JPH089385A - Dynamic image encoder - Google Patents

Abstract

PURPOSE:To reduce a data transfer amount of an image memory. CONSTITUTION:Two encoding processing parts 200A and 200B in a dynamic image encoder 200 are controlled by control parts in the respective encoding processing parts 200A and 200B, detect motion vectors and parallelly encode the columns of mutually adjacent macro blocks. While present image data and reference image data are required for motion vector detection, image data required between the respective encoding processing parts 200A and 200B in common among the reference image data are simultaneously read from an image memory 250 by the control of access means in the respective encoding processing parts 200A and 200B and written in image data buffers in the respective encoding processing parts 200A and 200B. Thus, the data transfer amount between the dynamic image encoder 200 and the image memory 250 is reduced.

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 transmitting moving pictures and the like.

[0002]

2. Description of the Related Art Conventionally, techniques in such a field include:
For example, some documents were described in the following documents. Reference 1: Optronics (1992-5) Optronics, Okubo “Videophone / Conference Encoding” P.74
-79,86-98 Reference 2; IEEE 1993 CUSTOM CIRCUITS CONFERENCE (19
93), Subroto Bose et al. “A Single Chip Multistand”
ard Video Codec ”P.11.4.1-11.4.4 In moving image coding, as described in Reference 2, one frame image to be coded is encoded, for example, in a macro block of 16 × 16 pixels (hereinafter, MB The following processes (1) to (5) are sequentially performed in small block units called "
Encode. (1) Motion Vector Detection The frame having the largest correlation with the coding target MB of the frame to be coded, that is, the coding frame is detected from the coding reference frame. As the encoded reference frame at this time, generally the frames before and after the encoded frame are used. In the area detection, the vicinity of the same spatial position as the coding target MB in the coding reference frame is set as the search area, and the area having the largest correlation with the coding target MB is obtained in the search area. Then, the difference between the spatial position in the reference frame of the region having the largest correlation and the spatial position of the encoding target MB is detected as a motion vector. (2) Discrete Cosine Transform (hereinafter referred to as DCT) A kind of orthogonal transform is applied to the encoding target MB, and the data of the encoding target MB is converted from spatial data to frequency spectrum data. Is the encoding target MB data of the encoding frame used as it is as the input data of the DCT,
Alternatively, the difference between the encoding target MB and the data of the reference frame of the area referred to by the motion vector is used. As the criterion, the data of the encoded frame is used as it is when the difference value is large, and the difference data is used when the difference value is small.

(3) Quantization The output data of the DCT is converted into a predetermined representative value. (4) Variable-length coding A code having a large length is given to a value having a low occurrence frequency in the quantized output data, and a code having a small length is assigned to a value having a high occurrence frequency for coding. It is output as data. (5) Inverse quantization and inverse discrete cosine transform (hereinafter referred to as IDCT
Inverse quantization performs inverse processing of quantization, and IDCT performs inverse processing of DCT. By these processes, the image data is reproduced as the spatial data and becomes the reference frame data for the frame to be encoded later. Here, since the quantization does not become a 1: 1 conversion, the quantized input and the dequantized output are not equivalent. Therefore, the original image and the reproduced image are different. The motion vector detection of (1) is performed on the luminance component data, and the processes of (2) to (5) are performed on the luminance component and the color component. The luminance component in 1 MB is 16 × 16 = 256 pixels, and the color component is half of 1
It has 28 pixels.

