recording, communication, transmission and broadcasting of a moving picture signal, and more particularly, to dividing one screen into a plurality of screens and coding each divided screen. And a video encoding device that performs

[0002]

2. Description of the Related Art Since a moving image signal has a huge amount of information,
In order to transmit or record simply by digitizing, a very wide transmission path and a large-capacity recording medium are required. Therefore, in a videophone, a video conference, a CATV, an image file device, and the like, a technology for compressing and encoding a moving image signal into a small data amount is indispensable.

[0003] As one of the compression coding techniques for a moving picture signal, a motion compensation predictive coding scheme is known. In the motion-compensated predictive coding method, a coded image is used as a reference screen, and a partial region of the input image that has the highest correlation with respect to a partial region of the input image is detected, so that the partial region of the input image is A motion vector indicating which partial region has been moved is obtained, and a prediction error signal which is a difference between the partial region of the input image and the partial region of the reference image indicated by the motion vector is encoded.

[0004] As moving images to be encoded by the moving image encoding apparatus, there are an image based on a standard TV system such as the NTSC system (standard TV image) and an HDTV (high definition television) image. Since an HDTV image has approximately six times the number of pixels as a standard TV image, the processing speed at the time of encoding and decoding needs to be approximately six times that of a standard TV image. Further, in order to make the motion vector detection follow the same motion as in the case of encoding a standard TV image when encoding an HDTV image, the search range of the motion vector needs to be approximately four times. Therefore, the processing speed required for the motion vector detection unit is about 24 times that of a standard TV image.

Therefore, in order to reduce the processing speed when encoding an HDTV image, the HDTV image is divided into a plurality of divided regions and encoded for each divided image signal, and the encoded data of each divided region is multiplexed. A moving picture coding apparatus for converting and outputting the same has been proposed (JP-A-8-9377). In this conventional moving picture coding apparatus, in order to reduce the prediction error and increase the coding efficiency, it is possible to detect a motion vector over a boundary of a divided picture signal by a method as shown in FIG.

First, as shown in FIG. 7A, an HDTV image signal, which is an input moving image signal, is divided into two in the horizontal and vertical directions, and divided into four divided areas A, B, C, and D.
Is generated. Next, as shown by hatching in FIG. 7B, a fixed pixel value (for example, at one end in the horizontal and vertical directions of the divided image signals of the divided areas A, B, C, and D).
“128”) dummy data is added to enlarge the divided image signal. On the other hand, the reference image signal used for motion vector detection is also enlarged as shown by the diagonal lines in FIG. Then, a motion vector between the enlarged divided image signal and the reference image signal is detected. By doing so, the boundary region of the divided image signal is located in the enlarged region, and it is possible to detect a motion vector that crosses the boundary. Therefore, the prediction error is smaller than the method of detecting only the motion vector in the divided region. Is reduced, and the coding efficiency is improved.

In this conventional moving picture coding apparatus, in order to detect a motion vector, four locally decoded picture signals obtained in the process of coding four divided picture signals are used as reference picture signals in each coding section. It is necessary to write simultaneously to the shared reference image memory. Further, when simultaneously searching for a motion vector between each enlarged reference image signal and each enlarged divided image signal in the reference image memory, the enlarged area of each reference image signal is as shown in FIG. Since the reference image signal overlaps with another reference image signal, it is necessary to simultaneously read the reference image signal from the reference image memory in order to detect the motion vector of each divided image signal also in this overlapping portion. That is, FIG.
In the example of (c), as the reference image signal for the divided image signal of the divided region A, divided regions B, C, and C which are horizontally and vertically adjacent to the reference image signal of the divided region A and the divided region A are used.
It is necessary to read out part of the reference image signal of D at the same time,
As a result, it is necessary to read four reference image signals from the reference image memory at the same time.

As described above, in the reference image memory, it is necessary to simultaneously write / read the reference image signals as many as the number of divisions when the input moving image signal is divided into a plurality of regions. Is required. This problem becomes more pronounced as the number of divisions increases.

In the above-mentioned conventional moving picture coding apparatus, the search range of the motion vector is expanded by the expansion of the divided image signal and the reference image signal, but the search range is expanded by the expansion of the divided image signal and the reference image signal. It is limited by the size of the enlarged area.

[0010]

As described above, a conventional moving picture encoding apparatus which divides an input moving picture signal into a plurality of divided areas inside a screen, encodes each divided picture signal, and then multiplexes the divided picture signals. Then, in order to enable motion vector detection across the boundary of the divided image signal, the divided image signal is enlarged by adding dummy data to each divided image signal, and the reference image signal is also enlarged. (A) Since it is necessary to write / read the reference image signal by the number of divisions when dividing the screen of the input moving image signal, very high-speed memory access is required. (B) The expansion of the search range of the motion vector is limited by the size of the enlarged area of the divided image signal and the reference image signal.

