JP2006080793A - Image coder, method, compputer program, and computer readable storage medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To suppress rapid switching over the reciprocal and nonreciprocal coding processes pixels per blocksand suppress the prominent block boundary of a coded image. <P>SOLUTION: A block divider 103 divides image data inputted from an image input 101 into pixel blocks of a specified size. A reciprocative and nonreciprocative coders 105, 107 code the divided pixel blocks to generate coded data, respectively. A select signal holder 110 holds a signal showing either of the reciprocative/nonreciprocative coded data is selected for pixel blocks just before continuously interested pixel blocks. When the reciprocative coded data are selected just before pixel blocks, a selector 109 weights either the reciprocative coded data quantity or nonreciprocative coded data quantity to make easy to select the reciprocative coded data in the interested pixel blocks and thereafter compares these code quantities to output coded data of a less code quantity by a signal line 112 with a signal 113 showing the selected coded data are either reciprocative or nonreciprocative data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image data encoding technique.

  Conventionally, an image is divided into blocks, and includes lossless encoding means for lossless encoding of the block and lossy encoding means for lossy encoding, and the final result of one of the encoding results for each block is provided. An image encoding method for outputting as encoded data has been proposed.

  In such a method, in a natural image region where image quality degradation is relatively unnoticeable, the lossy coding method is selected and applied to increase the compression rate. On the contrary, characters, line drawings, and CG portions where image quality degradation is easily noticeable Are based on suppressing visual image quality deterioration by using a lossless encoding method.

In order to suppress image quality degradation and reduce the amount of code, it is important to select an appropriate encoding method for each region. Refer to the amount of information and the number of colors in lossless encoded data, A method for selecting reversibility (see, for example, Patent Document 1) is disclosed.
Japanese Patent Laid-Open No. 08-167030

  However, in the above-described conventional image processing apparatus, when reversible and irreversible are selected and encoded for each block, there are cases where many portions where the reversible block and the irreversible block are adjacent are generated. In particular, there are problems such as frequent switching in a part of the image that has a characteristic that becomes the boundary of the condition for determining the coding method, and it becomes a visual disturbance factor such that the boundary of the block can be identified. is there.

  The present invention has been made in view of the above-described problems, and intends to provide an encoding technique that suppresses frequent switching of reversible and irreversible blocks and reduces image quality degradation.

In order to achieve the object of the present invention, for example, an image encoding apparatus of the present invention comprises the following arrangement. That is,
An image encoding device for encoding image data,
Dividing means for dividing the image data into pixel blocks of a predetermined size;
Lossless encoding means for losslessly encoding the pixel block and generating lossless encoded data;
Irreversible encoding means for irreversibly encoding the pixel block and generating irreversible encoded data;
Storage means for storing selection information indicating which lossless or lossy encoded data is selected as the encoded data of the pixel block;
The storage unit stores the code amount of the lossless encoded data generated by the lossless encoding unit corresponding to the pixel block of interest and the code amount of the lossy encoded data generated by the lossy encoding unit. Comparing means for weighting and comparing either one according to the selected information of the pixel block adjacent to the focused pixel block,
A selection output unit configured to select and output the encoded data of which the code amount is determined to be small as a result of the weighted comparison as the encoded data of the target pixel block;

  According to the configuration of the present invention, it is possible to suppress frequent switching between lossless encoding and lossy encoding for each pixel block, and it is possible to suppress conspicuous block boundaries in a decoded image.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

<First Embodiment>
FIG. 18 is a diagram showing a basic configuration of the image processing apparatus according to the present embodiment.

  In the figure, reference numeral 1401 denotes a CPU which controls the entire apparatus using programs and data stored in a RAM 1402 and a ROM 1403 and executes image encoding processing which will be described later.

  A RAM 1402 has an area for storing programs and data downloaded from the external device via the external storage device 1407, the storage medium drive 1408, or the I / F 1409, and when the CPU 1401 executes various processes. A work area is also provided.

  Reference numeral 1403 denotes a ROM which stores a boot program, a setting program for the apparatus, and data.

  Reference numerals 1404 and 1405 denote a keyboard and a mouse, respectively. Various instructions can be input to the CPU 1401.

  A display device 1406 includes a CRT, a liquid crystal screen, and the like, and can display information such as images and characters.

  Reference numeral 1407 denotes an external storage device, which is a large-capacity information storage device such as a hard disk drive device, which stores an OS, a program for image encoding processing to be described later, image data to be encoded, and the like. These programs and data are loaded into a predetermined area on the RAM 1402 by the control according to.

  A storage medium drive 1408 reads programs and data recorded on a storage medium such as a CD-ROM or DVD-ROM and outputs them to the RAM 1402 or the external storage device 1407. Note that a program for image encoding processing and an image to be encoded, which will be described later, may be recorded on this storage medium. In this case, the storage medium drive 1408 stores these programs and data under the control of the CPU 1401. A predetermined area on the RAM 1402 is loaded.

  Reference numeral 1409 denotes an I / F, which connects an external device to the present apparatus through the I / F 1409 and enables data communication between the present apparatus and the external apparatus. For example, an image scanner device or the like is connected via the I / F 1409, and image data to be encoded is input to the RAM 1402 or the external storage device 1407 of the apparatus, or conversely, the RAM 1402 or the external storage device of the apparatus. The encoded image data generated from 1407 can be output to the outside of the apparatus. A bus 1410 connects the above-described units.

