JP2011139175A - Image processor and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce a memory bandwidth in an image processor. <P>SOLUTION: The image processor includes a compression processing unit 1 which performs compression processing for input image data. In the compression processing unit 1, a variable length encoding unit 12 reads image data of a small area from an image storing unit 11, and encodes and compresses the image data for each small area by a variable length encoding method. A data construction unit 13 divides the compressed data into first data of a predetermined data size or below and second data exceeding the predetermined data size. A data construction unit 13 collects a plurality of small areas of the second data to create third data. The data construction unit 13 creates first block data based on the first data, second block data based on the third data, and size information on the third data. The first block data are written in a first block of a fixed size corresponding to a memory. The second block data are written in a second block corresponding to the memory. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing method.

  In the process of encoding and compressing moving image data, a process called motion compensation may be performed. Motion compensation is a technique for compressing data by predicting motion information of a compression target block from a frame that is close in time. When decoding the encoded data, the reference image data of the prediction source is read and used. At this time, the order and the size of the image for reading the reference image data may not be fixed. In that case, the memory controller used in the moving image encoding / decoding device is required to have random accessibility to the reference image. On the other hand, in recent years, with the spread of high-definition broadcasting and the like, the memory bandwidth used by the moving image processing circuit has increased.

  2. Description of the Related Art Conventionally, there is an image processing apparatus that performs image processing by storing image data in a memory in the form of variable-length compression codes. For example, there is an image processing apparatus configured to divide an input image into small areas and compress and encode image data for each small area. In this image processing apparatus, the memory is divided into a fixed area where the data storage position and capacity are fixed and a free area where the data is not fixed. Of the compressed image data, data corresponding to the capacity of the fixed area is stored in the fixed area. Data exceeding the capacity of the fixed area is stored in the free area. In the fixed area, address information indicating the storage position of each data stored in the free area in the free area is stored (for example, refer to Patent Document 1).

  In addition, the image is input from the image input device, the data is compressed, the image is stored in a storage medium, the data is compressed and stored as necessary, and the image is output from the image output device. There is an image signal encoding device. In this image signal encoding apparatus, fixed length compression processing is first performed by a fixed length compression / decompression circuit and written to a buffer memory, and then variable length compression processing is performed by a variable length compression / decompression circuit and then a hard disk or the like is transmitted via the buffer memory. There is a device that stores the information in a storage medium (for example, see Patent Document 2). In addition, a first encoding unit that generates lossy encoded data, a second encoding unit that generates lossless encoded data, and an attribute determination unit that detects the number of colors in a pixel block to be encoded are arranged in parallel. Thus, there is an image encoding device that performs processing on the same pixel block. In such an image encoding device, the encoding sequence control unit stores lossless encoded data in the first memory when the number of colors of the pixel block of interest is equal to or less than a predetermined number. When the number of colors exceeds a predetermined number, the shorter one of the lossy encoded data and the lossless encoded data is stored in the first memory (see, for example, Patent Document 3).

JP 2000-232587 A Japanese Unexamined Patent Publication No. 7-226848 JP 2006-166420 A

  However, the conventional image processing apparatus has the following problems when reading data of a plurality of fixed areas from the memory at once. Based on the address information stored in a plurality of fixed areas to be read, a plurality of read requests are issued to the free area. If the division unit of the small area is reduced and the random accessibility is increased, the number of times of reading data from the free area is increased accordingly, and the time until the data reading is completed is increased. Also, a read request is issued to the free area based on the address information read from the fixed area. That is, the data is read from the free area after the data is read from the fixed area. Therefore, the time until data reading is completed becomes long. Therefore, there is a problem that the memory bandwidth increases.

  An object of the present invention is to provide an image processing apparatus and an image processing method capable of reducing the memory bandwidth.

  The image processing apparatus includes a variable length coding unit and a data construction unit. In the image processing apparatus, an input image is divided into a plurality of small regions and processed. The variable length encoding unit encodes and compresses the image data for each small region by a variable length encoding method. The data construction unit divides the compressed data encoded by the variable length encoding unit into first data and second data. The first data is a data portion having a predetermined data size or less in the compressed data encoded by the variable length encoding unit. The second data is a data portion exceeding the predetermined data size in the compressed data encoded by the variable length encoding unit. The data construction unit collects the second data for the plurality of small areas as third data. The data construction unit generates first block data based on the first data. The first block data is written into a first block corresponding to the small area among a plurality of first blocks having a fixed memory size equal to or larger than the predetermined data size of the memory. The data construction unit generates second block data based on the third data. The second block data is written into the second block corresponding to the small area group in which the second data is collected, among the plurality of second blocks in the memory. The data construction unit generates size information of the third data.

  According to the image processing apparatus and the image processing method, the memory bandwidth can be reduced while maintaining random accessibility.

1 is a block diagram illustrating an image processing apparatus according to a first embodiment. FIG. 6 is a block diagram illustrating an image processing apparatus according to a second embodiment. FIG. 10 is a block diagram illustrating a developing unit of the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating an example of a variable-length encoding process according to the second embodiment. 10 is a flowchart illustrating an example of a variable-length encoding procedure according to the second embodiment. FIG. 10 is an explanatory diagram schematically illustrating a structure of data obtained by a variable length encoding process according to the second embodiment. FIG. 9 is an explanatory diagram illustrating a structure of output data of a variable length coding unit according to the second embodiment. FIG. 12 is an explanatory diagram illustrating an example of a memory storage structure of a D1 block by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating an example of a memory storage structure of a D2 block by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating an example of a D2 block by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating an example of a residual data field in each column by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating an example of a memory storage structure of a D2 table by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram of an example of a D2 table entry field by the image processing apparatus according to the second embodiment. 10 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the second embodiment. 10 is a flowchart illustrating an example of a data read procedure by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating a memory bandwidth reduction effect by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating a memory bandwidth reduction effect by the image processing apparatus according to the second embodiment. FIG. 10 is an explanatory diagram illustrating a data storage method in the image processing apparatus according to the third embodiment. FIG. 19 is a chart showing an IsValid flag corresponding to FIG. 18. FIG. FIG. 19 is a chart showing an IsNeighbor flag corresponding to FIG. 18. FIG. 12 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the third embodiment. 12 is a flowchart illustrating an example of a data read procedure by the image processing apparatus according to the third embodiment. FIG. 10 is an explanatory diagram illustrating a data storage method in the image processing apparatus according to the fourth embodiment. FIG. 24 is a chart showing an IsNeighbor flag corresponding to FIG. 23. FIG. 14 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the fourth embodiment. 10 is a flowchart illustrating an example of a data read procedure by the image processing apparatus according to the fourth embodiment. FIG. 10 is an explanatory diagram illustrating a memory bandwidth reduction effect by the image processing apparatus according to the fourth embodiment. 22 is a chart illustrating an example of an IsNeighbor flag in the fifth embodiment. 14 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the fifth embodiment. 14 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the fifth embodiment. 14 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the fifth embodiment. 22 is a chart showing an example of an IsNeighbor flag in Example 6. 16 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the sixth embodiment. 16 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the sixth embodiment. 16 is a flowchart illustrating an example of a data writing procedure by the image processing apparatus according to the sixth embodiment.

  Exemplary embodiments of an image processing apparatus and an image processing method will be described below in detail with reference to the accompanying drawings. In the following description of each embodiment, the same components are denoted by the same reference numerals, and redundant descriptions are omitted.

(Example 1)
In the image processing apparatus and the image processing method according to the first embodiment, an input image is divided into a plurality of small areas, and first data having a predetermined size or less and second remaining data from the data compressed for each small area , And the second data is grouped into a plurality of small areas as third data, and size information of the third data is generated.

FIG. 1 is a block diagram of an image processing apparatus according to the first embodiment. As illustrated in FIG. 1, the image processing apparatus includes a compression processing unit 1 that performs compression processing on input image data. The compression processing unit 1 includes an image storage unit 11, a variable length encoding unit 12, a data construction unit 13, and a compressed data storage unit 14. The image storage unit 11 stores input image data and divides it into a plurality of small area image data. The input image data is delivered from a request control unit (not shown), for example. The variable length encoding unit 12 reads out the image data of the small area from the image storage unit 11, and encodes and compresses the image data for each small area by the variable length encoding method.

  The data construction unit 13 divides the data compressed by the variable length coding unit 12 into first data and second data. The first data is a data portion having a data size equal to or smaller than a predetermined data size in the data compressed by the variable length encoding unit 12. The second data is a data portion of the data compressed by the variable length encoding unit 12 that exceeds a predetermined data size. When the data size of the data compressed by the variable length encoding unit 12 is equal to or smaller than a predetermined data size, only the first data is generated. The data construction unit 13 collects the second data for a plurality of small areas as third data.

  The data construction unit 13 generates first block data based on the first data. The first block data is written in a first block corresponding to the small area among a plurality of first blocks of a memory connected to an external memory access control unit (not shown), for example. The first block has a fixed memory size equal to or larger than the predetermined data size. The data construction unit 13 generates second block data based on the third data. The second block data is written into the second block corresponding to the small area group in which the second data is collected, among the plurality of second blocks in the memory. The data construction unit 13 generates size information of the third data.