SUMMARY OF THE INVENTION The present invention has been made in order to solve such a conventional problem, and it is possible to detect a motion vector over a boundary of a divided image signal, thereby improving coding efficiency or image quality. Moving image encoding apparatus capable of minimizing an increase in the memory access speed of the reference image memory, and further reducing the limitation of the search range of the motion vector to obtain higher encoding performance. The purpose is to provide.

[0012]

In order to solve the above problems, a moving picture coding apparatus according to the present invention divides an input moving picture signal into a plurality of divided areas arranged in one direction on a screen, and A dividing unit that generates a divided image signal for each region, a plurality of encoding units that encode each divided image signal generated by the dividing unit by motion compensation prediction using a motion vector, and output encoded data; Multiplexing means for multiplexing and outputting the coded data respectively output from the plurality of coding means.

Here, each of the encoding means includes a local decoding means for generating a locally decoded image signal, a local decoded image signal corresponding to a division area to be encoded by the encoding means, and a division of the encoding object. A reference image storage unit that stores at least a part of a locally decoded image signal corresponding to a divided region adjacent to the region in the screen as a reference image signal; and a divided image signal of the divided region to be encoded and the reference image storage unit. A motion vector detecting means for detecting a motion vector between the reference image signal and the stored reference image signal.

In the moving picture coding apparatus configured as described above, the locally decoded image signal corresponding to the divided area to be coded and the local decoded image signal corresponding to the divided area adjacent to the divided area to be coded in the screen. By using at least a part of the image signal as the reference image signal, it is possible to detect a motion vector that straddles the boundary of the divided image signal, and to improve coding efficiency or image quality.

The reference image signal to be simultaneously written into the reference image storage means is a local image corresponding to the divided area to be coded when at most the other divided area is adjacent to the divided area to be coded from both sides of the screen. Since the number of decoded image signals and local decoded image signals corresponding to the divided regions adjacent to the encoding target divided region from both sides in the screen are three, and the number of reference image signals to be written simultaneously is at most three, the The memory access speed at the time is reduced.

In particular, in the case where only the locally decoded image signal corresponding to an area smaller than half of the adjacent divided area among the locally decoded image signals corresponding to the divided area adjacent to the encoding target divided area is stored. Since the number of reference image signals to be simultaneously written may be at most two, the memory access speed at the time of writing is further reduced.

On the other hand, since the reference picture storage means is provided independently for each coding means, it is necessary to simultaneously read a plurality of reference picture signals from a shared reference picture memory as in a conventional moving picture coding apparatus. However, the memory access speed at the time of reading is also reduced.

Further, since the input moving image signal is divided into a plurality of divided areas in one direction in the screen, that is, only in the vertical direction or only in the horizontal direction, the search range of the motion vector is limited in the direction not divided. Disappears.

Here, if the input moving image signal is divided into a plurality of divided areas arranged in the vertical direction, the search range of the motion vector in the vertical direction is limited, but that in the horizontal direction is not limited. MPEG2 (Movingpicture expert group phas) where only the vector search range is limited
It is well suited to encoding methods such as e-2).

In another moving picture coding apparatus according to the present invention,
The reference image storage unit in each of the encoding units includes a locally decoded image signal corresponding to the divided region, and at least a part of the locally decoded image signal corresponding to the divided region adjacent to the encoding target divided region in the screen. Is stored as a first reference image signal, and at least a part of the divided image signals of the adjacent divided regions is stored as a part of a second reference image signal.

In this case, the motion vector detecting means detects a motion vector candidate between the divided image signal of the divided area to be encoded and the second reference image signal stored in the reference image storing means. A motion vector between the divided image signal of the divided area to be encoded and the first reference image signal stored in the reference image storage unit is detected with reference to the motion vector candidate.

With this arrangement, even if the first reference image signal is an image signal of a temporally separated image, for example, a future screen, with respect to the input moving image signal, that is, the divided image signal, the second reference image signal By obtaining the motion vector candidates in order from the input screen closer to the future reference screen using the signal, the motion vector search range can be narrowed, and the amount of calculation for detecting the motion vector is reduced.

Also in this case, among the locally decoded image signals corresponding to the divided region adjacent to the divided region to be encoded by the reference image storage means, the locally decoded image signal corresponding to an area less than half of the adjacent divided region. By storing only the divided image signals corresponding to an area equal to or less than half of the adjacent divided area among the divided image signals of the adjacent divided areas, the number of reference image signals to be written simultaneously is at most two. Thus, the increase in the memory access speed at the time of writing is reduced.