  In the above configuration, when the power of the apparatus is turned on, the CPU 1401 loads the OS from the external storage device 1407 to the RAM 1402 in accordance with the boot program stored in the ROM 1402. Then, the image processing application program stored in the external storage device 1407 is activated, and an image input via the interface 1409 is encoded. The encoding result is stored in a storage medium (such as a memory card) attached to the external storage device 1407 or the recording medium drive 1408.

  FIG. 1 is a functional configuration diagram of an apparatus when an image processing application according to the present embodiment is executed.

  As shown in FIG. 1, the image processing apparatus according to the present embodiment includes an image input unit 101, a line buffer 102, a block division unit 103, a tile buffer 104, a lossless encoding unit 105, a lossless code buffer 106, and an irreversible code. A non-reversible code buffer 108, a selector 109, a selection signal holding unit 110, and a code string forming unit 111. In the figure, reference numerals 112, 113, and 114 denote signal lines.

  Note that the buffers 102, 104, 105, and 108 and the selection signal holding unit 110 in the above configuration are secured in the RAM 1402 when this processing starts, and the other processing units are the main processing. Are constituted by each function program (subroutine program) called from. However, it is needless to say that the configuration shown in FIG. 1 may be realized by hardware. Note that the RAM 1402 is also used for processing in each processing unit other than the buffers 102, 104, 105, 108 and the selection signal holding unit 110 as appropriate.

  Hereinafter, with reference to FIG. 1, an image encoding process performed by the image processing apparatus according to the present embodiment will be described.

  Image data to be encoded by the image processing apparatus according to the present embodiment is RGB color image data, and each component (color component) is represented by a luminance value ranging from 0 to 255 in 8 bits. To do. Assume that the image data to be encoded is arranged in the order of dots, that is, in the raster scan order, and the pixels are arranged in the order of R, G, and B. The input image is composed of W pixels in the horizontal direction and H pixels in the vertical direction.

  The encoding target image data is input from the image input unit 101 in the raster scan order.

  The line buffer 102 has an area for storing a predetermined number of lines (Th) of image data, and sequentially stores the image data input from the image input unit 101. Hereinafter, data obtained by dividing the encoding target image data by the width of the Th line is referred to as a stripe. The capacity required for the line buffer 102, that is, the data amount of one stripe is W × Th × 3 (for RGB) bytes. For convenience of explanation, it is assumed that the number H of vertical pixels of an input image is an integral multiple of Th, and an incomplete stripe does not occur at the end of the image.

  When one line of image data, that is, image data for Th lines is stored in the line buffer 102, the block division unit 103 converts the image data for Th lines stored in the line buffer 102 into horizontal Tw pixels and vertical directions. The data is divided into rectangular blocks composed of Th pixels, read out in units of blocks, and stored in the tile buffer 104. Here again, for convenience of explanation, it is assumed that the horizontal pixel count W of an image is an integral multiple of Tw, and incomplete blocks do not occur when divided into rectangular blocks. Hereinafter, a rectangular block composed of horizontal direction Tw pixels and vertical direction Th pixels is referred to as a tile.

  FIG. 8 illustrates the relationship between image data to be encoded, stripes, and tiles. As shown in the figure, the i-th tile in the horizontal direction and the j-th tile in the vertical direction are denoted as T (i, j).

  The tile buffer 104 has an area for storing pixel data for one tile, and sequentially stores tile data output from the block dividing unit 103 every time encoding of one tile is started. The capacity required for the tile buffer 104 is Tw × Th × 3 (for RGB) bytes. The luminance value of the component c of the pixel at the horizontal pixel position x and the vertical pixel position y of the pixel data for one tile stored in the tile buffer 104 is defined as P (x, y, c). x is from 0 to Tw-1, y is from 0 to Th-1, and c is any one of R, G, and B.

  The lossless encoding unit 105 performs lossless encoding on the tile data stored in the tile buffer 104 to generate tile encoded data, and stores the tile encoded data in the lossless encoding buffer 106. Various methods can be applied as the lossless encoding processing in the lossless encoding unit 105. Here, an example in which predictive coding is applied is shown as an example, but in addition to this, a lossless mode of JPEG (ITU-T T.81 | ISO / IEC10918-1) recommended by ISO and ITU-T as an international standard system Or JPEG-LS (ITU-T T.87 | ISO / IEC14495-1) may be used.

  FIG. 2 is a block diagram showing the inside of the lossless encoding unit 105 in the embodiment. In the figure, 201 is a buffer, 202 is a predictor, 203 is a subtractor, 204 is a memory, 205 is a Huffman encoder, and 206 and 207 are signal lines. FIG. 2 shows an example of a lossless encoding unit that uses predictive conversion using neighboring pixels for series conversion processing for converting image data into prediction errors, and uses Huffman coding for encoding processing.

  The memory 205 stores a Huffman table generated based on a frequency distribution of prediction errors obtained by series conversion of some image data. As a general property of the frequency distribution of prediction errors, the appearance frequency is high centering on the prediction error 0, and the appearance frequency tends to decrease as the absolute value of the prediction error increases, so the Huffman code stored in the memory 204 is As shown in FIG. 4, a short code word is assigned near the prediction error 0, and a long code word is assigned to a portion where the absolute value of the prediction error is large.

  Hereinafter, the operation of the lossless encoding unit 105 will be described with reference to FIG.

  First, pixel data of the target tile is sequentially input from the tile buffer 104 via the signal line 206. It is assumed that tile data is input in raster scan order and each pixel is input in R, G, B order. The encoding process in the lossless encoding unit 105 is performed in units of pixels. Since the processing applied to each of R, G, and B is the same, the R component will be described as an example here.