  Hereinafter, the first data is called D1 data, the second data is called residual data, and the third data is called aggregated residual data. The first block data is called D1 block data, the second block data is called D2 block data, and the size information of the third data is called D2 size information. The compressed data storage unit 14 stores D1 block data, D2 block data, and D2 size information. The D1 block data, D2 block data, and D2 size information stored in the compressed data storage unit 14 are transferred to, for example, an external memory access control unit (not shown). In a memory (not shown), the first block is called a D1 block, and the second block is called a D2 block.

Image Processing Procedure The compression processing unit 1 stores, for example, input image data passed from a request control unit (not shown) in the image storage unit 11 and divides it into small region image data. Next, the compression processing unit 1 reads out the small area image data from the image storage unit 11 by the variable length encoding unit 12, encodes the image data by the variable length encoding method, and compresses the image data. Next, the compression processing unit 1 generates D1 data, residual data, aggregated residual data, D1 block data, D2 block data, and D2 size information from the data compressed by the variable length encoding unit 12 by the data construction unit 13 To do. Next, the compression processing unit 1 stores D1 block data, D2 block data, and D2 size information in the compressed data storage unit 14. The compression processing unit 1 outputs D1 block data, D2 block data, and D2 size information from the compressed data storage unit 14 to, for example, an external memory access control unit (not shown).

  According to the first embodiment, the variable length coding unit 12 and the data construction unit 13 generate D1 block data, D2 block data, and D2 size information. When reading data from a memory storing D1 block data, D2 block data, and D2 size information, it is possible to determine whether it is necessary to read data of the D2 block by checking the D2 size information. it can. As a result, when it is determined that there is no need to read the data of the D2 block, it is not necessary to read the data of the D1 block from the memory and read the data of the D2 block, so that the memory bandwidth can be reduced. If it is determined that it is necessary to read the data of the D2 block, the remaining data of a plurality of small areas are collected in the D2 block, so that the remaining data that has not been collected is read compared to reading one by one. The number of requests issued is reduced. Accordingly, the number of data read times is reduced, and the memory bandwidth can be reduced. In addition, since the process of reading the data of the D1 block and the process of reading the data of the D2 block can be performed in parallel, the time until the data reading is completed can be shortened. Therefore, the memory bandwidth can be reduced.

(Example 2)
The image processing apparatus and the image processing method according to the second embodiment are an image processing apparatus including the compression processing unit 1 according to the first embodiment and an image processing method that compresses image data according to the image processing procedure according to the first embodiment.

FIG. 2 is a block diagram of an image processing apparatus according to the second embodiment. As shown in FIG. 2, the image processing apparatus includes an encoder 2 and a memory controller 3. The encoder 2 outputs uncompressed rectangular image data (rectangular write data) from the decoded image output unit 4 to the memory controller 3. The encoder 2 receives uncompressed rectangular image data (rectangular read data) from the memory controller 3 through the reference image reading unit 5. Here, rectangular write data and rectangular read data are used, but shapes other than rectangular may be used.

  The memory controller 3 includes a request control unit 6, a compression processing unit 1 according to the first embodiment, a decompression unit 7, and an external memory access control unit 8. The request control unit 6 issues a write request and passes the rectangular write data received from the decoded image output unit 4 to the compression processing unit 1. The request control unit 6 issues a read request and passes the rectangular read data received from the expansion unit 7 to the reference image reading unit 5. The compression processing unit 1 is as described in the first embodiment. The rectangular write data is stored in the image storage unit 11 (see FIG. 1) of the compression processing unit 1. The compressed data stored in the compressed data storage unit 14 (see FIG. 1) of the compression processing unit 1 is passed to the external memory access control unit 8.

  The external memory access control unit 8 controls access to the external memory 9 connected to the memory controller 3, and exchanges compressed data between the memory controller 3 and the external memory 9. The expansion unit 7 expands the compressed data into uncompressed image data (rectangular read data) based on the D1 block data, the D2 block data, and the D2 table data of the compressed data read from the external memory 9. To do. The D2 table is a table storing D2 size information in the external memory 9. Data compressed by the compression processing unit 1 is written in the external memory 9. The external memory 9 is provided with a first area for storing D1 block data, a second area for storing D2 block data, and a D2 table. The external memory 9 includes, for example, SDRAM (Synchronous Dynamic Random Access Memory, synchronous DRAM), although not limited thereto.

FIG. 3 is a block diagram of the developing unit of the image processing apparatus according to the second embodiment. As shown in FIG. 3, the expansion unit 7 includes a D1 reading unit 21, a D2 table cache control unit 22, a D2 read determination unit 23, a D2 block cache control unit 24, a residual data acquisition unit 25, a variable length decoding unit 26, an image A storage unit 27, a table cache 28, and a D2 cache 29 are provided. The D1 reading unit 21 reads, from the external memory 9, data of a D1 block corresponding to, for example, a rectangular image area designated as a reading target. The table cache 28 is a storage unit that stores a part of the data of the D2 table. The D2 cache 29 is a storage unit that stores data of some D2 blocks.

  The D2 table cache control unit 22 determines whether or not the data (D2 size information) of the D2 table corresponding to, for example, a rectangular image area designated as a read target is stored in the table cache 28. When the desired D2 size information is stored in the table cache 28 (cache hit), the D2 table cache control unit 22 reads the desired D2 size information from the table cache 28. When the desired D2 size information is not stored in the table cache 28 (cache miss), the D2 table cache control unit 22 reads the desired D2 size information from the external memory 9 and stores it in the table cache 28.

  Whether the D2 read determination unit 23 needs to read data of a D2 block corresponding to, for example, a rectangular image area designated as a read target based on the D2 size information passed from the D2 table cache control unit 22 Determine whether or not. When the D2 size information indicates that the remaining data is stored in the corresponding D2 block, it is necessary to read the data of the D2 block. When the D2 size information indicates that the remaining data is not stored in the corresponding D2 block, it is not necessary to read the data of the D2 block.

  As a result of the determination by the D2 read determination unit 23, when it is necessary to read data of the D2 block, the D2 block cache control unit 24 determines whether the data of the D2 block that needs to be read is stored in the D2 cache 29 or not. To do. When the desired D2 block data is stored in the D2 cache 29 (cache hit), the D2 block cache control unit 24 reads the desired D2 block data from the D2 cache 29. When the desired D2 block data is not stored in the D2 cache 29 (cache miss), the D2 block cache control unit 24 reads the desired D2 block data from the external memory 9 and stores it in the D2 cache 29.

  The residual data acquisition unit 25 acquires desired residual data from the D2 block data passed from the D2 block cache control unit 24. The variable length decoding unit 26 combines the data of the D1 block read by the D1 reading unit 21 and the residual data passed from the residual data acquisition unit 25 to perform variable length decoding processing. When the D2 reading determination unit 23 determines that it is not necessary to read the data of the D2 block, the variable length decoding unit 26 performs the variable length decoding process on the data of the D1 block read by the D1 reading unit 21. I do. The image storage unit 27 stores the pixel data decoded by the variable length decoding unit 26. The pixel data stored in the image storage unit 27 is sent to the request control unit 6 (see FIG. 2). Note that all data of the D2 table may be stored in the table cache 28. All D2 block data may be stored in the D2 cache 29.

  As an example of the image processing apparatus according to the second embodiment, for example, H.264. An encoder device conforming to the H.264 standard can be mentioned. H. H.264 is one of the video coding standards established by ITU-T (International Telecommunication Union Telecommunication Standardization Sector, International Telecommunication Union Telecommunication Standardization Sector). H. The H.264 encoder device may be constituted by a semiconductor integrated circuit device, for example. In the configuration shown in FIG. 2, the memory controller 3 may be configured with a semiconductor integrated circuit device, or the memory controller 3 and the encoder 2 may configure one semiconductor integrated circuit device. Note that the image processing apparatus according to the second embodiment can be applied to an encoder apparatus of another standard or an apparatus other than the encoder apparatus. An example of another standard is, for example, the MPEG (Moving Picture Experts Group) -2 standard.

FIG. 4 is an explanatory diagram of an example of the variable length encoding process according to the second embodiment. In FIG. 4, the circle represents the pixel 31 and the numerical value (128 or 129) in the pixel 31 represents the pixel value. Numerical values (0 and −1) around the 16 pixels 31 are difference values of pixel values of adjacent pixels. Generally, since the values of adjacent pixels are correlated, the difference values between adjacent pixels are concentrated around 0. Using this characteristic, for example, a difference value is obtained between adjacent pixels, a short code (for example, 00, 100, 110, etc.) is assigned to a small difference value such as 0, and a long code (for example, 00, 100, 110, etc.) is assigned. Bit length can be reduced.