[0024]

Embodiments of the present invention will be described below with reference to the drawings. (First Embodiment) FIG. 1 shows a configuration of a moving picture coding apparatus according to a first embodiment of the present invention. For example, an HDTV image signal is input to the input terminal 11 as an input moving image signal. The input moving image signal is first sent to the screen division unit 12.
Is divided into a plurality of divided areas A, B, C and D arranged in one direction in the screen as shown in FIG. 2 (a) or (b).
The divided image signals 111a, 111a,
111b, 111c and 111d are generated. FIG.
(A) is an example in which divided areas A, B, C, and D are arranged in the vertical direction, that is, an example in which division is divided horizontally, and (b) is an example in which divided areas A, B, C, and D are arranged in the horizontal direction. This is an example in which the division is vertically divided.

The divided image signal 111 from the screen dividing unit 12
a, 111b, 111c, and 111d are the encoding units 13
a, 13b, 13c, and 13d, respectively, and are subjected to compression coding combining motion compensation prediction, DCT, and variable length coding. Encoding units 13a, 13b, 13
encoded data 118 respectively generated in c and 13d
a, 118b, 118c, and 118d are multiplexed by the multiplexing unit 14 and output from the output terminal 15 to a recording system or a transmission system (not shown).

Encoding units 13a, 13b, 13c, 13d
Are basically the same as in a normal moving picture coding apparatus, but each of the divided areas A, B, C,
D corresponding to D (the divided image signal 111
a, 111b, 111c, and 111d in the encoding process, an image signal obtained by local decoding), and the divided areas A, B, C, and D to be encoded in the screen (B for A) , B and A for C, B and D for C, and at least a part of the locally decoded image signal corresponding to D) in the reference image memory as reference image signals, The divided image signals 111a, 111b, 111
c, 111d and a reference image signal read from a reference image memory, a motion vector is detected, and motion compensation prediction is performed using this motion vector. It is different from the encoding device.

FIG. 3 shows the encoding units 13a, 13b, 13
3C is a block diagram illustrating a detailed configuration of an encoding unit 13b that encodes the divided image signal 111b of the divided region B in FIGS. 2A and 2B from among FIGS. Divided image signal 111b
Are temporarily stored in the input image memory 101 and are read out in a predetermined coded screen order. First, motion compensation prediction is performed on the divided image signal 112 read from the input image memory 101.

That is, in the motion compensation predictor 109, the divided image signal 112 from the input image memory 101 and the reference image signal composed of the locally decoded image signal of the already coded frame stored in the reference image memory 108 Are detected, and motion compensation is performed on the reference image signal using the motion vector to generate the predicted image signal 113. In particular,
The divided image signal is read out from the input image memory 101 in small block (partial area) units, and the small image block having the highest correlation with the reference image signal is detected for each small block of the divided image signal, thereby obtaining the divided image signal. Information indicating which small block of the reference image signal has caused each small block of the signal to be moved is detected as a motion vector. However,
In practice, the motion-compensated predictor 109 selects a preferred one of the motion-compensated prediction (inter-frame coding) and the intra-frame coding (predicted image signal = 0) that encodes the divided image signal 112 as it is. Is selected, and a predicted image signal 113 corresponding to the prediction mode is output.

Then, the divided image signal 112 read from the input image memory 101 is also input to the subtractor 102, where the difference between the divided image signal 112 and the predicted image signal 113 is obtained, thereby obtaining a prediction error signal. 114 is generated. The prediction residual signal 114 is subjected to discrete cosine transform (DCT) in the discrete cosine transformer 103 in units of blocks of a fixed size. DCT coefficient data obtained by the discrete cosine transform is quantized by the quantizer 104. The DCT coefficient data quantized by the quantizer 104 is bifurcated, input to the variable length encoder 110 on the one hand, and inversely quantized by the inverse quantizer 105 on the other hand,
Further, an inverse discrete cosine transform (inverse DCT) is performed by an inverse discrete cosine transformer 106.

The output from the inverse discrete cosine transformer 106 is added to the predicted image signal 113 by an adder 107 to form a local decoded image signal 116b, which is stored in the reference image memory 108 as a reference image signal. The reference image signal stored in the reference image memory 108 is read out by the motion compensation predictor 109, and the motion compensation prediction is performed as described above.

The motion vector and prediction mode information generated by the motion compensation predictor 109 are transmitted to the variable length encoder 11.
Input to 0. In the variable-length encoder 110, the quantized DCT coefficient, the motion vector, and the information on the prediction mode are each subjected to variable-length encoding, and the encoded data 118a obtained by this is sent to the multiplexing unit 14 in FIG. Sent out.