  The buffer 201 has a capacity for storing tile data input from the signal line 206 for two lines.

  When the R component of the pixel of interest is P (x, y, R), the predictor 202 determines from the buffer 201 the luminance value a (= P (x-1, y, R) of the R component of the pixel immediately before the pixel of interest. )) And the luminance value b (= P (x, y−1, R); see FIG. 3) of the R component of the pixel one line before, and pay attention by calculating p = (a + b) / 2 A predicted value p for the luminance value P (x, y, R) of the R component of the pixel is generated.

  When the target pixel position is the first line of the tile, there is no pixel data one line before. In this case, processing is performed assuming that the pixel data of the previous line is a predetermined value (for example, 0). Further, when the pixel of interest is at the head position of each line, since the luminance value a does not exist, it is assumed that a = b. Thus, the value assumed as the pixel value around the tile may be the same value at the time of encoding and at the time of decoding.

  The subtracter 203 outputs the difference value e between the luminance value P (x, y, R) of the R component of the encoding target pixel (target pixel) and the predicted value p. The Huffman encoder 205 refers to a Huffman table stored in advance in the memory 204 and outputs encoded data corresponding to the difference value e from the signal line 207.

  The above-described processing is performed for all R components in the tile, and reversible tile encoded data of the R component is generated and output. Then, following the encoding of the R component, the G component and B are similarly performed. As a result, the lossless encoding buffer 105 stores the lossless encoded data of all the R, G, and B components of the tile of interest.

  On the other hand, at the same time as the lossless encoding process by the lossless encoding unit 105, the lossy encoding unit 107 generates tile encoded data by encoding the tile data stored in the tile buffer 104 by the lossy encoding method, The data is stored in the lossy encoding buffer 108. Similar to the lossless encoding unit 105, various lossy encoding methods such as JPEG (ITU-T T.81 | ISO / IEC10918-1) and JPEG2000 (ITU-T T.800 | ISO / IEC15444-1). However, it is assumed here that the JPEG baseline method is applied. Since JPEG is publicly known, a description thereof is omitted here. However, the same Huffman table and quantization table used for JPEG encoding are used for all tiles, and the frame header, scan header, various tables, etc. that are common to all tiles are irreversible encoding buffers. It is assumed that only the encoded data portion is stored without being stored in 108. That is, only the entropy encoded data segment from immediately after the scan header to immediately before the EOI marker is stored in the configuration of the general JPEG baseline encoded data shown in FIG. For the sake of simplification of explanation, it is assumed that the restart interval is not defined by the DRI and RST markers, and the number of lines is not defined by the DNL marker.

  As described above, the lossless encoding unit 105 and the lossy encoding unit 107 perform encoding on the target tile data stored in the tile buffer 104. As a result, lossless and lossy encoded data are stored in the lossless encoding buffer 106 and the lossy encoding buffer 108, respectively.

  The selector 109 selects the code amount CLK of lossless tile encoded data stored in the lossless encoding buffer 106 and the code amount CLH of lossy tile encoded data stored in the lossy encoding buffer 108. In comparison, either one of the tile encoded data is selected and read out and sent to the code forming unit 111 via the signal line 112. At this time, the selection signal holding unit 110 holds a signal indicating which one has been selected. The selection signal holding unit 110 clears “0” when starting to encode one stripe.

  The principle of this encoded data selection process by the selector 109 will be described with reference to the flowchart shown in FIG.

  First, the code amount CLK of reversible tile encoded data stored in the lossless encoding buffer 106 is acquired (step S601). Next, the code amount CLH of the irreversible tile encoded data stored in the irreversible encoding buffer 108 is acquired (step S602).

  These coding amounts are easy to understand when considered as address pointers when storing the coded data in the respective buffers (reset to “0” every time one tile is coded).

  Subsequently, a selection signal S indicating whether reversible or lossy encoded data is selected as encoded data corresponding to the immediately preceding tile T (i−1, j), which is input from the selection signal holding unit 110 described later. (I-1, j) is compared with 0 (step S603), and if S (i-1, j) ≠ 0, a predetermined code amount N is added to the code amount CLH of the irreversible tile encoded data. (Step S604). When S (i-1, j) = 0, N is not added.

  Here, the selection signal S () = 0 indicates that lossy encoded data is selected as the encoded data of the tile T (), and S () = 1 indicates that lossless encoded data is selected. It shows that.

  The code amount CLK of the reversible tile encoded data is compared with the code amount CLH of the irreversible tile encoded data (step S605), and if CLK <CLH, the selection result of the tile of interest is set to S (i, j) as “ 1 (logical high) "is set and output to the signal line 113 (step S606). Then, the reversible tile encoded data stored in the lossless encoding buffer 106 is read and output to the signal line 112 (step S607).

  On the other hand, if CLK ≧ CLH, “0 (logic low)” is set to S (i, j), and this is output to the signal line 113 (step S608) and stored in the lossy encoding buffer 108. The lossy tile encoded data is read and output to the signal line 112 (step S609).

  When reversible tile encoded data is selected for the previous tile by the above processing, a predetermined code amount N is added to the code amount of the irreversible tile encoded data of the tile of interest to compare the code amount As a result, reversible tile encoded data is easily selected for the tile of interest. For this reason, when the reversible and irreversible code amounts antagonize, it is possible to suppress frequent switching of the coding method selected in the adjacent tile.