  In the example illustrated in FIG. 4, the pixel value of the right pixel is subtracted from the pixel value of the left pixel between adjacent pixels. For the leftmost vertical column, the pixel value of the upper pixel is subtracted from the pixel value of the lower pixel. The uppermost pixel in the leftmost vertical column is not compressed. As an example, if the small area is 8 pixels × 2 pixels (16 pixels in total, 8 pixels horizontally and 2 pixels vertically), and each pixel is expressed with a fixed length of 8 bits, all pixels are uncompressed in the illustrated example. Then, the sum of the pixel values in the small area is 128 bits (= 8 bits × 16 pixels). On the other hand, for example, when code 00 is assigned to the difference value 0 and code 110 is assigned to the difference value −1, the total pixel value of the small area is 40 bits (= 8 bits of fixed length + “00”). 2 bits × 13 pieces + “110” 3 bits × 2 pieces). That is, in the illustrated example, the sum of the pixel values in the small area can be compressed from 128 bits to 40 bits.

  Note that the variable-length coding processing method is not limited to the example described here, and various types of variable-length coding processing can be applied. The variable length encoding process may be reversible or irreversible. Also, as shown in FIGS. 5 and 6, the variable length encoding unit 12 performs a plurality of variable length encoding processes, and selects the smallest data size from among the obtained variable length codes. It is good also as composition to do.

  FIG. 5 is a flowchart of an example of the variable length coding procedure according to the second embodiment. In the example shown in FIG. 5, there are three types of compression methods, but two types may be used, or four or more types may be used. As illustrated in FIG. 5, when the variable length coding process is started in the variable length coding unit 12, the variable length coding unit 12 performs, for example, first compression on the given small area image data. Variable-length encoding processing is performed by the method, the second compression method, and the third compression method (step S1, step S2, and step S3). In addition to these compression modes, a non-compression mode may be provided in which non-compression, that is, variable length encoding processing is not performed on image data of a given small region (step S4). When data in each mode is obtained, the variable length coding unit 12 selects a mode with the smallest data size (step S5). Then, the variable length encoding unit 12 outputs the data of the selected mode (step S6), and ends the variable length encoding process.

  FIG. 6 is an explanatory diagram schematically illustrating the structure of data obtained by the variable-length encoding process according to the second embodiment. As shown in FIG. 6, the data has a configuration in which, for example, bits 45, 46, 47, and 48 representing the mode are added to the heads of the data 41, 42, 43, and 44 of each mode. The example shown in FIG. 6 is applied when image data is processed in a plurality of modes in the variable length encoding unit 12 as in the example shown in FIG. When the non-compression mode is provided as in the examples illustrated in FIGS. 5 and 6, the data size of the output data of the variable length encoding unit 12 is a flag indicating the compression mode in the non-compression data (original pixel data). The data size is less than the data size including bits.

FIG. 7 is an explanatory diagram of the structure of variable length code data output from the variable length coding unit according to the second embodiment. As shown in FIG. 7, when the data size of the variable-length code data 51 output from the variable-length encoding unit 12 exceeds the preset data size, the data 51 has a preset data size. The D1 data 52 is divided into the remaining data 53 exceeding the preset data size. The data size set in advance is not particularly limited, but may be, for example, 8 bytes. When the data size of the variable-length code data 51 is equal to or smaller than a preset data size, the data 51 is only D1 data 52.

FIG. 8 is an explanatory diagram of an example of the memory storage structure of the D1 block by the image processing apparatus according to the second embodiment. As shown in FIG. 8, in the first area 61 of the external memory 9, a D1 block 62 is provided for each small area of the input image. The D1 block data is stored in the D1 block 62. In the example shown in FIG. 8, symbols such as A11, A12, and A21 are given to individual D1 blocks for convenience of explanation. In addition, the first area 61 is divided into groups of 4 × 4 D1 blocks and given different alphabets (A, B, C...), But the correspondence with the D2 block described later is clear. It is only grouped to make it, and has no special meaning. The number of D1 blocks in each group is not limited to 4 × 4. In order to avoid the figure from becoming complicated, the individual D1 blocks were omitted in groups other than the group marked with A.

  In the first area 61, the position of each D1 block 62 is predetermined for each small area. In the first area 61, the memory size assigned to each D1 block is determined in advance to a constant value. The memory size allocated to the D1 block is not particularly limited, but may be 8 bytes, for example.

FIG. 9 is an explanatory diagram of an example of the memory storage structure of the D2 block by the image processing apparatus according to the second embodiment. As shown in FIG. 9, in the second area 63 of the external memory 9, a D2 block 64 is provided for each group in which a plurality of D1 blocks are grouped (a group given an alphabet in FIG. 8). In the D2 block 64, D2 block data is stored. For example, “D2 block for A11 to A44” in FIG. 9 includes the remaining data accompanying the D1 data stored in, for example, 16 D1 blocks marked with A in the first area 61 shown in FIG. Stored. The same applies to other D2 blocks such as “D2 block for B11 to B44”. In the second area 63, the position of each D2 block is determined in advance for each group in which a plurality of D1 blocks are grouped. The memory size assigned to the D2 block is determined in advance to a constant value. Accordingly, in the second area 63, the D2 block in which the residual data associated with the D1 block is stored is uniquely determined.

  FIG. 10 is an explanatory diagram of an example of the D2 block by the image processing apparatus according to the second embodiment. As shown in FIG. 10, the D2 block 64 is provided with, for example, a residual data field 65 for each column and a residual data total value field 66 for each column. The residual data field 65 of each column stores the residual data of each column. The residual data total value field 66 of each column stores the total value RowSize of the residual data of each column. For example, the case where the D2 block 64 is a collection of the residual data associated with the D1 block marked with A in FIG. 8 will be described as an example. In the D2 block 64 shown in FIG. 10, the “residual data field in the first column”, which is one of the residual data fields 65 in each column, has residual data associated with the D1 blocks A11, A21, A31, and A41. Is stored. The same applies to the remaining data fields in the other columns such as “residual data field in second column”. In the “RowSize1 field”, which is one of the remaining data total value fields 66 of each column, the residual data associated with each D1 block of A11, A21, A31, and A41 included in the “residual data in the first column” is stored. The total value RowSize1 of the data size is stored. The same applies to the remaining data total value fields of other columns such as “RowSize2”.

  For example, in FIG. 8, when there is no residual data associated with each D1 block of A12, A22, A32, and A42 in the second column, the corresponding “second column residual data field” is stored in the D2 block 64. Absent. The value RowSize2 of the “RowSize2 field” is 0. Accordingly, in the individual D2 block 64, for example, when the remaining data of each column is packed and stored from the head address of the D2 block 64, the offset is offset by “RowSize1 field” from the head address of the D2 block 64. Thus, the address of the remaining data in the second column can be obtained. By offsetting the head address of the D2 block 64 by “RowSize1 field” + “RowSize2 field”, the address of the remaining data in the third column can be obtained. The same applies to the remaining data in the fourth column.

  FIG. 11 is an explanatory diagram of an example of the residual data field of each column by the image processing apparatus according to the second embodiment. As shown in FIG. 11, the residual data field 65 of each column stores, for example, a residual data field 67 of each row, an IsValid flag field 68, and a data size field 69 of residual data of each row. For example, a case will be described as an example where the remaining data field 65 of each column is targeted for each D1 block of A11, A21, A31, and A41 in FIG. 8, that is, the first column of the group to which A is attached. In the residual data field 65 of each column shown in FIG. 11, the “residual data field of the first row”, which is one of the residual data fields 67 of each row, has a residual associated with the D1 block of A11 of the first row. Data is stored. The same applies to the remaining data fields in other rows such as “residual data field in second row”.

  The IsValid flag field 68 stores an IsValid flag. The IsValid flag indicates whether or not residual data is attached to the D1 block of each row in each column. The number of bits in the IsValid flag field 68 is determined in advance. For example, if the D1 blocks A11, A21, and A41 are accompanied by residual data, but the D1 block of A31 is not accompanied by residual data, the value of the IsValid flag field 68 is 1101. In this case, in the residual data field 65 of each column, there is no residual data in the third row, and the residual data in the fourth row is arranged after the residual data in the second row.

  In the data size field 69 of the residual data of each row, the data size DiffSize of the residual data of each row includes the data size of the residual data of the first row, the data size of the residual data of the second row, and the residual of the third row. The data size is stored, for example, in units of bytes in the order of the data size of the data and the data size of the remaining data in the fourth row. For the lines with no residual data, the data size is not stored in the data size field 69 of the residual data of each line. The number of bits for indicating the size of the row in which the remaining data exists is determined in advance.

  Accordingly, in the residual data field 65 of each individual column, for example, by offsetting by “IsValid flag field” + “data size field of residual data in each row” from the start address of the field, the residual in the first row The address of the data can be obtained. By offsetting from the head address of the residual data field 65 of each column by “IsValid flag field” + “data size field of residual data of each row” + “data size of residual data of the first row”, The address of the remaining data on the second row can be obtained. The same applies to the remaining data on the third row and the remaining data on the fourth row.

