JP2019047154A - Image coding apparatus and control method thereof - Google Patents

Abstract

Encoded data that guarantees the generation of an image with a constant image quality while ensuring that the amount of encoded data per block is equal to or less than the target code amount. Image data is divided into blocks composed of a plurality of pixels to generate lossless encoded data, and when the code amount of lossless encoded data exceeds a target code amount, the excess data is excluded. First lossy encoded data is generated as encoded data of the target block, and when the code amount of the lossless encoded data is equal to or smaller than the target code amount, the lossless encoded data is generated as encoded data of the target block. Each time encoded data less than the target code amount is generated in the block encoding by the encoding unit and the first encoding unit, the surplus data capacity for the target code amount is counted, and A code amount control unit that outputs the encoded data corresponding to the surplus data capacity in addition to the encoded data obtained by the first encoding unit when encoding by the first encoding unit is completed. [Selection] Figure 1

Description

  The present invention relates to an image data encoding technique.

  2. Description of the Related Art Conventionally, a fixed-length encoding technique is known that encodes image data so that the entire image or a predetermined code amount (or less) is determined for each predetermined unit. As a fixed-length encoding method, there is a method for realizing fixed-length encoding for each block by having an image feature ID for each block and a quantization value corresponding to the image feature (Patent Document 1). According to this method, appropriate quantization is performed on each block even for images with both high-resolution, low-gradation blocks and low-resolution, high-gradation blocks. Can be encoded.

  In addition, when the lossless encoding is within the target code amount for each predetermined unit, the lossless encoded data is output, and when the target code amount is exceeded, the code amount control is performed by performing quantization or truncation of the encoded data. There is also a method of performing (Patent Document 2). According to this method, code amount control can be performed while assigning an appropriate code length for each predetermined unit.

Japanese Patent Laid-Open No. 2006-63282 JP 2001-86502 A

  However, in patent document 1, it encodes with the same code length about all the blocks in an image. Therefore, in the case of a block with low entropy such that the block is composed of one color, an extra code is allocated. Conversely, for high entropy blocks such as complex textures, there is a possibility that image quality is excessively degraded. Patent Document 2 mentions a mechanism that does not reduce the code more than necessary for a predetermined unit with a small code amount, and a coefficient that greatly affects the image quality. It does not have a mechanism that can be guaranteed.

  The present invention has been made in view of the above problems, and is encoding that performs encoding in units of blocks, and ensures that the amount of encoded data per block is equal to or less than the target code amount and is constant. An object of the present invention is to provide a technique for generating encoded data that guarantees the generation of an image of high image quality.

In order to solve this problem, for example, an image encoding device of the present invention has the following configuration. That is,
An image encoding device for encoding an image, comprising:
A dividing means for dividing the image data into blocks composed of a plurality of pixels;
When lossless encoded data is generated from the block of interest obtained by the dividing means and the code amount of the lossless encoded data exceeds the target code amount, the excess data is stored in the memory and the excess data is removed. When the lossless encoded data is generated as encoded data of the block of interest and the code amount of the lossless encoded data is less than or equal to the target code amount, the lossless encoded data is converted to the code of the block of interest. First encoding means for generating as encoded data;
Second encoding means for encoding the excess data stored in the memory;
Counting means for counting an excess data capacity with respect to the target code amount each time encoded data less than the target code amount is generated in the block encoding by the first encoding unit;
Encoded data corresponding to the surplus data capacity counted by the counting unit among the encoded data generated by the second encoding unit when the encoding by the first encoding unit is completed for the image data Is added to the encoded data obtained by the first encoding means and output.

  According to the present invention, encoding is performed in units of blocks, and the amount of encoded data per block is ensured to be equal to or less than the target code amount, and the generation of an image with a constant image quality is guaranteed. Encoded data can be generated.

1 is a block diagram showing a configuration of an image encoding device according to a first embodiment. The flowchart which shows the process of the fixed length encoding part which concerns on 1st Embodiment. The flowchart which shows the process of the overbit encoding part which concerns on 1st Embodiment. The figure which shows the example of an encoding object block. The figure which shows the example of the encoding data of an encoding object block. The figure which shows the relationship between the encoding object block and the data length of encoding data. The figure which shows the relationship between an overblock and overbit coding data. The figure which shows the relationship between the image data of encoding object, and a block. The figure which shows the structure of the encoding data which concern on 1st Embodiment. The figure which shows the structure of the coding data which concern on 2nd Embodiment. The flowchart which shows the process of the image coding apparatus which concerns on 2nd Embodiment. The figure which shows the structure of the coding data which concern on 3rd Embodiment. The block diagram which shows the structure of the image coding apparatus which concerns on 4th Embodiment. The flowchart which shows the flow of the encoding process which concerns on 4th Embodiment. The figure which shows an example of subsampling. The figure which shows the example whose fluctuation | variation of complexity is flat. The figure which shows the example of the setting value of the fixed length for every block with respect to complexity. The figure which shows the relationship between ROI and a block. The figure which shows the example of the setting value of the fixed length of 5th Embodiment.