  The selection signal holding unit 110 holds the selection signal S (i, j) input via the signal line 113, and when selecting the method for the next tile, this selection signal S (i−1) of the immediately preceding tile is selected. , J). It is assumed that the tile selection signal S (−1, j) just before the tile required for selecting the method of the leftmost tile T (0, j) of the image is 0. For this reason, the selection signal holding unit 110 is cleared to “0” when encoding one stripe is started.

  The code string forming unit 111 combines the selection signal S (i, j) of each tile input through the signal line 113 and the selected tile encoded data input through the signal line 112, and adds necessary additions. Information is added to form a code string to be output from the image processing apparatus and output from the signal line 114.

  FIG. 9 shows the data format of the output code string of the image processing apparatus. At the beginning of the output code string, information necessary for decoding the image, such as the number of pixels in the horizontal direction, the number of pixels in the vertical direction, the number of components, the number of bits of each component, the width of the tile, the height, etc. Additional information is attached as a header. The header portion includes not only information about the image data itself but also information related to encoding such as a Huffman encoding table and a quantization table used in common for each table.

  FIG. 10 is a diagram showing the configuration of the output code string of each tile. In the case of the embodiment, since the reversible and lossy encoded data are mixed in the image data, the corresponding tile selection signal S (i) is used for the determination as shown in FIGS. , J). The information added to the head may be 1 bit, or 1 byte in order to facilitate concatenation with tile encoded data. Although not shown in the configuration of the tile encoded data in FIG. 10, a special marker is set so that a predetermined value does not occur in the encoded data, and the beginning or end of each tile data is set. Random access in units of tiles can be made possible by placing a marker on or managing the length of the code string of each tile by including it in the head of the tile or the header part of the head of the encoded data. .

  Next, the flow of encoding processing of image data to be encoded by the image processing apparatus according to the embodiment will be described with reference to the flowchart of FIG.

  Note that the program according to the figure is loaded in the RAM 1402, and the CPU 1401 executes it.

  First, data for one stripe of image data to be encoded is stored in the line buffer 102 (step S501). Next, for the stripe stored in the line buffer 102, one tile data is taken out by the block dividing unit 103 and stored in the tile buffer 104 (step S502). Subsequently, the lossless encoding process (step S503) of the tile data by the lossless encoding unit 105 and the lossy encoding process (step S504) by the lossy encoding unit 107 are performed in parallel, and the lossless encoding buffer 106 is loaded. The lossy encoded tile data is stored in the lossy encoding buffer 108. If parallel processing cannot be performed, the process of step S504 may be performed after (or before) step S503.

  In any case, when the tile data encoding process is completed in both reversible and lossy manner, the selector 109 refers to the selection signal S (i−1, j) input from the selection signal holding unit 110 and is reversible. The code amount CLK of the tile encoded data is compared with the code amount CLH of the irreversible tile encoded data (step S505), and one of the tile encoded data is selected and read (step S506). This comparison and selection process is performed by the process shown in FIG.

  Next, it is determined whether or not the encoded tile is the last tile in the stripe (step S507). If it is not the last tile (NO), the processing target is transferred to the adjacent tile and the processing is continued from S502. If it is the last tile (YES), the process proceeds to step S508.

  In step S508, it is determined whether or not the encoded stripe is the last stripe of the image. If it is not the last stripe (NO), the process moves to step S501 to read the data of the next stripe and continue the process. . If it is the last stripe, the encoding process for the target image data is terminated.

  The encoded tile data output from the selector 109 in step S506 is appropriately combined by the code string forming unit 111 to form encoded image data that is output from the image processing apparatus according to the present embodiment.

  In step S604 in FIG. 6, the predetermined number N (positive value) is added to CLH. However, since it is a relative value, the predetermined number N may be subtracted from CLK.

  In addition, in order to decode the encoded data generated by the image processing apparatus of the present embodiment, a reversible and irreversible decoder is selected and switched with reference to S (i, j) indicated at the head of each tile, Each tile may be decoded according to the reverse procedure of the encoding process.

  As described above, the image processing apparatus according to the present embodiment divides an image into tiles, compares the lossless and irreversible code amounts for each tile, and uses the smaller encoded data of the tiles. By comparing the code amount with the reversible and irreversible selection status of the previous tile, it is possible to prevent frequent switching of the coding method in the part where the lossless and irreversible code amount antagonize. can do. As a result, it is possible to reduce visual interference caused by the adjoining reversible and irreversible tiles.

  If JPEG baseline encoding is used as lossy encoding and JPEG-LS is used as lossless encoding, the following synergistic effect can be achieved.

  JPEG baseline encoding is a compression encoding technique that is particularly excellent for natural images, while JPEG-LS is a compression encoding technique that can realize a high compression rate for character / line images, on the contrary. Therefore, when encoding an image in which natural images and character / line images are mixed, the probability that irreversible encoded data is selected in the natural image region by comparing the encoded data of each tile and selecting and outputting the smaller one. Therefore, the probability that lossless encoded data is selected in the character / line drawing region is high. In other words, when JPEG baseline encoding and JPEG-LS are used, encoded data corresponding to each image area is selected without a special image area determination circuit (or its determination processing program). Thus, it is possible to suppress the lossless and lossy encoded data from occurring alternately to some extent. In addition, by performing the processing as described above, the occurrence of such a problem can be further suppressed, so that encoded data excellent in image quality can be generated. The same applies to the second and third embodiments described below. The same effect can be obtained when the JPEG extension method for 12-bit images instead of the JPEG baseline or JPEG2000 is applied, such as when the bit accuracy of the image data to be encoded is 8 bits or more. It is done.