FIG. 2 is a block diagram showing an example of a conventional moving picture coding apparatus. The flow of coding processing for a certain MB data will be described with reference to this drawing. The moving picture coding apparatus 100 includes an image data buffer 101 that stores image data necessary for coding a coded MB, and a motion vector detecting unit 102 and a DCT unit are provided on the output side of the image data buffer 101. 103
Are connected. A control unit 104 is connected to the output side of the motion vector detection unit 102. The output side of the DCT 103 is connected to the IDCT unit 105 and the quantization unit 106, and the output data of the IDCT unit 105 is output to the image data buffer 101. Quantizer 106
The dequantization unit 107 and the variable length coding unit 108 are provided on the output side of
Are connected. The output signal of the inverse quantization unit 107 is also ID
It is configured such that it is input to the CT unit 105 and the output of the variable length encoding unit 108 is output as encoded data DT _out .
The control unit 104 controls the sequence of the encoding process of the moving image encoding apparatus 100, and the control signal s1
It has a function of transmitting 04a, s104b, and s104c. The control signal s104a from the control unit 104 is DC
The T section 103, the IDCT section 105, the quantization section 106, the dequantization section 107, and the variable length coding section 108 are input,
The control signal s104b is input to the image data buffer 101. The control signal s104c is a connection supplied to an external image memory 150 (not shown).
The motion vector detecting unit 102 is the image data buffer 101.
Necessary data for motion vector detection is input from the control unit 104 and the motion vector d102 as a motion vector detection result is input to the control unit 104.
Output to. The DCT 103 inputs the encoding target frame data (current image data) and the reference frame data (reference image data) at the position indicated by the motion vector from the image data buffer 101, and uses the current image data as DCT input data. It is determined whether to use the current image data and the reference image data. The DCT 103 outputs the determination result to the IDCT unit 105 as the intraframe / interframe prediction selection result s103 and performs DCT to perform D
The CT result data d103 is sent to the quantization unit 108.

The quantizing unit 106 inputs the DCT output d103 and quantizes it, and the data d106 converted into a representative value by this quantizing is supplied to the variable length coding unit 108.
The variable length coding unit 108 receives the data d106 from the quantization unit.
, A code corresponding to the occurrence frequency is assigned to the encoded data D
Generate and output T _out . The inverse quantization unit 107 inversely quantizes the quantized data d106, and supplies the inverse quantization result data d107 to the IDCT unit 105. ID
When the CT unit 105 performs IDCT on the data d107 and the intraframe / interframe prediction selection result s103 in the DCT unit 103 selects intraframe prediction, the IDCT unit 105 directly reproduces the IDCT result as the reproduced image data d105. To the image data buffer 101. Further, when the inter-frame prediction is selected in the DCT unit 103, the IDCT unit 105 inputs the reference image data from the image data buffer 101 and outputs the IDC.
Add to T result. The data obtained by adding the reference image data to the IDCT result is output to the image data buffer 101 as reproduced image data d105. The image data buffer stores reproduced image data, and the stored reproduced image data is output as image data to an external image memory. The control unit 104 includes the motion vector detection unit 102 and the DCT unit 1.
03, the IDCT unit 105, the quantization unit 106, the dequantization unit 107, and the variable length coding unit 108, the coding control signal s104a is sent to control the coding processing sequence. The content of the encoding control signal s104a is composed of an address and data, and the section specified by the address carries out the calculation of the encoding process. The data portion of the encoded control signal s104a serves as a parameter given to the specified portion. Further, the control unit 104 sends a buffer control signal s104b for controlling input / output to / from the image data buffer 101 to the image data buffer 101 to control input / output of data between the external image memory 150 and the image data buffer 101. Control signal s1
04c is sent to the external image memory 150.

FIG. 3 is a diagram for explaining the data transfer of the image data buffer in FIG. In the following (i) to (iii),
The data transfer amount between the external image memory 150 and the image data buffer 101 required for 1 MB encoding processing will be described with reference to FIG. (I) Reading Reference Image Data In order to detect the motion vector for the luminance component, it is necessary to read the image data from the image memory 150 for the area used as the search range in the reference frame. For example, when the search range is ± 16 pixels in the horizontal direction and ± 16 pixels in the vertical direction for the MB to be encoded, it is necessary to read image data of 48 × 48 pixels. However, since encoding is usually performed in the order of spatial positions, there is a portion where the MB to be encoded and the MB processed immediately before overlap in the search range. Since the image data of this overlapping portion is already stored in the image data buffer 101, if only the data newly used in the motion vector detection for the encoding target MB is read out, as shown in FIG.
The x48 pixels are read from the external image memory. On the other hand, since the color component is read according to the motion vector detection result, it has 16 × 8 pixels.