Here, FIG.
(A) In addition to the local decoded image signal 116b corresponding to the divided image signal 111b of the divided region B in (b), FIG.
In (b), the locally decoded image signals 116a and 116c corresponding to the divided image signals 111a and 111c of the divided regions A and C adjacent to the divided region B are also input from the encoding units 13a and 13c. 116c is stored as a part of the reference image signal. Other encoding unit 13
The same applies to the reference image memories in a, 13c, and 13d. This is shown in FIG. FIGS. 4A and 4B
(C) and (d) show the divided areas A, B, C, and D in FIG.
Encoding unit 13 when arranged in the vertical direction as shown in FIG.
The area in the screen of the reference image signal stored in the reference image memories in a, 13b, 13c, and 13d is shown.

That is, the reference image memory in the encoding unit 13a for encoding the divided image signal 111a of the divided area A in FIG. 2A has the divided area A as shown by the hatched portion in FIG. The locally decoded image signal corresponding to the whole and the locally decoded image signal corresponding to a part of the divided area B adjacent to the divided area A are stored as reference image signals. FIG.
The reference image memory 108 in the encoding unit 13b that encodes the divided image signal 111b of the divided region B in FIG.
As shown by the hatched portion in FIG. 4B, the local decoded image signal corresponding to the entire divided region B and the local decoded image signals corresponding to a part of each of the divided regions A and C adjacent to the divided region B are different from each other. It is stored as a reference image signal. Similarly, FIG.
The reference image memory in the encoding unit 13c that encodes the divided image signal 111c of the divided area C in FIG.
As shown by the hatched portion in (c), the local decoded image signal corresponding to the entire divided area C and the local decoded image signals corresponding to a part of each of the divided areas B and D adjacent to the divided area C are the reference image. It is stored as a signal. Further, the reference image memory in the encoding unit 13d that encodes the divided image signal 111d of the divided region D in FIG. 2A corresponds to the entire divided region D as indicated by the hatched portion in FIG. And a locally decoded image signal corresponding to a part of the divided area C adjacent to the divided area D are stored as reference image signals.

The storage method of the reference image signal in the reference image memory 108 will be described in more detail. For the divided image signals 111a, 111b, 111c and 111d, the encoding units 13a, 13b, 13c and 13d use the divided areas A and B. , C, and D are coded in order from the top, the area to be stored as the reference image signal is 1/1/2 of the entire divided area and the adjacent divided areas.
Limited to 2 or less.

That is, the divided image signal 11 of the divided area A
The area to be stored as the reference image signal for 1a is shown in FIG.
As shown in the shaded area in FIG. 7A, the division area A is limited to the whole and the upper half of the division area B adjacent to the division area A.
Also, the area to be stored as the reference image signal for the divided image signal 111b of the divided area B is adjacent to the entire divided area B and the divided area B of the divided area A as shown by the hatched portion in FIG. It is limited to the lower half and the upper half of the divided area C adjacent to the divided area B. Similarly, the area to be stored as the reference image signal for the divided image signal 111c of the divided area C is the whole of the divided image C and the lower half of the divided area B adjacent to the divided area C as shown in FIG. Is limited to the upper half of the divided area D adjacent to the divided area C. Further, as shown in FIG. 4D, the area to be stored as the reference image signal for the divided image signal 111d of the divided area D is the whole of the divided area D and the lower half of the divided area C adjacent to the divided area D. Limited.

In this way, the reference image memory 108
In the same timing, only the locally decoded image signal obtained by one or two encoding units may be written as a reference image at the same timing, and the locally decoded image signals obtained by three or more encoding units may be simultaneously written as in the related art. No need to write. This will be described with reference to FIG.

FIG. 4E shows a case where the reference image signal is recorded in the reference image memory 108 in the encoding unit 13b for encoding the divided image signal 111b of the divided area B as shown in FIG. 4B. FIG. 4 is a diagram showing a writing method of FIG. As described above, the encoding unit 13b sequentially encodes the small blocks in the divided areas A, B, and C from the top in the vertical direction.

Here, when the local decoded image signal of the small block at the position indicated by the circle in the divided area B is written into the reference image memory 108, the small block at the upper end in the vertical direction indicated by the circle in the divided area A is simultaneously written. Write the locally decoded image signal. With the progress of encoding, the position of the mark moves downward in the vertical direction, and when writing the locally decoded image signal of the small block at the lower end in the vertical direction in the divided area B to the reference image memory 108,
At the same time, the local decoded image signal of the small block at the lower end in the vertical direction in the divided area A is written. During this time, the local decoded image signal is not written for the divided area C.