[Second Embodiment]
In the first embodiment, when selecting the lossless and lossy encoded data of the target tile, only the selection result of the previous tile is referred to, but a plurality of already-encoded tiles around the target tile are selected. An example of referring to the result will be described as a second embodiment.

  FIG. 11 is a block diagram illustrating a functional configuration of the image processing apparatus according to the second embodiment. Blocks that are the same as those in FIG. 1 described in the first embodiment are given the same numbers, and descriptions thereof are omitted.

  The image processing apparatus according to the second embodiment includes an image input unit 101, a line buffer 102, a block division unit 103, a tile buffer 104, a lossless encoding unit 105, a lossless encoding buffer 106, an irreversible encoding unit 107, A lossy code buffer 108, a selector 1101, a selection signal buffer 1102, a code amount weight calculation unit 1103, and a code string formation unit 111 are provided. In the figure, reference numerals 112, 113 and 114 denote signal lines. The basic configuration of the image processing apparatus according to the second embodiment is the configuration shown in FIG. 18 as in the first embodiment, and is realized by a program that causes the computer to realize the functions of the units shown in FIG. It is assumed that this program is loaded from the external device into the RAM 1402 via the external storage device 1407, the storage medium drive 1408, or the I / F 1409.

  Hereinafter, processing performed by the image processing apparatus according to the second embodiment will be described with reference to FIG.

  The image data to be encoded by the image processing apparatus according to the second embodiment is RGB color image data as in the first embodiment, and each component (color) is 8 bits, and is 0 to 255. It is assumed to be composed of pixel data representing a range of luminance values. Assume that the image data to be encoded is arranged in the order of dots, that is, in the raster scan order, and the pixels are arranged in the order of R, G, and B. The image is composed of horizontal W pixels and vertical H pixels as in the first embodiment.

  In the image processing apparatus according to the second embodiment, as in the image processing apparatus according to the first embodiment, the image data is divided into tiles, and the lossless encoding unit 105 and the lossy encoding unit for each tile. In 107, lossless and lossy encoding is performed, and lossless and lossy tile encoded data are stored in the lossless encoding buffer 106 and the lossy encoding buffer 108, respectively.

  The selector 1101 subtracts the code amount weight W input from the code amount weight calculation unit 1103 from the code amount CLK of the reversible tile encoded data stored in the lossless encoding buffer 106, and converts this to the lossy encoding buffer 108. Compared with the code amount CLH of the irreversible tile encoded data stored in, one tile encoded data with a smaller code amount is read and output from the signal line 112. Further, when reversible tile encoded data is selected, “1 (logical high)” is output from the signal line 113 as the selection signal S (i, j) of the tile T (i, j) of interest and is irreversible. Similarly, when the tile encoded data is selected, a signal “0 (logic low)” is output.

  FIG. 13 shows a code selection processing procedure by the selector 1101 in the second embodiment.

  First, the code amount CLK of the reversible tile encoded data stored in the lossless encoding buffer 106 is acquired (step S1301). Next, the code amount CLH of the irreversible tile encoded data stored in the irreversible encoding buffer 108 is acquired (step S1302). Subsequently, the code amount weight coefficient W (detailed later) input from the code amount weight calculation unit 1103 described later is subtracted from the reversible code amount CLK, and the result is set to CLK ′ (step S1303).

  Next, CLK ′ after subtraction is compared with a predetermined threshold value TH1 (step S1304). If CLK ′ <TH1 (YES), the process proceeds to step S1306. If CLK ′ ≧ TH1, the process proceeds to step S1305 (NO).

  In step S1305, CLK 'after subtraction is compared with CLH (step S1305). If CLK' <CLH (YES), the process proceeds to step S1306. If CLK'≥CLH (NO), the process proceeds to step S1308.

  In step S 1306, “1” is set in S (i, j) as a result of selecting the tile of interest, and this is output to the signal line 113. Subsequently, the reversible tile encoded data stored in the lossless encoding buffer 106 is read out and output to the signal line 112 (step S1307).

  On the other hand, in step S1308, S (i, j) is set to “0”, and this is output to the signal line 113, followed by reading the lossy tile encoded data stored in the lossy encoding buffer 108. Is output to the signal line 112 (step S1309).

  When CLK ′ after subtracting the code amount weight W by the above processing is smaller than a predetermined threshold value TH1, reversible tile encoded data is output regardless of the magnitude relationship with CLH, and is equal to TH1? If it is more than that, CLK ′ and CLH are compared, and the tile encoded data with the smaller code amount is selected and output.

  Next, the calculation principle of the “code amount weighting coefficient W” in the above description will be described.

  The selection signal buffer 1102 has a capacity that can store the selection signals S of the number of tiles (W / Tw) included in one stripe. When the tile of interest is represented by T (i, j), the tile T (i-1, j) immediately preceding the same stripe from the tile T (i, j-1) at the same horizontal position of the immediately preceding stripe. The selection signals S (i, j−1) to S (i−1, j) for the tiles up to (the tiles indicated by halftone dots in FIG. 12) are held. Note that S (i, −1), which is referred to when the tile T (i, 0) in the first stripe in the image is the target tile, and the first tile T (0, j) in the stripe are focused. It is assumed that S (−1, j) referred to in the case of a tile is 0.