(Ii) Current image data 16 × 16 pixels for luminance component and 16 × 8 for color component
Pixels are read out. (iii) Reproduced image data The luminance component is 16 × 16 pixels and the color component is 16 × 8 pixels.
When the above (i) to (iii) are summed up, (16 × 48 + 16 × 8) + (16 × 16 + 16 × 8) + (16 × 16 + 16 × 8) = 1664 pixels. FIG. 4 is a block diagram showing the moving picture coding apparatus and the picture memory of FIG. FIG. 4 shows an image memory 150 connected to the moving picture coding apparatus 100 of FIG. The input data DT _in of the image memory 150 is data obtained by analog / digital conversion of moving image data input in time series.
T _in is written. An image memory control signal S104c is output from the control unit 104 in the moving image encoding apparatus 100,
Based on the image memory control signal s104c, the image data d150 is the moving image encoding device 100 and the image memory 150.
Transferred to and from. The content of the image data d150 is composed of reference image data, current image data, and reproduced image data.

[0008]

However, the conventional moving picture coding apparatus has the following problems. When the size of the processed image becomes large, the moving picture coding apparatus 100
Processing capacity, moving picture coding apparatus 100, and image memory 15
There was a problem that the data transfer rate between 0s was limited. For example, in the standard video standard, the size of one frame is 720 × 480 pixels, and the frame frequency is 30 Hz.
Is. The number of MBs in one frame is 720 × 480 / (16 × 16) = 1350MB, and the maximum allowable processing time for 1MB is 1 / 30sec / 1350 = 24.6μs, which is a very short time. Therefore, a large amount of processing may occur and one moving image coding device may not satisfy the performance. Further, the cycle per pixel in data transfer between the moving picture coding apparatus 100 and the image memory 150 is 24.6 μs / 1664 pixels = 14.78 ns. Even if four pixels are transferred at the same time in order to improve the transfer capability per pixel, it is necessary to transfer at 59.1 ns cycle. Considering the transfer overhead of about 25%, it is necessary to input / output data at a cycle of 47.3 ns, that is, 21.1 MHz. When either the coding processing performance or the image transfer performance cannot be satisfied in the moving picture coding device, a system in which the coding calculation amount and the image data transfer amount are appropriately adjusted by dividing and processing the image data, It is described in Reference 2.

FIG. 5 is a block diagram showing a moving picture coding apparatus for dividing image data. This image encoding device includes, for example, two encoding processing units 100A and 100b having the same configuration as the moving image encoding device shown in FIG. Image memories 150A and 150B are connected to the encoding processing units 100A and 100B, respectively. Two
In each of the image memories 150A and 150B, 720 × 24
The image data DT _{in of} 0 pixel is written respectively. Each encoding processing means 100A, 100B and image memory 150
The data d150A and d150B are respectively transferred between A and 150B, and the respective coding processing units 100A and 100B are transferred.
Perform encoding processing respectively. FIG. 6 shows M in the frame.
It is a figure which shows B. FIG. 6 shows A0 to G4 of a plurality of MBs that are part of one frame. For example, the image data DT _in corresponding to A0 to D4 of MB is stored _{in the} image memory 1
50A, and the encoding processing unit 100A writes A0 of MB.
~ D4, that is, the encoding corresponding to the upper MB of the frame is performed. Similarly, the image data DT _in corresponding to E0 to G4 of MB is written _{in the} image memory 150B, and the encoding processing unit 100B performs encoding corresponding to E0 to G4 of MB, that is, the MB above the frame. . According to such encoding processing, since the load on the moving image encoding apparatus is divided, it is possible to avoid concentration of the processing amount.
However, in the MB that is the boundary between the upper and lower parts of the frame, the motion vector detection range cannot be sufficiently secured, the motion vector detection performance deteriorates, and the coding efficiency decreases. For example, in the motion vector detection for the upper MB D2, the data corresponding to the lower E0 to E4 MBs could not be referred to.