  By providing the D1 block and the D2 block in the external memory 9 with the above-described configuration, the D2 block storing the residual data associated with the D1 block is uniquely determined for the D1 block. A column including the desired residual data is uniquely determined, and the desired residual data is uniquely determined in the column. That is, in the external memory 9, the storage position of the residual data corresponding to a certain D1 block is uniquely determined. In the D2 block, instead of collecting the residual data in each column, the remaining data may be combined in each row.

FIG. 12 is an explanatory diagram of an example of a memory storage structure of the D2 table by the image processing apparatus according to the second embodiment. As shown in FIG. 12, the D2 table 71 is provided with a D2 table entry field 72 for each D2 block. The D2 table entry field 72 stores D2 table entry data. The D2 table entry data includes, for example, D2 size information. In the external memory 9, the position of each D2 table entry field 72 is determined in advance. In the external memory 9, the memory size assigned to each D2 table entry field 72 is determined in advance to a constant value. Therefore, in the external memory 9, the D2 table entry data corresponding to the D2 block is uniquely determined.

  FIG. 13 is an explanatory diagram of an example of the D2 table entry field by the image processing apparatus according to the second embodiment. As shown in FIG. 13, in each D2 table entry field 72, for example, a data size field 73 and an IsValidRow flag field 74 of a D2 block are provided. In the data size field 73 of the D2 block, the data size D2Size of the D2 block is stored. The data size D2Size of the D2 block is the total value of the residual data field 65 of each column and the residual data total value field 66 of each column in the D2 block 64 shown in FIG. Therefore, if the value of the data size D2Size of the D2 block is equal to the value when the residual data is not stored in the D2 block 64, the compressed data is decoded (not decoded in the case of non-compressed data). There is no need to read the data. Note that the data size D2Size of the D2 block may be the total value of the remaining data field 65 of each column in the D2 block 64 shown in FIG.

  The IsValidRow flag field 74 stores an IsValidRow flag. The IsValidRow flag indicates whether or not residual data is attached to the D1 block in each column of the D2 block 64. For example, description will be given by taking the first column and the second column of the group with A in FIG. 8 as an example. When residual data is attached to any of the D1 blocks (A11, A21, A31, A41) in the first column, in the IsValidRow flag field 74 of the D2 table entry field 72 corresponding to the group to which A is attached, The value corresponding to the first column is 1. In the case where none of the D1 blocks (A12, A22, A32, A42) in the second column is accompanied by residual data, in the IsValidRow flag field 74 of the D2 table entry field 72 corresponding to the group to which A is attached, The value corresponding to the second column is 0. The same applies to the third column and the fourth column. Therefore, if each bit of the IsValidRow flag field 74 is 0, it is not necessary to read the data of the D2 block when decoding the compressed data (not decoding in the case of non-compressed data).

  The D2 table entry field 72 only needs to store the data size D2Size of the D2 block, but an IsValidRow flag or other information may be added. Examples of other information include, for example, a flag indicating whether or not residual data is attached to the D1 block in each row of the D2 block 64, a flag indicating whether or not residual data is attached to each D1 block 62, and the like. Can be mentioned. As shown in FIG. 13, when the data size D2Size of the D2 block and the IsValidRow flag are stored in the D2 table entry field 72, when determining whether it is necessary to read the data of the D2 block, Either may be confirmed. In the table cache 28 (see FIG. 3), some data of the D2 table entry field 72 may be cached, or all the data of the D2 table entry field 72 may be cached. Alternatively, only a part of the data in the D2 table entry field 72, for example, the data size D2Size of the D2 block may be cached in the table cache 28.

FIG. 14 is a flowchart of an example of a data writing procedure performed by the image processing apparatus according to the second embodiment. As shown in FIG. 14, when the writing process of data to the external memory 9 is started, first, the compression processing unit 1 stores a rectangular input image in the image storage unit 11 until one writing request is completed. Data is stored in order (step S11, step S12: No). When one writing request is completed (step S12: Yes), the compression processing unit 1 divides the input image data stored in the image storage unit 11 into, for example, small regions of 8 pixels × 2 pixels, and images of the small regions Data is read (step S13).

  Next, the compression processing unit 1 selects a compression mode (including a non-compression mode) by the variable length encoding unit 12 (step S14). The compression processing unit 1 uses the variable length encoding unit 12 to compress the small area image data in the selected mode (including the case of non-compression). In the compression processing unit 1, the data construction unit 13 generates D1 data and residual data (step S15). The compression processing unit 1 stores the generated D1 data and residual data in the compressed data storage unit 14. At this time, the compression processing unit 1 obtains the data size (DiffSize) of the remaining data by the data construction unit 13. The data size of the residual data is stored in the data size field 69 of the residual data in each row when the D2 block data is written in the external memory 9 (see FIG. 11).

  In the example shown in FIG. 8, the compression processing unit 1 performs steps S14 to S16 until the generation of the D1 data and the residual data is completed for the small region for one column of the group with the same alphabet (step S16: No). repeat. In the example shown in FIG. 8, when the processing is completed for the small region for one column of the group with the same alphabet (step S16: Yes), the compression processing unit 1 uses the data construction unit 13 to store the remaining data for the column. A bit corresponding to the total value (RowSize) and the IsValidRow flag is generated (step S17, see FIG. 13). The total value of the residual data is the total of the data size (DiffSize) of the residual data obtained in step S15 until the processing for the small area for one column is completed. The total value of the residual data is stored in the residual data total value field 66 of each column when the D2 block data is written to the external memory 9 (see FIG. 10). The bits of the IsValidRow flag are stored in the corresponding D2 table entry field 72 when the D2 size information is written in the external memory 9 (see FIG. 13).

  In the example shown in FIG. 8, the compression processing unit 1 repeats steps S13 to S18 until the processing is completed for the small regions for all the columns of the group given the same alphabet (step S18: No). In the example shown in FIG. 8, when the processing is completed for the small regions for all the columns of the group with the same alphabet (step S18: Yes), the compression processing unit 1 writes the D1 block data in the external memory 9 (step S18). S19).

  When the compressed data storage unit 14 has all the remaining data necessary for constructing one D2 block data (step S20: Yes), the compression processing unit 1 causes the data construction unit 13 to execute the D2 block. The data size (D2Size) of data is obtained. The data size of the D2 block is stored in the data size field 73 of the D2 block when the D2 size information is written in the external memory 9 (see FIG. 13). In addition, the compression processing unit 1 causes the data construction unit 13 to generate a D2 header based on the data of the corresponding D2 table entry and the total value (RowSize) of the residual data obtained in step S17 (step S21). For example, in the D2 block 64 shown in FIG. 10, the D2 header includes a residual data total value field 66 of each column. The D2 block is constructed by the D2 header and all the remaining data necessary for constructing one D2 block. The compression processing unit 1 writes the D2 block data and the corresponding D2 table entry data in the external memory 9 (step S22), and ends the process. Note that when writing data of the D2 table entry in the external memory 9, data for a plurality of D2 blocks may be written together. This is preferable because the number of write requests to the external memory 9 is reduced.

Data Reading Process FIG. 15 is a flowchart of an example of a data reading procedure performed by the image processing apparatus according to the second embodiment. As shown in FIG. 15, when the process of reading data from the external memory 9 is started, the developing unit 7 first obtains all D1 blocks corresponding to the image area specified by the read request, and reads D1. The unit 21 reads all necessary D1 block data from the external memory 9 (step S31). Further, the expansion unit 7 checks whether the data of the D2 table entry corresponding to the image area specified by the read request is stored in the table cache 28 by the D2 table cache control unit 22 (step S32). If not stored (step S32: No), the data of the desired D2 table entry is read from the external memory 9 (step S33).

  Next, the expansion unit 7 refers to the data size D2Size or IsValidRow flag of the D2 block included in the data of the D2 table entry given from the table cache 28 or the external memory 9 by the D2 read determination unit 23, and the D2 block It is determined whether it is necessary to read the data. Here, description will be made assuming that the data of the D2 block needs to be read. The expansion unit 7 checks whether or not the data of the D2 block that needs to be read is stored in the D2 cache 29 by the D2 block cache control unit 24 (step S34), and if not stored (step S34: No) Then, the desired D2 block data is read from the external memory 9 (step S35).

  Next, the expansion unit 7 uses the residual data acquisition unit 25 to calculate the address of the residual data field 65 (see FIG. 10) of the column including the desired residual data based on the residual data total value RowSize of the D2 block. Thus, a column including desired residual data is obtained (step S36). Next, the expansion unit 7 uses the residual data acquisition unit 25 to acquire the address of the desired residual data from the IsValid flag of the data in each column and the data size DiffSize of the residual data (see FIG. 11). Thereby, desired residual data is obtained (step S37). Until the remaining data for one column is obtained (step S38: No), the developing unit 7 repeats steps S37 to S38. When the acquisition of the residual data for one column is complete (step S38: Yes), the expansion unit 7 performs steps S36 to S39 until the acquisition of the residual data for all the columns necessary for decoding is completed (step S39: No). repeat.