The code amount weight calculation unit 1103 selects the selection signal S (i, j-1) of the tile immediately above the tile T (i, j) of interest and the selection signal T (i-1, j) of the previous tile. The data is read from the buffer 1102, and the code amount weighting coefficient W is calculated by the following formula and output.
W = {S (i−1, j) + S (i, j−1)} × M
Here, “M” is a predetermined constant.

  According to the above equation, the code amount weighting coefficient W is obtained when both S (i−1, j) and S (i, j−1) are “0”, that is, T (i−1, j) and T ( If lossy encoding is selected in both i, j-1), W = 0. When either one is lossless encoding and the other is lossy encoding, W = M, and when both lossless encoding is selected, W = 2M.

  FIG. 14 is a diagram illustrating the relationship between the lossless code amount CLK (before subtraction of the code amount weight W), the irreversible code amount CLH, and the determination result of the encoding method by the selector 1101.

  A hatched area indicates an area where lossless encoding is selected, and the other indicates an area where lossy encoding is selected. The code amount weight W = 0, M, or 2M input to the selector 1101, FIG. 14A shows the case where W = 0, FIG. 14B shows the case where W = M, FIG. ) Shows the case of W = 2M. It can be seen that as W increases, the range in which the reversible tile encoded data is selected in the selector 1101 (the hatched area in the figure) is widened, and it is easy to select lossless encoding.

  By determining the weighting factor W as described above and performing the processing of FIG. 13, either one of lossless and lossy encoded data is selectively output for each tile, and the main image is generated by the code string forming unit 111. A code string serving as an output of the processing device is formed and output from the signal line 114. The configuration of the code string output in the second embodiment is the same as that in FIGS. 9 and 10 used in the description of the first embodiment.

  The flow of the encoding process of the encoding target image data by the image processing apparatus according to the second embodiment is almost the same as that in FIG. 5 used in the description of the first embodiment, but the process of step S505 is the same. The selector 1101 is different from the code amount weight calculation unit 1103 in reference to the code amount weight W.

  Further, according to the second embodiment, lossless encoded data is selected when the lossless encoded data amount falls below a predetermined threshold TH1. The reason for this is that the lossless encoded data is literally reversible, so image quality does not deteriorate, and when the code amount is small, the influence on the encoded data amount of the entire image is small. Therefore, the technique for providing this threshold value TH1 may also be applied to the first embodiment.

  As described above, according to the second embodiment, when an image is divided into tiles and reversible and irreversible data are selected and output for each tile, the tiles that are close to the target tile are already selected. The weighting coefficient is obtained with reference to the signal, and when reversible is selected for tiles adjacent vertically or horizontally to the target aisle, reversible is easily selected for the target tile, and the reversible and irreversible code amount antagonize. It is possible to suppress frequent switching of the encoding method in such a part. As a result, it is possible to reduce the visual disturbance caused by the adjacent reversible and irreversible tiles in both the vertical and horizontal directions.

In the above embodiment, the description has been made with reference to the selection signal on the left side of the target pixel and the selection signal of the tile in the same horizontal position of the immediately preceding stripe.
W = {S (i−1, j) + S (i−1, j−1) + S (i, j−1) + S (i + 1, j−1)} × M ′ (selection) The capacity of the signal holding unit 110 is increased by 1 bit). In this case, selection considering the oblique direction is performed.

[Third Embodiment]
In the image processing apparatuses according to the first and second embodiments, weighting is performed in such a manner that the lossless and irreversible code amount is increased or decreased by a predetermined number of bytes, but weighting is performed by multiplying the code amount by a predetermined coefficient. Can also be done. In the third embodiment, such an example will be described.

  FIG. 15 is a block diagram illustrating a functional configuration of the image processing apparatus according to the third embodiment.

  In the figure, reference numeral 1501 denotes a selector, and 1502 denotes a weight coefficient calculation unit. As shown in the figure, the image processing apparatus according to the third embodiment is almost the same as the image processing apparatus described in the second embodiment, and the selector 1101 replaces the selector 1501 and the code amount weight calculation unit 1103. Is different from the weight coefficient calculation unit 1502 only.

  When reversible tile coding data is selected, “1” is output from the signal line 113 as the selection signal S (i, j) of the tile T (i, j) of interest, and the lossy tile coding is performed. Similarly to the second embodiment, when data is selected, “0” is output.

  FIG. 17 is a flowchart showing the processing procedure of the selector 1501 in the third embodiment.

  First, the code amount CLK of the reversible tile encoded data stored in the lossless encoding buffer 106 is acquired (step S1701). Next, the code amount CLH of the irreversible tile encoded data stored in the irreversible encoding buffer 108 is acquired (step S1702). Subsequently, the code amount weight coefficient e (detailed later) input from the code amount weight calculation unit 1103 described later is multiplied by the reversible code amount CLK, and the result is set to CLK ′ (step S1703).

  Next, CLK ′ after multiplication is compared with a predetermined threshold value TH1 (step S1704). If CLK ′ <TH1 (YES), the process proceeds to step S1706. If CLK ′ ≧ TH1, the process proceeds to step S1705 (NO).

  In step S1705, CLK ′ and CLH after multiplication are compared (step S1705). If CLK ′ <CLH (YES), the process proceeds to step S1706, and if CLK ′ ≧ CLH (NO), the process proceeds to step S1708.

  In step S 1706, “1” is set in S (i, j) as a result of selecting the tile of interest, and this is output to the signal line 113. Subsequently, the reversible tile encoded data stored in the lossless encoding buffer 106 is read and output to the signal line 112 (step S1707).