[0010]

In order to solve the above problems, the present invention shares an image memory for storing image data, and an image of a block of a predetermined size in a current image stored in the image memory. Moving picture coding for carrying out coding on a moving picture, comprising a plurality of coding processing units for carrying out motion vector detection using data and image data of a block of a reference picture and coding the current picture in block units The device has the following configuration. That is, the control means for causing the block strings in the current image adjacent to each other to correspond to the respective encoding processing units and performing the encoding in parallel by the respective encoding processing units, and the control unit used in the motion vector detection An access for simultaneously reading, from the image memory, the image data commonly used by the respective encoding processing units among the image data of the block of the reference image to all the encoding processing units commonly using the image data. And means are provided.

[0011]

According to the present invention, since the moving picture coding apparatus is constructed as described above, each coding processing section, under the control of each control means, parallelizes the block strings in the current picture adjacent to each other. Encode. At this time, the image data of the block of the reference image that is commonly used by the encoding processing units is simultaneously read from the image memory by the access unit to all the encoding processing units that use it. Each coding processing unit further reads, from the image memory, the image data of the block of the reference image that is not commonly used by the coding processing units and the image data of the block of the current image, and detects the motion vector. Encoding in block units. Therefore, the above problem can be solved.

[0012]

1 is a block diagram of a moving picture coding apparatus according to an embodiment of the present invention. This moving picture coding device 200
Includes two encoding processing units 200A and 200B configured by using a CPU (central processing unit) and the like, and these encoding processing units 200A and 200B store image data DT _in input in time series. The image memory 250 is connected. The encoding processing units 200A and 200B perform motion vector detection in units of blocks of the current image and perform encoding, respectively.
Data is transferred between B and the image memory 250, and the encoded data DT _out is output from the output side of the encoding processing units 200A and 200B. Further, the encoding processing units 200A and 200B share the image memory 250, and the image memories 2 of the encoding processing units 200A and 200B are shared.
Access to 50 is controlled by the synchronization signal s200. FIG. 7 is a configuration block diagram showing the inside of the encoding processing unit in FIG. Encoding processing units 200A, 2
00B has the same configuration and includes an image data buffer 201. The image data buffer 201 is connected to the image memory 250, and each encoding processing unit 20.
It has a function of storing image data necessary for encoding in 0A and 200B. On the output side of the image data buffer 201, a motion vector detection unit 202 and a DCT unit 203
Are connected. A control unit 204, which is a control unit, is connected to the output side of the motion vector detection unit 202. D
The IDCT unit 205 and the quantization unit 2 are provided on the output side of the CT 203.
06 is connected, and the output data of the IDCT unit 205 is output to the image data buffer 201. An inverse quantizer 207 and a variable length encoder 208 are connected to the output side of the quantizer 206. The output data of the inverse quantization unit 207 is also input to the IDCT unit 205, and the output of the variable length encoding unit 208 is output as encoded data DT _out . The control unit 204 controls the sequence of the encoding process of the encoding processing unit, and has a function of transmitting the control signals s204a, s204b, s204c. Control signal s204a from the control unit 104
Are the DCT unit 203, the IDCT unit 205, and the quantization unit 20.
6, the dequantization unit 207, and the variable length coding unit 208, and the control signal s204b is input to the image data buffer 20.
1 is input. The control signal s104c is a connection supplied to the image memory 250. A synchronization control unit 300 is provided in each of the encoding processing units 200A and 200B, and the synchronization control unit 300 is connected to the control unit 204. The synchronization controller 220 constitutes access means,
The image memory 2 of each of the encoding processing units 200A and 200B according to the synchronization control signal s200 transmitted from the synchronization control unit 220.
This is a configuration in which access to 50 is controlled.