  When the acquisition of the residual data of all the columns necessary for decoding is completed (step S39: Yes), the decompressing unit 7 combines the data of the D1 block with the residual data necessary for decoding by the variable length decoding unit 26, for example, Compressed data (including uncompressed data) in units of 8 pixels × 2 pixels is generated (step S40). Then, the decompressing unit 7 decodes the compressed data by the variable length decoding unit 26 (in the case of non-compressed data, the decoding is not performed, step S41). The expansion unit 7 stores the decoded pixel data in the image storage unit 27. The decompressing unit 7 repeats steps S41 to S42 until all the compressed data is decompressed for the area of the image designated by the read request (step S42: No). When all the compressed data are expanded (step S42: Yes), the expansion unit 7 transfers the pixel data to the encoder 2 (step S43), and ends the process. In step S41, a process of decoding a plurality of compressed data may be performed in parallel. In addition, the process of decoding the compressed data in step S41, the process of reading the data of the D2 block from the external memory 9 in step S35, the process of acquiring a column including the desired residual data in step S36, the desired process in step S37 You may perform in parallel with the process which acquires residual data.

  According to the second embodiment, the same effect as the first embodiment can be obtained. For example, when the pixel value of one pixel is represented by 8 bits, the area of 8 pixels × 2 pixels is a small area, and the memory size of the D1 block is 8 bytes, the memory band reduction effect in the reading process is as follows: is there. As shown in FIG. 16, when the image area 81 (area surrounded by a thick line) specified by the read request coincides with the boundary of the small area 82, the data for 128 pixels of the image area 81 specified by the read request. Is 128 bytes in the uncompressed case. On the other hand, when it is not necessary to read out the data of the D2 block in the second embodiment, the data for 128 pixels in the image area 81 specified by the read request (data for eight D1 blocks) is 64 bytes (= 8 8 bytes). Therefore, according to the second embodiment, the memory bandwidth can be reduced by 50%. Also, as shown in FIG. 17, when the image area 81 (area surrounded by a thick line) specified by the read request does not coincide with the boundary of the small area 82, 128 pixels of the image area 81 specified by the read request. In order to read out the data, it is necessary to read out data of 12 D1 blocks. Since the data of the 12 D1 blocks is 96 bytes (= 8 bytes × 12), if it is not necessary to read the data of the D2 block in the second embodiment, the memory bandwidth can be reduced by 25%.

(Example 3)
The third embodiment stores the remaining data of compressed data (including the case of non-compressed data) of a certain small area in the D2 block of the memory and corresponds to the small area adjacent to the small area in the second embodiment. It is also stored in the unused area of the D1 block. The third embodiment is different from the second embodiment in part of the process when generating the D1 block data in the data writing process and part of the process in reading the data of the D2 block in the data reading process. Other configurations and processes of the third embodiment are the same as those of the second embodiment. Only differences from the second embodiment will be described below.

  FIG. 18 is an explanatory diagram illustrating a data storage method in the image processing apparatus according to the third embodiment. In FIG. 18, FIG. 91 on the upper side is a schematic diagram showing a state of compressed data (including uncompressed data) in each small area. As shown in FIG. 18, in the third embodiment, the remaining data 92, 93, 94, 95, 96, 97 of compressed data (including non-compressed data) in a certain small area is stored in the first memory of the external memory 9. In the area 61, it is also stored in the unused area of the D1 block corresponding to the adjacent small area. In the example shown in FIG. 18, the residual data 92, 93, 94, 95, 96, 97 of a certain small area is stored in the D1 block corresponding to the small area adjacent to the lower side, but not limited to the lower side. The upper side, the left side, or the right side may be used.

  FIG. 19 is a chart showing IsValid flags corresponding to the state of compressed data (including the case of non-compressed data) in each small area shown in the upper side of FIG. In the chart 101 shown in FIG. 19, the correspondence relationship between the compressed data (including the case of non-compressed data) of each small region shown in FIG. , And so on, from the left to the right, the first row, the second row, and so on. In the illustrated example, the IsValid flag is 1 when there is residual data, and the IsValid flag is 0 when there is no residual data.

  FIG. 20 is a chart showing the IsNeighbor flag corresponding to the state of the first area of the external memory shown on the lower side of FIG. The IsNeighbor flag indicates whether or not residual data of a small area adjacent to the small area corresponding to the D1 block is stored in the D1 block. In the illustrated example, the IsNeighbor flag is 1 if the residual data of the adjacent small area is stored, and the IsNeighbor flag is 0 if the residual data of the adjacent small area is not stored. In the chart 102 shown in FIG. 20, the correspondence with the state of the first area 61 shown in FIG. 18 is the first line, the second line,... From the top to the bottom, from the left to the right. The first row, the second row, and so on. For example, the IsNeighbor flag is stored in the D2 table entry field 72 of the D2 table 71 (see FIG. 12). The IsValid flag may be stored in the D2 table entry field 72 of the D2 table 71, for example.

Data Writing Process FIG. 21 is a flowchart of an example of a data writing procedure performed by the image processing apparatus according to the third embodiment. As shown in FIG. 21, when the process is started, first, the compression processing unit 1 compresses the image data in the small area (uncompressed) in the same manner as in step S15 of the data writing process in the second embodiment. Including step S51). The small area processed at this time is set as a self-small area. Next, the compression processing unit 1 uses the data construction unit 13 to reduce the data size of the compressed data (including the case of non-compressed data) in its own small area to be equal to or smaller than the memory size of the D1 block of the external memory 9 corresponding to the self-small area. It is determined whether or not there is (step S52).

  When the data size is equal to or smaller than the memory size of the D1 block (step S52: Yes), there is an unused area in the D1 block corresponding to the self-small area. Therefore, the compression processing unit 1 sets the IsValid flag corresponding to the self-small region to 0 by the data construction unit 13 (step S53), and compresses the compressed data of the small region adjacent to, for example, the upper side (example in FIG. 18) of the self-small region. It is determined whether or not there is residual data (including the case of uncompressed data) (step S54). When there is corresponding residual data (step S54: Yes), the compression processing unit 1 can store the residual data of the adjacent small area in the unused area of the D1 block corresponding to the small area by the data construction unit 13. It is determined whether or not (step S55).

  When the data can be stored (step S55: Yes), the compression processing unit 1 uses the data construction unit 13 to combine the compressed data of the small area (including the case of non-compressed data) with the residual data of the adjacent small area. D1 block data is generated (step S56), and the IsNeighbor flag corresponding to the self-small area is set to 1 (step S57). If there is no corresponding residual data (step S54: No), or if the data size is large and cannot be stored in the unused area (step S55: No), the compression processing unit 1 uses the data construction unit 13 to reduce the size of the data. D1 block data is generated from the compressed data of the area (including the case of non-compressed data) (step S58), and the IsNeighbor flag corresponding to the self-small area is set to 0 (step S59).

  On the other hand, when the data size of the compressed data (including the case of non-compressed data) is larger than the memory size of the D1 block of the external memory 9 (step S52: No), the compression processing unit 1 uses the data construction unit 13 to reduce itself. The IsValid flag corresponding to the area is set to 1, and the IsNeighbor flag corresponding to the self-small area is set to 0 (step S60). The compression processing unit 1 uses the data construction unit 13 to divide the compressed data in the small area (including the case of non-compressed data) into D1 data and residual data, and generate D1 block data using the D1 data (step S61). . The compression processing unit 1 adds the residual data to the D2 block data by the data construction unit 13 (step S62). This residual data is used when the D1 block corresponding to the adjacent small area is not used when the processes from step S51 to step S62 are performed on the small area adjacent to, for example, the lower side of the own small area (example in FIG. 18). May be stored in the area. In the data write processing procedure shown in FIG. 14, the compression processing unit 1 performs the above-described processing for all the small areas included in the image area specified by the write request, and D1 block data corresponding to each small area, D2 block data and D2 size information are generated.

Data Reading Process FIG. 22 is a flowchart of an example of a data reading procedure performed by the image processing apparatus according to the third embodiment. As shown in FIG. 22, when the process is started, the developing unit 7 first acquires the coordinates and size of, for example, a rectangular image area designated by the read request (step S71). Next, the expansion unit 7 sets the D2_Read flag to 0 (step S72). The D2_Read flag indicates whether it is necessary to read data of the D2 block. The D2_Read flag is held by, for example, the development unit 7. Next, the developing unit 7 determines whether or not there is a small region whose IsValid flag is 1 among a plurality of small regions included in, for example, a rectangular image region designated as a read target (step S73). ). In the small area where the IsValid flag is 1, residual data is generated.

  When there is no small area whose IsValid flag is 1 (step S73: No), the developing unit 7 determines whether or not the D2_Read flag is 1 (step S77). Since the D2_Read flag is 0 (step S77: No), all the data of a plurality of small areas included in, for example, a rectangular image area designated as a reading target are stored in the D1 block. Therefore, the expansion unit 7 determines that there is no need to read the data of the D2 block.