When writing the locally decoded image signal of the small block at the upper end in the vertical direction indicated by the mark x in the divided area B to the reference image memory 108, the small picture at the upper end in the vertical direction indicated by the cross mark in the divided area C is simultaneously written. Write the locally decoded image signal of the block. As the encoding progresses, the position of the cross moves downward in the vertical direction, and when the locally decoded image signal of the small block at the lower end in the vertical direction in the divided area B is written in the reference image memory 108, the divided area C is simultaneously written. Write the locally decoded image signal of the small block at the lower end in the vertical direction. During this time, the divided area A
No local decoded image signal is written.

When the reference image signal is written into the reference image memory 108 in this manner, at most two reference image signals are written at the same time. The increase in the memory access speed at the time of writing is significantly reduced as compared with the image encoding device.

Since the reference image memory 108 is provided independently for each of the encoding units 13a, 13b, 13c and 13d, a plurality of reference image memories are provided from a common reference image memory as in a conventional moving image encoding apparatus. It is not necessary to simultaneously read out the reference image signal, and there is no increase in the memory access speed at the time of reading.

Further, according to the present invention, as shown in FIGS. 2A and 2B, an input moving image signal is divided into a plurality of divided regions only in one direction in the screen, that is, only in the vertical direction or only in the horizontal direction. Therefore, the search range of the motion vector is not restricted in the direction in which the motion vector is not divided. Here, in particular, FIG.
When the input moving image signal is divided into a plurality of divided regions arranged in the vertical direction as in (a), the vertical motion vector search range is limited, but the horizontal motion vector search range is not limited. There is an advantage that it is well suited to an encoding method such as MPEG2 (Moving picture expert group phase-2) in which only the search range is limited.

(Second Embodiment) FIG. 5 shows a second embodiment of the present invention.
1 shows a configuration of a video encoding device according to the embodiment. For example, an HDTV image signal is input to the input terminal 21 as an input moving image signal. The input moving image signal is first divided by the screen dividing unit 22 into a plurality of divided areas A, B, C, and D arranged in one direction in the screen as shown in FIG. 2 (a) or (b). The divided image signals 111a, 111b, 111c, 111d corresponding to the divided areas are generated.

The divided image signal 111 from the screen dividing unit 22
a, 111b, 111c, and 111d are the encoding units 23
a, 23b, 23c, and 23d, respectively, and are subjected to compression coding combining motion compensation prediction, DCT, and variable length coding. Encoders 23a, 23b, 23
encoded data 118 respectively generated in c and 23d
a, 118b, 118c, and 118d are multiplexed by the multiplexing unit 24 and output from the output terminal 25 to a recording system or a transmission system (not shown).

Encoding units 23a, 23b, 23c, 23d
, The local decoded image signals (the divided image signals 111a, 111b, and 111b) corresponding to the respective encoding target divided regions A, B, C, and D, as in the first embodiment.
1c, 111d) and a divided area adjacent to the encoding target divided areas A, B, C, and D in the screen (B for A, B for B) A and C, at least a part of the locally decoded image signal corresponding to B and D for C, and C for D) is stored as a first reference image signal in a reference image memory. And at least a part of the divided image signals of the adjacent divided regions (B for A, A and C for B, B and D for C, and C for D) Is stored as a reference image signal.

In this embodiment, the divided image signals 111a, 111a of the divided areas A, B, C, D to be encoded are set.
b, 111c, 111d and a motion vector candidate between the second reference image signal stored in the reference image memory are detected, and the first reference stored in the reference image memory is referred to by referring to the motion vector candidate. A motion vector between the image signal and the image signal is detected, and motion compensation prediction is performed using the motion vector.

FIG. 6 shows the encoding units 23a, 23b, 23
3C is a block diagram illustrating a detailed configuration of an encoding unit 23b that encodes the divided image signal 111b of the divided region B in FIGS. 2A and 2B among c and 23d. In this embodiment, the same parts as those in FIG. 3 are denoted by the same reference numerals. In this embodiment, the divided image signals 111a and 111a of the divided areas A and C adjacent to the encoding target divided area B of the encoding unit 13b are stored in the input image memory 101. 11
1c is newly written as a second reference image signal from the encoding units 13a and 13c. The same applies to the input image memories in the other encoding units 23a, 23c and 23d. Further, it is assumed that the area in the screen stored as the second reference image signal is the same as the reference image signal in the first embodiment (the first reference image signal in the second embodiment).

Further, in this embodiment, a motion estimator 120 having a motion vector candidate detecting function and a motion vector candidate delay memory 121 for delaying the output and supplying the output to the motion compensation estimator 109 are added.

Hereinafter, the divided areas A, B, C, and D are shown in FIG.
As shown in (a), when arranged in the vertical direction,
Each encoding unit 23a, 23b, 23 in the present embodiment
4 (a), (b), (c), and (d) show a method of storing the first and second reference image signals in the reference image memories in c and 23d.
This will be described below.