Next, the operation of the moving picture coding apparatus shown in FIG. 1 will be described with reference to the drawings. Encoding processing units 200A and 200B
Share the image memory 250, access by time division, and carry out encoding respectively. The synchronization control signal s200 determines which encoding processing unit inputs / outputs data to / from the image memory 250. The motion vector detection unit 202 inputs the data necessary for motion vector detection from the image data buffer 201, and outputs the motion vector data d202 of the motion vector detection result to the control unit 204. DCT2
03 receives the current image data and the reference image data at the position indicated by the motion vector from the image data buffer 201, and determines whether to use the current image data as the DCT input data or the current image data and the reference image data. To do. The DCT 203 uses the determination result as the intraframe / interframe prediction selection result s203.
5 and performs DCT and sends the DCT result data d203 to the quantizer 206. Quantizer 206
Receives the DCT output d203 and quantizes it, and the data d206 converted into a representative value by this quantization is supplied to the variable length coding unit 208. The variable length coding unit 208 allocates a code corresponding to the occurrence frequency of the data d206 from the quantizing unit, generates coded data DT _out , and outputs it. The inverse quantization unit 207 inversely quantizes the quantized data d206, and supplies the inverse quantization result data d207 to the IDCT unit 205. In the IDCT unit 205, the data d20
7 is subjected to IDCT, and the intra-frame / inter-frame prediction selection result s203 in the DCT unit 203 selects intra-frame prediction, the IDCT unit 205 determines that the ID
The CT result is directly output to the image data buffer 201 as the reproduced image data d205. In addition, the DCT unit 20
If inter-frame prediction is selected in 3, ID
The CT unit 205 inputs the reference image data from the image data buffer 201 and adds it to the IDCT result. The data obtained by adding the reference image data to the IDCT result is output to the image data buffer 201 as reproduced image data d205. The image data buffer stores reproduced image data, and the stored reproduced image data is output to the image memory 250 as image data d250.

The control unit 204 is a motion vector detecting unit 202.
, DCT section 203, IDCT section 205, quantization section 206, inverse quantization section 207, and variable length coding section 20.
8, the coding control signal s204a is sent out to control the sequence of the coding process. Encoding control signal s204a
The content of is composed of an address and data, and the part specified by the address carries out the calculation of the encoding process. The data of the encoded control signal s204a becomes a parameter to be given to the specified part. In addition, the control unit 204
Is a buffer control signal s104b for controlling input / output to / from the image data buffer 201.
The image memory control signal s204c for controlling input / output of data between the external image memory 250 and the image data buffer 201 is transmitted to the external image memory 250. Each of the encoding processing units 200A and 200B selects a block to be processed according to the image memory control signal s204c. Each encoding processing unit 200A, 200B of this embodiment
Encodes adjacent block sequences in the current image in parallel. For example, the encoding processing unit 200A is shown in FIG.
In A to A4, C0 to C4, and E0 to E4 are encoded, and the encoding processing unit 200B sets B0 to B0.
Coding is performed on MBs of B4, D0 to D4, and F0 to F4. Each A0-A4 of MB, each B0-B4, each C0-C
4 and each of D0 to D4, and each of E0 to E4 and each of F0 to F4 are encoded at the same time while being slightly shifted in timing.

FIG. 8 is a diagram for explaining the data transfer of the encoding processing unit of FIG. This figure shows MB C1 and D1
3 shows a sequence of encoding of. Each encoding processing unit 2
Of the processing time in 00A and 200B, about 1/3 is used for motion vector detection, and about 2/3 is used for DCT and quantization. If the reference data of the luminance component for C1 is ± 16 pixels in both the horizontal and vertical directions, the MB of the required reference image is B0, B1, B2, C0,
C1, C2, D0, D1, and D2, which are B0, B1, C0, C1, and D in the encoding process of C0 performed previously.
Since 0 and D1 are used, they are already stored in the image data buffer 201. Therefore, the newly required M
The image data of B is the image data of B2, C2, D2. Similarly, the image data newly required for D1 is
It is the image data of MB of C2, D2, and E2. here,
The image data of C2 and D2 are encoded by the respective encoding processing units 200A and 2A.
00B, it is necessary to detect both motion vectors, and thus is read from the image memory 250 and written to the image data buffers 201 of both encoding processing units 200A and 200B at the same time.