  On the other hand, when there is a small area whose IsValid flag is 1 (step S73: Yes), the developing unit 7 sets all the IsValid flags that are included in, for example, a rectangular image area designated as the target of reading. The processing from the next step S74 to step S76 is performed for the small area. The developing unit 7 determines whether or not a small area where the residual data is stored is included in, for example, a rectangular image area designated as a reading target (step S74). When the small area where the residual data is stored is included (step S74: Yes), the expansion unit 7 determines whether or not the IsNeighbor flag corresponding to the small area where the residual data is stored is 1 (step S74: Yes). Step S75). When the small area of the residual data storage destination is not included (step S74: No), and the corresponding IsNeighbor flag is 0 even if the residual data storage destination small area is included (step S75: No) The expansion unit 7 sets the D2_Read flag to 1 (step S76).

  When the processing from step S74 to step S76 is completed for all the small regions whose IsValid flag is 1 included in the rectangular image region designated as the read target, for example, the expansion unit 7 has the D2_Read flag set to 1. Whether or not (step S77). When the D2_Read flag is 0 (step S77: No), all the residual data of a plurality of small areas included in the rectangular image area designated as the read target are all stored in the D1 block. . Therefore, the expansion unit 7 determines that there is no need to read the data of the D2 block. When the D2_Read flag is 1 (step S77: Yes), the expansion unit 7 issues a request for reading the data of the D2 block (step S78), and reads the data of the D2 block.

  The processing after the expansion unit 7 reads the data of the D2 block by the above processing is the same as the data reading processing in the second embodiment. However, when the data of the D2 block is not read, in step S40 of the data reading process in the second embodiment, the data of the D1 block corresponding to the self-small area and the lower side of the self-small area (example in FIG. 18). ) And the remaining data of the small area read from the D1 block corresponding to the small area adjacent to ().

  According to the third embodiment, the remaining data of the self-small area is stored in the D1 block corresponding to the small area adjacent to the self-small area, so that the D1 block corresponding to the small area adjacent to the self-small area is stored. When the data is read, the remaining data in the small area is also read. Accordingly, it is not necessary to read out data of the D2 block corresponding to the self-small area, so that the memory bandwidth can be reduced.

Example 4
In the fourth embodiment, a part of the D1 data of the compressed data (including the case of non-compressed data) in a certain small area is replaced with the unused area of the D1 block corresponding to the small area adjacent to the small area in the second embodiment. Is also stored. A part of the D1 data written in the unused area of the D1 block corresponding to the adjacent small area is called prescribed data. The data size of the prescribed data may be determined in advance, for example. For example, the data size of the prescribed data may be a data size that can expand the first line of D1 data. The fourth embodiment is different from the second embodiment in part of processing when generating D1 block data in the data writing process and part of processing when reading data of the D1 block in the data reading process. Other configurations and processes of the fourth embodiment are the same as those of the second embodiment. Only differences from the second embodiment will be described below.

  FIG. 23 is an explanatory diagram illustrating a data storage method in the image processing apparatus according to the fourth embodiment. 23, FIG. 111 on the upper side is a schematic diagram showing a state of compressed data (including non-compressed data) in each small area. As shown in FIG. 23, in the fourth embodiment, a part of D1 data 112, 113, 114, 115, 116, 117, 118, 119, 120 of compressed data (including non-compressed data) in a certain small area. , 121 and 122 are also stored in the unused area of the D1 block corresponding to the adjacent small area in the first area 61 of the external memory 9. In the example shown in FIG. 23, a part of D1 data 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122 of a certain small region is a D1 block corresponding to the small region adjacent to the upper side. However, it is not limited to the upper side, and may be the lower side, the left side, or the right side.

  FIG. 24 is a chart showing the IsNeighbor flag corresponding to the state of the first area of the external memory shown at the bottom of FIG. The IsNeighbor flag indicates whether or not a part of D1 data of a small area adjacent to the small area corresponding to the D1 block is stored in the D1 block. In the illustrated example, the IsNeighbor flag is 1 when D1 data of an adjacent small region is stored, and the IsNeighbor flag is 0 when D1 data of an adjacent small region is not stored. In the chart 103 shown in FIG. 24, the correspondence relationship with the state of the first region 61 shown in FIG. 23 is the first line, the second line,... From the top to the bottom, from the left to the right. The first row, the second row, and so on. For example, the IsNeighbor flag is stored in the D2 table entry field 72 of the D2 table 71 (see FIG. 12). The state of the IsValid flag corresponding to the state of the compressed data (including the case of non-compressed data) of each small area shown on the upper side of FIG. 23 is as shown in FIG. The IsValid flag may be stored in the D2 table entry field 72 of the D2 table 71, for example.

Data Writing Process FIG. 25 is a flowchart of an example of a data writing procedure performed by the image processing apparatus according to the fourth embodiment. As shown in FIG. 25, when the processing is started, first, the compression processing unit 1 compresses the image data in the small area (uncompressed) in the same manner as in step S15 of the data write processing in the second embodiment. Step S81). The small area processed at this time is set as a self-small area. Next, the compression processing unit 1 uses the data construction unit 13 to reduce the data size of the compressed data (including the case of non-compressed data) in its own small area to be equal to or smaller than the memory size of the D1 block of the external memory 9 corresponding to the self-small area. It is determined whether or not there is (step S82).

  When the data size is equal to or smaller than the memory size of the D1 block (step S82: Yes), the compression processing unit 1 sets the IsValid flag corresponding to the self-small area to 0 by the data construction unit 13 (step S83). The compression processing unit 1 can store, in the unused area of the D1 block corresponding to the self-small area, the specified data of the small area adjacent to, for example, the lower side of the self-small area (example in FIG. 23) by the data construction unit 13. It is determined whether or not (step S84).

  When the data can be stored (step S84: Yes), the compression processing unit 1 uses the data construction unit 13 to combine the compressed data (including the case of non-compressed data) of its own small area with the specified data of the adjacent small area. D1 block data is generated (step S85), and the IsNeighbor flag corresponding to the self-small area is set to 1 (step S86). When the data size of the corresponding specified data is large and the specified data cannot be stored in the unused area (step S84: No), the compression processing unit 1 uses the data construction unit 13 to compress the compressed data (uncompressed data) of its own small area. D1 block data is generated (step S87), and the IsNeighbor flag corresponding to its own small area is set to 0 (step S88).

  On the other hand, when the data size of the compressed data (including the case of non-compressed data) is larger than the memory size of the D1 block of the external memory 9 (step S82: No), the compression processing unit 1 uses the data construction unit 13 to reduce itself. The IsValid flag corresponding to the area is set to 1, and the IsNeighbor flag corresponding to the self-small area is set to 0 (step S89). The compression processing unit 1 uses the data construction unit 13 to divide the compressed data (including the case of non-compressed data) in its own small area into D1 data and residual data, and generate D1 block data with the D1 data (step S90). . The compression processing unit 1 adds the residual data to the D2 block data by the data construction unit 13 (step S91). The part of the prescribed data in the D1 data of the self-small area is obtained when the small area adjacent to, for example, the upper side (example in FIG. 23) of the self-small area is subjected to the processing of the adjacent small area. It may be stored in the unused area of the D1 block corresponding to the area.

  In the data write processing procedure shown in FIG. 14, the compression processing unit 1 performs the above-described processing for all the small areas included in the image area specified by the write request, and D1 block data corresponding to each small area, D2 block data and D2 size information are generated. In the example shown in FIG. 23, the prescribed data of a certain small area is stored in the D1 block corresponding to the small area adjacent to the upper side thereof, so that the small area from the bottom to the top in the image area specified by the write request. It is better to process.

Data Reading Process FIG. 26 is a flowchart illustrating an example of a data reading procedure performed by the image processing apparatus according to the fourth embodiment. As shown in FIG. 26, when the process is started, the developing unit 7 first acquires the coordinates and size of, for example, a rectangular image area designated by the read request (step S101). Next, the expansion unit 7 sets the D1_IsNeighbor_In flag to 0 (step S102). The D1_IsNeighbor_In flag indicates whether or not the prescribed data of the self-small area is stored in the D1 block corresponding to the small area adjacent to the self-small area. The D1_IsNeighbor_In flag is held by, for example, the expansion unit 7. Next, the expansion unit 7 determines that the vertical data size of the portion included in the image area to be read out of the small areas located at the lower end of the rectangular image area designated as the read target is the data size of the specified data, for example. It is determined whether or not the following is true (step S103).

  When it is larger than the data size of the prescribed data (step S103: No), the expansion unit 7 determines whether or not the D1_IsNeighbor_In flag is 1 (step S107). Since the D1_IsNeighbor_In flag is 0 (step S107: No), the data of the portion included in the image area to be read out for all the small areas positioned at the lower end of the rectangular image area designated as the read target, for example, Therefore, it is not stored in the D1 block corresponding to the small area adjacent to the upper side (example in FIG. 23). Accordingly, the expansion unit 7 issues a request to read out the data of the D1 block so as to include all the small areas positioned at the lower end of, for example, the rectangular image area designated as the read target (Step S109). Read data of D1 block.