The divided image signal 1 of the divided area A in FIG.
The reference image memory in the encoding unit 23a that encodes the image data 11a includes the divided area A as shown by the hatched portion in FIG.
Is stored as a first reference image signal, and a local decoded image signal corresponding to a part of the divided area B adjacent to the divided area A is stored as a first reference image signal.
01 includes the entire divided image signal 111a of the divided area A,
Part of the divided image signal 111b of the divided area B adjacent to the divided area A is stored as a second reference image signal. FIG.
The reference image memory 108 in the encoding unit 23b that encodes the divided image signal 111b of the divided region B in FIG.
As shown by the hatched portion in FIG. 4B, the local decoded image signal corresponding to the entire divided region B and the local decoded image signals corresponding to a part of each of the divided regions A and C adjacent to the divided region B are different from each other. The second divided image signal 111b is stored as the first reference image signal, and the divided image signals 111a and 111c of the divided regions A and C adjacent to the divided region B and the divided image signals 111a and 111c of the divided regions A and C adjacent to the divided region B are further stored in the second divided image signal. It is stored as a reference image signal. The divided image signal 111c of the divided area C in FIG.
In the reference image memory in the encoding unit 23c that encodes the divided region C, a local decoded image signal corresponding to the entire divided region C and a divided region B adjacent to the divided region C are stored as shown by hatched portions in FIG. , D are stored as the first reference image signal, and the input image memory 101 stores the entire divided image signal 11 of the divided region C in the input image memory 101.
1c and a part of the divided image signals 111b and 111d of the divided regions B and D adjacent to the divided region C are stored as the second reference image signals. The divided area D in FIG.
In the reference image memory in the encoding unit 23d that encodes the divided image signal 111d, a locally decoded image signal corresponding to the entire divided region D as indicated by a hatched portion in FIG. The local decoded image signal corresponding to a part of the adjacent divided region C is stored as a first reference image signal, and the input image memory 101 is adjacent to the entire divided image signal 111d of the divided region D and the adjacent divided region D. Division area C
Are stored as the second reference image signal.

In this embodiment, when a motion vector is detected, first, each of the encoding units 23a, 23b, 23c, 23
motion predictor 1 having a function of detecting a motion vector candidate in d
20, the divided image signals 111 stored in the input image memory 101 in order from the screen close to the reference screen.
a, 111b, 111c, and 111d are sequentially read out for each small block composed of a predetermined number of pixels, and each of the coding units 2 is similarly read based on a motion vector candidate detected one screen before.
Input image memory 10 in 3a, 23b, 23c, 23d
A plurality of motion vector candidates are detected with reference to the second reference image signals respectively stored in the motion vector candidate memory 1 and input to the motion vector candidate delay memory 121.

Next, each of the encoding units 23a, 23b, 23
The motion-compensated predictors 109 in c and 23d again generate the divided image signals 111a, 111b, 111c and 1 stored in the input image memory 101 again according to the coding screen order.
11d is sequentially read out for each small block composed of a predetermined number of pixels, and a motion vector between the reference image signal and the first reference image signal stored in the reference image memory 108 is detected with reference to the motion vector candidate.

When the motion vector detection is performed in this manner, the input moving image signal, that is, the divided image signals 111a, 111b, 111c, and 111d from the input image memory 101 are obtained.
On the other hand, even if the first reference image signal is a temporally separated image signal of, for example, a future screen, the second reference image signal is used in advance to sequentially input images closer to the future reference screen. By obtaining a motion vector candidate, the motion vector search range can be narrowed. As a result, the amount of calculation for detecting a motion vector can be reduced. The processing after detecting the motion vector in this way is the same as in the first embodiment, and a description thereof will be omitted.

The second reference image signal is also converted to the first reference image signal.
In the same manner as the reference image signal, the region to be stored as the second reference image signal is limited to the whole of each divided region and a half or less of the divided region adjacent thereto.
As described with reference to FIG. 4E, the reference image memory 1
In 08, at the same timing, only the locally decoded image signal obtained by one or two encoding units may be written as a reference image, and the locally decoded image signal obtained by three or more encoding units as in the related art may be written. Since there is no need to write at the same time, an increase in the memory access speed at the time of writing is significantly reduced.

Since the reference image memory 108 is provided independently for each of the encoding units 13a, 13b, 13c and 13d, a plurality of reference image memories are provided from a common reference image memory as in a conventional moving image encoding apparatus. It is not necessary to simultaneously read out the reference image signal, and there is no increase in the memory access speed at the time of reading.