The reading of the luminance data of the reference image starts at time t _{0 in} FIG. 8, and the reading of the current image data is subsequently performed. When reading the current image data, the current image data of C1 is read first, and the current image data of D1 is read later. After the reading of the current image data of C1 is completed, the encoding processing unit 200A starts the motion vector detection for C1. After the reading of the current image data of D1 is completed, the encoding processing unit 200B starts the motion vector detection for D1.
In addition, after the reading of the current image data of D1 is completed, C already calculated and stored in each image data buffer 201 is displayed.
The reproduced image data of 0 and D0 are written in the image memory 250 in order. At the end of writing the reproduced image data of D0, C
Since the motion vector detection for 1 has been completed, the reference region data C1c of the color component indicated by the detected motion vector is written in the image data buffer 201 of the encoding processing unit 200A. Subsequently, the data D1c of the color component reference area for D1 is read and the encoding processing unit 20
It is written in the image data buffer 201 of 0B. The image memory 2 is processed by the encoding process for C1 and D1 of two MBs.
The data transfer amount between 50 and each image data buffer 201 is 281 between time t1 and time t2 shown in FIG.
It becomes 6 pixel data, and it is 14 when converted per 1MB.
It becomes the data of 08 pixels. This value is reduced to 11/13 as compared with the conventional data transfer amount of 1664 pixels, which means that the processing time can be shortened.

FIG. 9 is a diagram showing a change in access right between the encoding processing units in FIG. 1, and using this figure, each encoding processing unit 200A at time t1 to time t2 in FIG.
Access to the image memory 250 of 200B will be described. One of the encoding processing units 200A and 200B is a master and the other is a slave. When the number of encoding processing units is three or more, one encoding processing unit becomes a master and all the other encoding processing units become slaves. Here, the encoding processing unit 200A is a master and the encoding processing unit 200B is a slave. When the data required in common to both the encoding processing units 200A and 200 is read from the image memory 250,
The image memory control signal s204c is output from the encoding processing unit 200A on the master side and is not output from the slave side. Which of the encoding processing units is the master or the slave is set by giving the synchronization control interface signal sI from the control unit 204 to the synchronization control 220. Time t1
At, the encoding processing unit 200A abandons the access right to the image memory 250. The synchronization control interface signal sI is supplied from the control unit 204 in the encoding processing unit 200A to the synchronization control unit 220 of the encoding processing unit 200A,
A synchronization control signal s200 is sent from the synchronization control unit 220 to the synchronization control unit 220 in the encoding processing unit 200B.
The synchronization control interface signal sI is input from the synchronization control unit 220 in the encoding processing unit 200B to the control unit 204 in the encoding processing unit 200B, and it is recognized that the control unit 204 has obtained the access right to the image memory 250. .
The control unit 204 outputs the image memory control signal S204c for reading the current image data of D1 which is MB to the image memory 250, and the corresponding data is read from the image memory 250. After the reading of the current image data of D1 is completed, the control unit 204 of the code processing unit 200B abandons the access right to the image memory 250. That is, the code processing unit 200
The B control unit 204 supplies the synchronization control interface signal sI to the synchronization control unit 220 of the code processing unit 200B. The synchronization control signal s200 is transmitted from the synchronization control unit 220 of the code processing unit 200B to the synchronization control unit 220 of the encoding processing unit 200A.

In the same procedure as described below, the encoding processing unit 200A
Writing of reproduced image data of C0 by the encoding processing unit 2
00B for writing the reproduced image data of D0, the encoding processing unit 200A for reading the C1 color component C1c reference image data, and the encoding processing unit 200B for the D1 color component D1.
c Reference image data reading and reference image data reading of the luminance component of C3 by the encoding processing unit 200A are performed. The luminance component which is the reference image data of C3 of MB is read from the image memory 250, and at the same time, each encoding processing unit 20
It is written in the image data buffers 201 of 0A and 200B, respectively. Here, the procedure for writing the reference image data of C3 will be described. The control unit 204 of the encoding processing unit 200A, which is the master, uses the common data access request signal as the synchronization control interface sI.
To the synchronization control unit 220. This synchronization control unit 220
Sends a synchronization control signal s200 to the synchronization control unit 220 of the encoding processing unit 200B. Synchronization control signal s2
00, the image memory 25 is delayed by a predetermined cycle.
The reading from 0 is started. The number of cycles of the predetermined cycle is determined by the two encoding processing units 200A and 20A.
It is preset between 0B. The control unit 204 of the code processing unit 200B inputs the common data access signal via the synchronization control unit 220. The data transfer to the image data buffer 201 is started at the same time when the predetermined cycle elapses. Control unit 204 of encoding processing unit 200A that is a master
Sends the image memory control signal s204c to the image memory 250.
And the buffer control signal s204b to the image data buffer 201. On the other hand, the control unit 204 of the encoding processing unit 200B controls the buffer control signal s204b.
Is supplied to the image data buffer 201 only. After the reading of the luminance component that is the reference image data of C3 of MB is completed, the luminance component of D3 is read as reference image data in the same procedure. Subsequently, the encoding processing unit 200B
E3 reference image data of MB, encoding processing unit 20
Each reading of the current image data of C2 of MB by 0A is performed.