  On the other hand, when the data size is equal to or smaller than the data size of the prescribed data (step S103: Yes), the developing unit 7 sets the D1_IsNeighbor_In flag to 1 (step S104). The development unit 7 performs the following steps S105 to S106 for the small regions adjacent to the upper side (example in FIG. 23) of all the small regions positioned at the lower end of the rectangular image region designated as the read target, for example. Perform the process. The expansion unit 7 determines whether or not the IsNeighbor flag corresponding to the small area adjacent to the upper side (example in FIG. 23) of the small area located at the lower end of the rectangular image area designated as the read target is 1, for example. Is determined (step S105). When the IsNeighbor flag is 0 (step S105: No), the expansion unit 7 sets the D1_IsNeighbor_In flag to 0 (step S106). When the processing from step S105 to step S106 is completed for the small areas adjacent to the upper side (example in FIG. 23) of all the small areas positioned at the lower end of the rectangular image area designated as the read target, for example, the developing unit 7 determines whether the D1_IsNeighbor_In flag is 1 (step S107).

  When the D1_IsNeighbor_In flag is 1 (step S107: Yes), the data of the portion included in the image area to be read is stored in all the small areas positioned at the lower end of the rectangular image area designated as the read target, for example. These are stored in the D1 block corresponding to the small area adjacent to each upper side (example in FIG. 23). Therefore, the expansion unit 7 issues a request to read the data of the D1 block for the part of the image area specified by the original read request excluding all the small areas located at the lower end of the image area (step S108). ) Read the data of the corresponding D1 block. For example, the data of the D1 block corresponding to the small area located at the lower end of the rectangular image area designated as the read target is not read.

  When the D1_IsNeighbor_In flag is 0 (step S107: No), for example, at least one small area of the small area located at the lower end of the rectangular image area designated as the read target is the image area to be read. The data of the portion included in is not stored in the D1 block corresponding to the small area adjacent to the upper side (example in FIG. 23). Accordingly, the expansion unit 7 issues a request to read out the data of the D1 block so as to include all the small areas positioned at the lower end of, for example, the rectangular image area designated as the read target (Step S109). Read data of D1 block. For example, the data of the D1 block corresponding to the small area located at the lower end of the rectangular image area designated as the reading target is read.

  The processing after the expansion unit 7 reads the data of the D1 block by the above processing is the same as the data reading processing in the second embodiment. However, if the data of the D1 block corresponding to the small area located at the lower end of the rectangular image area designated as the read target is not read, the data of the D1 block corresponding to the small area located at the lower end Is cut out from the data of the D1 block corresponding to the small area adjacent to the upper side (example in FIG. 23).

  According to the fourth embodiment, since a part of the D1 data in the small area is stored in the D1 block corresponding to the small area adjacent to the small area, when the data of the D1 block is read out, The number of D1 blocks is reduced. Accordingly, since the amount of data read from the D1 block is reduced, the memory bandwidth can be reduced. For example, as shown in FIG. 27, an image area 81 (area surrounded by a thick line) designated by a read request is divided into nine small areas 82 from the Mth row and the Nth column to the M + 2 row and the N + 2 column. In the case where the lower end of the region 81 of the image that is straddled and does not match the boundary of the small region 82 is specified as follows. When the fourth embodiment is not applied, it is necessary to read data of the D1 block corresponding to the nine small regions 82 from the Mth row and the Nth column to the M + 2 row and the N + 2 column. On the other hand, the data of the D1 block corresponding to the small area 82 from the (M + 2) th row and the Nth column to the (M + 2) th row and the (N + 2) th column located at the lower end of the image area 81 specified by the read request in the fourth embodiment. If it is not necessary to read out, the data of the D1 block corresponding to the six small regions 82 from the Mth row and the Nth column to the M + 1th row and the N + 2th column may be read. Therefore, according to the fourth embodiment, the memory bandwidth can be reduced.

(Example 5)
Example 5 is a combination of Example 3 and Example 4. In the fifth embodiment, a part of the process when generating the D1 block data in the data writing process and a part of the process when reading the data of the D1 block and the data of the D2 block in the data reading process are described in the second embodiment. However, the other configurations and processes of the fifth embodiment are the same as those of the second embodiment. Only differences from the second embodiment, the third embodiment, and the fourth embodiment will be described below.

  FIG. 28 is a table illustrating an example of an IsNeighbor flag according to the fifth embodiment. As shown in the chart 104 in FIG. 28, the IsNeighbor flag is represented by, for example, 2 bits for each small area. For example, the upper bit of the two bits of the IsNeighbor flag indicates whether or not the D1 block stores residual data of a small area adjacent to the small area corresponding to the D1 block. For example, the upper bit of the IsNeighbor flag is 1 when the residual data of the adjacent small region is stored, and the upper bit of the IsNeighbor flag is 0 when the residual data of the adjacent small region is not stored. For example, the lower bits of the two bits of the IsNeighbor flag indicate whether or not the D1 block has prescribed data for D1 data in a small area adjacent to the small area corresponding to the D1 block. For example, the lower bit of the IsNeighbor flag is 1 when the specified data of the adjacent small area is stored, and the lower bit of the IsNeighbor flag is 0 when the specified data of the adjacent small area is not stored.

Data Writing Process FIGS. 29 to 31 are flowcharts illustrating an example of a data writing procedure by the image processing apparatus according to the fifth embodiment. 30 and 31 are a continuation of FIG. As an example, the residual data of the small area adjacent to the upper side of the small area corresponding to the D1 block is stored in a certain D1 block, and the D1 data of the small area adjacent to the lower side of the small area corresponding to the D1 block is stored. Suppose that you want to store specified data. In addition, it is good also as a structure which stores the remainder data of the small area | region adjacent to the side other than upper side, and the prescription | regulation data of D1 data of the small area | region adjacent to the side other than the lower side in a certain D1 block.

  As shown in FIG. 29, when the processing is started, first, the compression processing unit 1 performs the same processing from step S51 to step S55 of the data writing processing in the third embodiment, and the data construction unit 13 It is determined whether or not the remaining data of the small area adjacent on the upper side can be stored in the unused area of the D1 block corresponding to the self-small area (steps S111 to S115). When the data can be stored (step S115 in FIG. 29: Yes), the compression processing unit 1 uses the data construction unit 13 to add a small area adjacent to the lower side to the remaining unused area of the D1 block corresponding to the self-small area. It is determined whether or not the specified data can be stored (step S116 in FIG. 30). The remaining unused area of the D1 block is the remainder obtained by subtracting the data size required to store the remaining data of the small area adjacent to the upper side from the data size of the unused area of the D1 block corresponding to the own small area. It is the area of.

  When the data can be stored (step S116 of FIG. 30: Yes), the compression processing unit 1 uses the data construction unit 13 to compress the compressed data of the small area (including the case of non-compressed data) and the small area adjacent to the upper side. The D1 block data is generated by combining the remaining data and the prescribed data of the small area adjacent to the lower side (step S117 in FIG. 30). The compression processing unit 1 sets the IsNeighbor flag corresponding to its own small area to 11 (step S118 in FIG. 30). When the data size of the corresponding specified data is large and the specified data cannot be stored in the remaining unused area (step S116 in FIG. 30: No), the compression processing unit 1 uses the data construction unit 13 to compress the small area. D1 block data is generated from the data (including the case of non-compressed data) and the residual data of the small area adjacent on the upper side (step S119 in FIG. 30). The compression processing unit 1 sets the IsNeighbor flag corresponding to the self-small area to 10 (step S120 in FIG. 30).

  In step S114 of FIG. 29, there is no residual data in the compressed data (including the case of non-compressed data) in the small area adjacent to the upper side of the small area (step S114: No), and the data size is large. If the data cannot be stored in the unused area (step S115 in FIG. 29: No), the compression processing unit 1 uses the data construction unit 13 to set the small area adjacent to the unused area of the D1 block corresponding to the own small area. It is determined whether or not the prescribed data can be stored (step S121 in FIG. 31). When the data can be stored (step S121 in FIG. 31: Yes), the compression processing unit 1 uses the data construction unit 13 to store the compressed data of the small area (including the case of non-compressed data) and the small area adjacent to the lower side. The D1 block data is generated together with the prescribed data (step S122 in FIG. 31). The compression processing unit 1 sets the IsNeighbor flag corresponding to the self-small area to 01 (step S123 in FIG. 31). When the data size of the corresponding specified data is large and the specified data cannot be stored in the unused area (step S121 of FIG. 31: No), the compression processing unit 1 uses the data construction unit 13 to compress the compressed data ( D1 block data is generated (including the case of uncompressed data) (step S124 in FIG. 31). The compression processing unit 1 sets the IsNeighbor flag corresponding to the self-small area to 00 (step S125 in FIG. 31).