Further, as shown in FIGS. 2A and 2B, since the input moving image signal is divided into a plurality of divided regions only in one direction in the screen, a search for a motion vector is performed in a direction not divided. If the input moving image signal is divided into a plurality of divided regions arranged in the vertical direction as shown in FIG. 2A, the range of the motion vector search in the vertical direction is limited, but that in the horizontal direction is not limited. MPEG2 in which only the vertical motion vector search range is restricted because it is not restricted
There is an advantage that it is well adapted to an encoding method such as

[0057]

As described above, according to the present invention, the locally decoded image signal corresponding to the divided area to be coded and the local decoded image signal corresponding to the divided area adjacent to the divided area to be coded in the screen are provided. By using at least a part of the image signal as the reference image signal, it is possible to detect a motion vector that straddles the boundary of the divided image signal, and to improve coding efficiency or image quality.

The reference image signal to be simultaneously written into the reference image storage means includes at most a locally decoded image signal corresponding to the divided region to be encoded and a divided region adjacent to the divided region to be encoded from both sides in the screen. , And in particular, among the locally decoded image signals corresponding to the divided regions adjacent to the encoding target divided region, the locally decoded image signals corresponding to less than half of the adjacent divided regions. When only image signals are stored, the number of reference image signals to be written at the same time is at most two, so that an increase in memory access speed at the time of writing can be significantly reduced.

On the other hand, since the reference picture storage means is provided independently for each coding means, it is necessary to simultaneously read a plurality of reference picture signals from a shared reference picture memory as in a conventional moving picture coding apparatus. However, the memory access speed at the time of reading can be reduced.

Further, by dividing the input moving image signal into a plurality of divided regions only in one direction in the screen, it is possible to eliminate the limitation of the search range of the motion vector in the direction not divided, thereby achieving higher encoding performance. Is obtained. Here, if the input moving image signal is divided into a plurality of divided regions arranged in the vertical direction, the vertical motion vector search range is limited, but the horizontal motion vector search range is not limited. Is well suited to an encoding system such as MPEG2 in which is restricted.

Further, in the present invention, a locally decoded image signal corresponding to a divided region to be encoded and at least a part of a locally decoded image signal corresponding to a divided region adjacent to the divided region to be encoded in a screen are displayed. Is stored as a first reference image signal, and at least a part of the divided image signals of the adjacent divided regions is stored as a second reference image signal.
A motion vector candidate between the divided image signal of the divided region to be encoded and the second reference image signal is detected, and the divided image signal of the divided region to be encoded and the first The first reference image signal is an image signal of, for example, a future screen that is temporally separated from the divided image signal that is the input moving image signal by detecting a motion vector between the input image signal and the reference image signal However, the motion vector search range can be narrowed by obtaining motion vector candidates in order from the input screen close to the future reference screen using the second reference image signal in advance, and the amount of calculation for motion vector detection is reduced. It becomes possible to reduce.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a moving picture coding apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing two methods of screen division of an input moving image signal according to the present invention.

FIG. 3 is a block diagram showing a detailed configuration of one encoding unit in the embodiment.

FIG. 4 is a diagram showing a region in a screen of a reference image signal stored in a reference image memory in each encoding unit in the embodiment.

FIG. 5 is a block diagram showing a configuration of a moving picture coding apparatus according to a second embodiment of the present invention.

FIG. 6 is a block diagram showing a detailed configuration of one encoding unit in the embodiment.

FIG. 7 is an explanatory diagram of an operation of a conventional moving picture encoding device.

[Explanation of symbols]

 11, 21 ... moving image signal input terminals 12, 22 ... screen division units 13a to 13d, 23a, 23d ... encoding units 14, 24 ... multiplexing units 15, 25 ... encoded data output terminals 101 ... input image memory 102 ... Subtractor 103 Discrete cosine transformer 104 Quantizer 105 Inverse quantizer 106 Inverse discrete cosine transformer 107 Adder 108 Reference image memory 109 Motion compensation predictor 110 Variable length encoder 111a 111d: divided image signal 112: divided image signal 113: predicted image signal 114: prediction error signal 115: DCT coefficient data 116a to 116d: locally decoded image signal 117: motion vector and prediction mode information 118a to 118d: encoded data 120: Motion predictor 121 for detecting motion vector candidates 121... Delay memory

Claims (8)