  On the other hand, in step S1708, “0” is set to S (i, j), and this is output to the signal line 113. Subsequently, the irreversible tile encoded data stored in the irreversible encoding buffer 108 is read. Is output to the signal line 112 (step S1709).

  Here, the weighting coefficient e calculated in step S1703 is obtained as follows.

First, the weighting factor calculation unit 1502 selects the selection signal S (i, j-1) of the tile selection signal S (i, j-1) immediately above the tile T (i, j) of interest and the immediately preceding tile. The selection signal S (i−1, j) of T (i−1, j) is read from the buffer 1102, and the weighting coefficient value e is calculated and output by the following equation.
e = 1.0− {S (i−1, j) + S (i, j−1)} × E
Here, E is a predetermined positive constant.

  Therefore, when S (i-1, j) and S (i, j-1) are both 0, that is, both T (i-1, j) and T (i, j-1) are irreversible codes. E = 1.0 when the conversion is selected. When either one is lossless encoding and the other is lossy encoding, e = 1.0-E, and when both are lossless encoding, 1.0-2E. .

  Here, if E = 0.2 is defined, e is one of 1.0, 0.8, and 0.6, and either lossless encoded data or lossy encoded data in FIG. The boundary conditions to be selected are as shown in FIGS. 16 (a), (b), and (c), respectively.

  FIG. 16A shows the case where e = 1.0. At this time, if the determination in step S1705 is expressed using CHK before multiplication, it is determined whether or not “CLK <CHK” is true.

  FIG. 16B shows the case where e = 0.8. At this time, if the determination in step S1705 is expressed using CHK before multiplication, it is determined whether or not “CLK × 0.8 <CHK” is true. From FIG. Easy to select data.

  FIG. 16C shows the case where e = 0.6. At this time, step S1705 determines whether or not “CLK × 0.6 <CHK” is true, and it is possible to make the condition that lossless encoded data is more easily selected than in FIG. Become.

In the above embodiment, the example of multiplying CLK has been described. However, since it is a relative one, the reciprocal (1 / e) may be multiplied by CHK. Further, in the above embodiment, the weight coefficient value e is set to be smaller as the number of tiles that are losslessly encoded in the adjacent tile groups is larger, but on the contrary, lossless encoding is performed as in the following equation. The value may be increased as the number increases, and CHK may be multiplied by CHK (or the inverse thereof may be multiplied by CLK).
e = 1.0 + {S (i−1, j) + S (i, j−1)} × E

  That is, as the number of tiles that are losslessly encoded in the tile group adjacent to the tile of interest increases, the possibility that the lossless encoded data is selected as the encoded data of the tile of interest can be increased.

  Note that the code string forming unit 111 in the third embodiment forms a code string similar to that in the first and second embodiments, and outputs it from the signal line 114. The structure of the code string at that time is also the same as in FIGS. 9 and 10 used in the description of the first embodiment.

  As described above, according to the third embodiment, in an encoding method in which an image is divided into tiles and reversible or irreversible is selected and encoded for each tile, it is directly above the previous tile. The reversible and irreversible selection of tiles is weighted according to the reversible and irreversible code amount of the target tile. Can be easily selected, and frequent switching of the encoding method in a portion where the reversible and irreversible code amounts antagonize can be suppressed. As a result, it is possible to reduce the visual disturbance caused by the adjacent reversible and irreversible tiles in both the vertical and horizontal directions.

  Further, in the third embodiment, only the selection signal for the tile adjacent to and on the left of the tile of interest has been described. However, as described in the second embodiment, the determination condition is determined with reference to the oblique direction as well. It doesn't matter.

<Modification>
The present invention is not limited to the embodiments described above. For example, as an example of the lossless encoding method used in the lossless encoding unit 105, predictive conversion is performed using a fixed prediction formula and the prediction error is Huffman encoded. However, several prediction formulas are prepared and dynamically switched. Such a configuration may be adopted. Further, instead of the Huffman code, other entropy coding such as Golomb code or arithmetic code may be used. Furthermore, other lossless encoding methods such as JPEG lossless mode, JPEG-LS, and JPEG2000 may be applied. Further, although an example using the JPEG baseline method as the irreversible encoding method used in the irreversible encoding unit 107 has been shown, other irreversible encoding methods such as the JPEG extension method and JPEG2000 may be used.

  However, as described above, the encoding adopted for lossless encoding and lossy encoding is different from each other in the image areas that the encoding is good at, such as JPEG baseline and JPEG-LS. Is more convenient.

  In the embodiment, an example in which a comparison operation is performed in order to determine which lossless encoded data or lossy encoded data is selected as encoded data to be output from the tile of interest has been described. A selection table may be obtained by preparing a lookup table storing determination conditions (values of 1 or 0) as shown in FIG. 16 and supplying reversible and irreversible data amounts as addresses. .

  Further, as is clear from the above-described embodiment, the present invention is not limited to hardware, and it is clear that a computer program is included in its category. In general, a computer program can be realized by setting a computer-readable storage medium such as a CD-ROM in the computer and copying or installing it in the system. Obviously, it is within the scope of the present invention.