  If the data size of the compressed data (including the case of non-compressed data) in the small area is larger than the memory size of the D1 block corresponding to the small area (step S112: No) in FIG. The unit 1 sets the IsValid flag corresponding to the self-small area to 1 and sets the IsNeighbor flag corresponding to the self-small area to 00 (step S126 in FIG. 29). The compression processing unit 1 uses the data construction unit 13 to divide the compressed data in the small area (including the case of non-compressed data) into D1 data and residual data, and generate D1 block data with the D1 data (FIG. 29). Step S127). In the compression processing unit 1, the data construction unit 13 adds the remaining data to the D2 block data (step S128 in FIG. 29). This residual data may be stored in the unused area of the D1 block corresponding to the adjacent small area when the same processing is performed on the small area adjacent to the small area.

Data reading process In the data reading process, the process for determining whether or not to read the data of the D2 block is the same as in the third embodiment. When determining whether or not to read the data of the D2 block, it is only necessary to check the upper bit of the IsNeighbor flag. In the data read process, the process for determining whether or not to read the data of the D1 block corresponding to the small area located at the lower end of the image area designated as the read target is the same as in the fourth embodiment. When determining whether or not to read the data of the D1 block, the lower bit of the IsNeighbor flag may be confirmed.

(Example 6)
In the sixth embodiment, the remaining data of the compressed data (including the case of non-compressed data) in a certain small area in the third embodiment is represented by D1 corresponding to the small area adjacent to either the upper, lower, left or right side of the small area. It is also stored in the unused area of the block. The sixth embodiment is different from the second embodiment in part of the process when generating the D1 block data in the data writing process and part of the process in reading the data of the D2 block in the data reading process. Other configurations and processes of the sixth embodiment are the same as those of the second embodiment. Only differences from the second and third embodiments will be described below.

  FIG. 32 is a table illustrating an example of an IsNeighbor flag according to the sixth embodiment. As shown in the chart 105 in FIG. 32, the IsNeighbor flag is represented by, for example, 2 bits for each small area. For example, the IsNeighbor flag is 10 when the residual data of the small area adjacent on the upper side is stored in the D1 block of the self-small area, and 01 when the residual data of the small area adjacent on the lower side is stored. And 11 when the residual data of the small area adjacent on the left side is stored. For example, if the residual data of the adjacent small area is not stored in the D1 block of the self-small area, the IsNeighbor flag is 00.

Data Writing Process FIGS. 33 to 35 are flowcharts illustrating an example of a data writing procedure by the image processing apparatus according to the sixth embodiment. 34 and 35 are a continuation of FIG. As shown in FIG. 33, when the processing is started, first, the compression processing unit 1 performs the same processing from step S51 to step S53 of the data writing processing in the third embodiment, and the compression processing unit 1 The data construction unit 13 sets the IsValid flag corresponding to the self-small area to 0 (steps S131 to S133). The compression processing unit 1 determines whether or not there is residual data in the compressed data (including the case of non-compressed data) in the small area adjacent to the upper side of the small area by the data construction unit 13 (step in FIG. 33). S134).

  When there is corresponding residual data (step S134 in FIG. 33: Yes), the compression processing unit 1 uses the data construction unit 13 to set the small region adjacent to the upper side to the unused region of the D1 block corresponding to the self-small region. It is determined whether or not the remaining data can be stored (step S135 in FIG. 33). When the data can be stored (step S135 in FIG. 33: Yes), the compression processing unit 1 uses the data construction unit 13 to compress the compressed data of the small area (including the case of non-compressed data) and the small area adjacent to the upper side. Together with the remaining data, D1 block data is generated (step S136 in FIG. 33), and the IsNeighbor flag corresponding to the self-small area is set to 10 (step S137 in FIG. 33).

  If there is no residual data of the small area adjacent to the upper side (step S134 in FIG. 33), even if the data size is large and cannot be stored in the unused area (step S135 in FIG. 33: No), the compression processing unit 1, the data construction unit 13 determines whether there is residual data in the compressed data (including the case of non-compressed data) of the small area adjacent to the lower side of the own small area (step S138 in FIG. 34). . When there is corresponding residual data (step S138 in FIG. 34: Yes), the compression processing unit 1 uses the data construction unit 13 to place a small area adjacent to the unused area of the D1 block corresponding to the self-small area on the lower side. It is determined whether or not the remaining data can be stored (step S139 in FIG. 34). When the data can be stored (step S139 in FIG. 34: Yes), the compression processing unit 1 uses the data construction unit 13 to store the compressed data of the small area (including the case of non-compressed data) and the small area adjacent to the lower side. Are combined with the remaining data to generate D1 block data (step S140 in FIG. 34), and the IsNeighbor flag corresponding to the self-small area is set to 01 (step S141 in FIG. 34).

  If there is no residual data in the small area adjacent to the lower side (step S138: No in FIG. 34), or if the data size is large and cannot be stored in the unused area (step S138: No in FIG. 34), the compression process The unit 1 uses the data construction unit 13 to determine whether or not there is residual data in the compressed data (including the case of non-compressed data) in the small area adjacent to the left side of the small area (step S142 in FIG. 34). . When there is corresponding residual data (step S142 in FIG. 34: Yes), the compression processing unit 1 causes the data construction unit 13 to set the unused area of the D1 block corresponding to the own small area to the small area adjacent to the left side. It is determined whether or not the remaining data can be stored (step S143 in FIG. 34). When the data can be stored (step S143 in FIG. 34: Yes), the compression processing unit 1 uses the data construction unit 13 to compress the compressed data of the small area (including the case of non-compressed data) and the small area adjacent to the left side. Together with the remaining data, D1 block data is generated (step S144 in FIG. 34), and the IsNeighbor flag corresponding to the self-small area is set to 11 (step S145 in FIG. 34).

  If there is no residual data in the small area adjacent to the left side (step S142 in FIG. 34), or if the data size is large and cannot be stored in the unused area (step S143 in FIG. 34: No), the compression processing unit 1, the data construction unit 13 generates D1 block data with compressed data (including uncompressed data) in its own small area (step S146 in FIG. 35), and sets the IsNeighbor flag corresponding to its own small area to 00. (Step S147 in FIG. 35). On the other hand, when the data size of the compressed data (including the case of non-compressed data) is larger than the memory size of the D1 block of the external memory 9 (step S132: No in FIG. 33), the compression processing unit 1 uses the data construction unit 13 Accordingly, the IsValid flag corresponding to the self-small area is set to 1, and the IsNeighbor flag corresponding to the self-small area is set to 00 (step S148 in FIG. 33). The compression processing unit 1 uses the data construction unit 13 to divide the compressed data in the small area (including the case of non-compressed data) into D1 data and residual data, and generate D1 block data with the D1 data (FIG. 33). Step S149). The compression processing unit 1 adds the residual data to the D2 block data by the data construction unit 13 (step S150). This residual data may be stored in the unused area of the D1 block corresponding to the adjacent small area when the same processing is performed on the small area adjacent to the small area.

Data read processing The data read processing is the same as in the third embodiment. When determining the storage destination of the remaining data, the IsNeighbor flag may be confirmed.

DESCRIPTION OF SYMBOLS 9 Memory 12 Variable length encoding part 13 Data construction part 23 Judgment part 28,29 Storage part

Claims (5)

A variable length encoding unit that divides an input image into a plurality of small regions and encodes and compresses the image data for each small region by a variable length encoding method;
The compressed data encoded by the variable-length encoding unit is divided into first data having a predetermined data size or less and second data exceeding the predetermined data size, and the second data is divided into a plurality of pieces. Corresponding to the small area among a plurality of first blocks having a fixed memory size equal to or larger than the predetermined data size of the memory based on the first data. Based on the first block data and the third data written to the first block, the second block corresponding to the group of small areas grouped out of the plurality of second blocks of the memory is written. A data construction unit for generating second block data and size information of the third data;
An image processing apparatus comprising:   The data construction unit generates storage position information of each of the second data in the second block of the memory, and includes the storage position information in the second block data. The image processing apparatus according to 1.   The second block having a plurality of the first blocks and a plurality of the second blocks and storing the second data is uniquely associated with each small area, and The image processing apparatus according to claim 1, wherein the image processing apparatus is connected to a memory having a fixed memory size of the two blocks. A storage unit for storing part or all of the second block data and part or all of the size information of the third data;
A determination unit that determines whether to read out the second block data from the memory based on the size information of the second block data and the third data stored in the storage unit;
The image processing apparatus according to claim 1, further comprising: Dividing the input image into a plurality of small regions, and encoding and compressing the image data for each small region by a variable length coding method;
The compressed data encoded by the variable-length encoding unit is divided into first data having a predetermined data size or less and second data exceeding the predetermined data size, and the second data is divided into a plurality of pieces. Corresponding to the small area among a plurality of first blocks having a fixed memory size equal to or larger than the predetermined data size of the memory based on the first data. Based on the first block data and the third data written to the first block, the second block corresponding to the group of small areas grouped out of the plurality of second blocks of the memory is written. Generating size information of second block data and the third data;
An image processing method comprising:

Images