[Claims] A dividing unit that divides an input moving image signal into a plurality of divided regions arranged in one direction in a screen and generates a divided image signal for each divided region; and a divided image generated by the dividing unit. A plurality of coding means for coding a signal by motion compensation prediction using a motion vector and outputting coded data; and a multiplexing means for multiplexing and outputting coded data respectively output from the plurality of coding means. Each of the encoding means includes: a local decoding means for generating a local decoded image signal; a local decoded image signal corresponding to a division area to be encoded by the encoding means; and a division of the encoding object. Reference image storage means for storing at least a part of a locally decoded image signal corresponding to a divided region adjacent to a region in a screen as a reference image signal, and a divided image signal of the divided region to be encoded; Video encoding apparatus characterized by having a motion vector detecting means for detecting a motion vector between the reference image signal stored in the irradiation image storing means. 2. Dividing means for dividing an input moving image signal into a plurality of divided areas arranged in one direction in a screen, and generating a divided image signal for each divided area; and each divided image generated by the dividing means. A plurality of coding means for coding a signal by motion compensation prediction using a motion vector and outputting coded data; and a multiplexing means for multiplexing and outputting coded data respectively output from the plurality of coding means. Each of the encoding means includes: a local decoding means for generating a local decoded image signal; a local decoded image signal corresponding to a division area to be encoded by the encoding means; and a division of the encoding object. At least a part of the locally decoded image signal corresponding to the divided region adjacent to the region in the screen is stored as a first reference image signal, and at least a part of the divided image signal of the adjacent divided region is stored in a second reference image signal.
Reference image storage means for storing as a reference image signal, and detecting a motion vector candidate between the divided image signal of the divided area to be encoded and the second reference image signal stored in the reference image storage means A motion vector detecting a motion vector between the divided image signal of the divided region to be encoded and the first reference image signal stored in the reference image storage unit, with reference to the motion vector candidate A moving picture coding apparatus comprising: a detection unit. 3. The reference image storage means stores only a locally decoded image signal corresponding to an area equal to or less than half of the adjacent divided area among the locally decoded image signals of the divided area adjacent to the encoding target divided area. 2. The method according to claim 1, wherein the information is stored.
Or the video encoding device according to 2. 4. The apparatus according to claim 1, wherein said reference image storage means stores only a divided image signal corresponding to an area equal to or less than half of the adjacent divided area among the divided image signals of the adjacent divided area. 3. The video encoding device according to item 2. 5. A dividing means for dividing an input moving image signal into a plurality of divided areas arranged in a vertical direction in a screen, and generating a divided image signal for each divided area, and each divided image generated by said dividing means. A plurality of coding means for coding a signal by motion compensation prediction using a motion vector and outputting coded data; and a multiplexing means for multiplexing and outputting coded data respectively output from the plurality of coding means. Each of the encoding means includes: a local decoding means for generating a local decoded image signal; a local decoded image signal corresponding to a division area to be encoded by the encoding means; and a division of the encoding object. A local decoded image signal corresponding to at least one of the divided region adjacent to the region on the upper side of the screen and the divided region adjacent to the encoding target divided region on the lower side of the screen is stored as a reference image signal. Reference image storage means, and a motion vector detection means for detecting a motion vector between the divided image signal of the divided area to be encoded and the reference image signal stored in the reference image storage means. Video encoding device. 6. A dividing means for dividing an input moving image signal into a plurality of divided areas arranged in a vertical direction in a screen, and generating a divided image signal for each divided area, and each divided image generated by said dividing means. A plurality of coding means for coding a signal by motion compensation prediction using a motion vector and outputting coded data; and a multiplexing means for multiplexing and outputting coded data respectively output from the plurality of coding means. Each of the encoding means includes: a local decoding means for generating a local decoded image signal; a local decoded image signal corresponding to a division area to be encoded by the encoding means; and a division of the encoding object. A local decoded image signal corresponding to at least one of the divided region adjacent to the region on the upper side of the screen and the divided region adjacent to the encoding target divided region on the lower side of the screen is recorded as a first reference image signal. A reference image storage unit for storing at least a part of the divided image signal of the adjacent divided region as a second reference image signal; and a divided image signal of the encoding target divided region and the reference image storage unit. And detecting a motion vector candidate between the second reference image signal and the reference image signal stored in the reference image storage unit. And a motion vector detecting means for detecting a motion vector between the first reference image signal and the first reference image signal. 7. The reference image storage means includes at least one of a divided region adjacent to the encoding target divided region on the upper side of the screen and a divided region adjacent to the encoding target divided region on the lower side of the screen. Of the locally decoded image signals corresponding to
6. A local decoding image signal corresponding to at least one of a lower half of the divided area adjacent on the upper side and an upper half of the divided area adjacent on the lower side.
Or the moving picture encoding device according to 6. 8. The reference image storage means, of at least a part of the divided image signals of the adjacent divided areas, includes a lower half of the upper adjacent divided area and an upper half of the lower adjacent divided area. 7. The moving picture coding apparatus according to claim 6, wherein only the divided picture signal corresponding to at least one of the following is stored.