As described above, in this embodiment, the two encoding processing units 200A and 200 sharing the image memory 250 are shared.
Since the moving image coding apparatus is configured by B, the processing amount is not concentrated and the load on the coding processing unit is reduced. In addition, the encoding processing units 200A and 200B respectively encode the columns of MBs adjacent to each other in the current image in parallel. Therefore, the area of the reference image is not limited and the data of the common reference image can be used, and the amount of data transfer between the image memory 250 and the moving image encoder 200 can be prevented while preventing the deterioration of the image quality. Can be reduced. The present invention is not limited to the above embodiment, and various modifications can be made. For example, the number of encoding processing units 200A and 200B may not be two, and may be three or more.
In that case, the load on each encoding processing unit is further reduced. The size of the MB can be changed according to the purpose of the image processing, and the same effect as that of the above-described embodiment can be obtained regardless of the size of the MB.

[0020]

As described in detail above, according to the present invention, a plurality of coding processing units sharing an image memory are provided with a moving picture coding device and a control means, and each of them is controlled by the control means. The encoding processing unit is configured to encode adjacent block strings in parallel. Further, the moving picture coding apparatus is provided with an access means, and the data of the reference image commonly used by the respective coding processing sections by the access means is simultaneously read from the image memory to each coding processing section using the reference picture data. To be done. Therefore, it is possible to use common reference image data without limiting the reference image area, and it is possible to reduce the amount of data transfer between the image memory and the moving image encoding device.

[Brief description of drawings]

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

FIG. 2 is a configuration block diagram showing an example of a conventional moving image encoding device.

FIG. 3 is a diagram illustrating data transfer of an image data buffer in FIG.

FIG. 4 is a block diagram showing a moving picture coding apparatus and an image memory of FIG.

[Fig. 5] Fig. 5 is a block diagram illustrating a moving image encoding device that divides image data.

FIG. 6 is a diagram showing an MB in a frame.

7 is a configuration block diagram showing the inside of the encoding processing unit in FIG. 1. FIG.

FIG. 8 is a diagram illustrating data transfer of the encoding processing unit in FIG. 1.

9 is a diagram showing a change in access right between the encoding processing units in FIG.

[Explanation of symbols]

200 moving picture coding apparatus 200A, 200B coding processing section 250 image memory s200 synchronization control signal s204c image memory control signal s250 transfer data DT _out coded data 201 image data buffer 202 motion vector detection section 203 DCT section 204 control section 220 synchronization Control unit s204b Buffer control signal sI Synchronous control interface signal

Claims (1)

[Claims] 1. A motion vector is detected by sharing an image memory for storing image data and using image data of a block of a predetermined size and image data of a block of a reference image in the current image stored in the image memory. In the moving picture coding apparatus, which includes a plurality of coding processing units for coding the current picture in block units, and which performs coding for the moving picture, Control means corresponding to each of the encoding processing units and causing the encoding processing units to perform the encoding in parallel, and each encoding processing unit of the image data of the block of the reference image used in the motion vector detection. Access means for simultaneously reading out image data commonly used between all the encoding processing units that commonly use the image data from the image memory, Moving image encoding process and wherein the digit.