It is a block block diagram which shows the function structure of the image processing apparatus which concerns on 1st Embodiment. 3 is a block diagram illustrating a configuration of a lossless encoding unit 105. FIG. It is a figure which shows the position of the surrounding pixels a, b, and c with respect to the encoding object pixel x. 4 is a diagram illustrating an example of a Huffman table stored in a memory 204. FIG. It is a flowchart which shows the flow of the encoding process of the encoding target image data by the image processing apparatus which concerns on this embodiment. It is a flowchart which shows the flow of the code selection process of the selector in 1st Embodiment. It is a figure which shows the data structure of JPEG encoding data. It is a figure which shows the mode of the tile division | segmentation of the image in embodiment. It is a figure which shows the structure of the output code sequence of this image processing apparatus. It is a figure which shows the structure of the code sequence of one tile. It is a block block diagram which shows the function structure of the image processing apparatus which concerns on 2nd Embodiment. FIG. 10 is a diagram illustrating a storage state of a selection signal S in a selection signal buffer 1102. It is a flowchart which shows the flow of the code selection process of the selector in 2nd Embodiment. It is a figure which shows the relationship between the code amount CLK of lossless encoding data in the 2nd Embodiment, the code amount CLH of lossy encoding data, and the boundary condition selected by the selector 1101. It is a block block diagram which shows the function structure of the image processing apparatus which concerns on 3rd Embodiment. FIG. 10 is a diagram illustrating a relationship between a code amount CLK of lossless encoded data, a code amount CLH of lossy encoded data, and a boundary condition selected by a selector 1501 in the third embodiment. It is a flowchart which shows the flow of the code selection process of the selector in 2nd Embodiment. It is a figure which shows the basic composition of the image processing apparatus in embodiment.

Claims (9)

An image encoding device for encoding image data,
Dividing means for dividing the image data into pixel blocks of a predetermined size;
Lossless encoding means for losslessly encoding the pixel block and generating lossless encoded data;
Irreversible encoding means for irreversibly encoding the pixel block and generating irreversible encoded data;
Storage means for storing selection information indicating which lossless or lossy encoded data is selected as the encoded data of the pixel block;
The storage unit stores the code amount of the lossless encoded data generated by the lossless encoding unit corresponding to the pixel block of interest and the code amount of the lossy encoded data generated by the lossy encoding unit. Comparing means for weighting and comparing either one according to the selected information of the pixel block adjacent to the focused pixel block,
An image encoding apparatus comprising: selection output means for selecting and outputting encoded data of which the code amount is determined to be small as a result of the weighted comparison as encoded data of the pixel block of interest.   The image encoding apparatus according to claim 1, wherein the selection output means unconditionally selects the lossless encoded data when the amount of lossless encoded data of the pixel block of interest is equal to or less than a predetermined amount.   The comparison unit adds a predetermined positive value to the encoded data amount generated by the lossy encoding unit of the target pixel block when the selection result of the pixel block immediately before the target pixel block is lossless encoded data. Alternatively, a predetermined positive value is subtracted from the encoded data amount generated by the lossless encoding means, and the calculated lossy encoded data amount and the lossless encoded data amount are compared. The image coding apparatus according to claim 1 or 2. The positive value determined according to the already selected information of the pixel block adjacent to the target pixel block stored in the storage unit is α, the code amount generated by the lossless encoding unit of the target pixel block is CLK, When the code amount generated by the lossless encoding means is CLH,
The comparison means includes
CLK-α and CLH
Or
CLK and CLH + α
The image coding apparatus according to claim 1, wherein the image coding apparatuses are compared with each other. Code amount generated by the lossless encoding unit of the pixel block of interest, where α is a positive value of 1.0 or less determined according to the already selected information of the pixel block adjacent to the pixel block of interest stored in the storage unit Is CLK, and the code amount generated by the lossy encoding means is CLH,
The comparison means includes
CLK × α and CLH
Or
CLK and CLH x (1 / α)
The image coding apparatus according to claim 1, wherein the image coding apparatuses are compared with each other.   6. The image according to claim 1, wherein one of the lossless encoding unit and the lossy encoding unit is an encoding unit suitable for a natural image and the other for a character / line image. Encoding device. An image encoding method for encoding image data, comprising:
A division step of dividing the image data into pixel blocks of a predetermined size;
A lossless encoding step of losslessly encoding the pixel block to generate lossless encoded data;
An irreversible encoding step of irreversibly encoding the pixel block to generate irreversible encoded data;
A storage step of storing, in a predetermined storage unit, selection information indicating which lossless or lossy encoded data is selected as the encoded data of the pixel block;
The storage unit stores the code amount of the lossless encoded data generated in the lossless encoding step corresponding to the pixel block of interest and the code amount of the lossy encoded data generated in the lossy encoding step. A comparison step of weighting and comparing either one according to the selected information of the pixel block adjacent to the focused pixel block
An image encoding method comprising: a selection output step of selecting and outputting encoded data of which the code amount is determined to be small as a result of the weighted comparison as encoded data of the pixel block of interest. A computer program that functions as an image encoding device that encodes image data by being read and executed by a computer,
Dividing means for dividing the image data into pixel blocks of a predetermined size;
Lossless encoding means for losslessly encoding the pixel block and generating lossless encoded data;
Irreversible encoding means for irreversibly encoding the pixel block and generating irreversible encoded data;
Storage means for storing selection information indicating whether reversible or lossy encoded data is selected as the encoded data of the pixel block in a predetermined storage means;
The storage unit stores the code amount of the lossless encoded data generated by the lossless encoding unit corresponding to the pixel block of interest and the code amount of the lossy encoded data generated by the lossy encoding unit. Comparing means for weighting and comparing either one according to the selected information of the pixel block adjacent to the focused pixel block,
A computer program that functions as a selection output unit that selects and outputs encoded data of which the code amount is determined to be small as a result of the weighted comparison as encoded data of a pixel block of interest.   A computer-readable storage medium storing the computer program according to claim 8.

Images