  Embodiments according to the present invention will be described below in detail with reference to the accompanying drawings. The following embodiments do not limit the present invention, and all the combinations of features described in the present embodiment are not necessarily essential to the solution means of the present invention. In addition, about the same structure, the same code | symbol is attached | subjected and demonstrated.

[First Embodiment]
In the first embodiment, when encoding an image, encoding that is not more than a fixed length in units of blocks and encoding of bits that exceed the fixed length are performed separately. When encoding at a fixed length or less, processing in bit units is not required, and encoding can be performed with a simple method. For encoding of bits exceeding the fixed length, the data is held without compression, and control is performed in units of bits at the time of code amount control. As a result, it is possible to control the amount of code so that the code length becomes a fixed length or less for each predetermined unit, while ensuring a certain level of image quality. A specific example will be described below.

  The image encoding apparatus in the embodiment will be described as being built in an imaging apparatus such as a digital camera. That is, the encoded counter image is an image obtained by the imaging unit included in the imaging device. In addition, you may apply to apparatuses other than an imaging device. Also, the image to be encoded may be stored in a storage medium or stored in a file server on the network. It is to be understood that the present invention is applied to the imaging apparatus in order to facilitate the understanding of the technology of the embodiment by being embodied, and is merely an example.

  FIG. 1 is a block diagram showing a configuration of an encoding process according to the first embodiment. As illustrated in FIG. 1, the encoding apparatus according to the present exemplary embodiment includes an input unit 101, an encoding unit 100, a code amount control unit 106, a code string forming unit 107, and an output unit 108. The encoding unit 100 includes a block dividing unit 102, a block encoding unit 103, an overbit encoding unit 104, and a buffer (memory) 105.

  The input unit 101 inputs image data captured by an image sensor (not shown) included in the digital camera. The input image is RAW image data composed of pixels in a Bayer array of R, G0, G1, and B, and one pixel (one component) is represented by 14 bits. Further, the input unit 101, from the input RAW image data, R plane composed only of R component pixels, G0 plane composed only of G0 component pixels, G1 plane composed only of G1 component pixels, Then, a B plane composed only of B component pixels is generated. The input unit 101 supplies the generated planes to the encoding unit 100. That is, the input unit 101 supplies each plane generated in four steps from one piece of RAW image data obtained by the image sensor to the encoding unit 100. The order of the planes to be generated is not particularly limited, but the order of R, G0, G1, and B planes is used.

  As described above, the encoding unit 100 according to the embodiment encodes the image data supplied from the input unit 101 as monochrome image data of one pixel and one component.

  In this embodiment, the image data to be encoded is RAW image data composed of pixels in a Bayer array. However, one pixel is an RGB component and each component is N-bit image data or other color representation. It can also be applied to the used image data. For example, for image data in which each RGB component is represented by 8 bits, it is only necessary to encode each color component of R, G, B as a monochrome image of 8 bits per pixel. A luminance / color difference component in which one pixel is represented by Y, Cb, and Cr and each component is represented by 8 bits may be encoded as an 8-bit monochrome image of each component of Y, Cb, and Cr. The above are only examples.

  The block division unit 102 in the encoding unit 100 divides the image data (any one of R, G0, G1, and B planes) input from the input unit 101 into rectangular blocks of horizontal Bw pixels and vertical Bh pixels. . For convenience of explanation, it is assumed that the number W of pixels arranged in the horizontal direction of the image to be encoded is an integer multiple of Bw. Similarly, pixel values H arranged in the vertical direction are also integer multiples of Bh, and incomplete blocks do not occur when divided into rectangular blocks. Hereinafter, the rectangular block (block of interest) composed of the horizontal Bw pixels and the vertical Bh pixels is referred to as block data or simply a block.

In the present embodiment, description will be made assuming that the size of one block is 4 × 4 pixels. FIG. 8 illustrates the relationship between image data to be encoded and blocks. As shown, the upper left corner of the image to be encoded is the origin. The horizontal i-th block and the vertical j-th block are denoted as B (i, j). Further, the pixel value at the coordinates (m, n) in the block B (i, j) is denoted as B _{i, j} (m, n). For example, the pixel at the upper left corner of block B (2, 3) is denoted as B _2,3 (0, 0).

  The block encoding unit 103 performs an encoding process for generating encoded data having a predetermined fixed code length or less from the pixel data for one block supplied from the block dividing unit 102. In describing the block encoding unit 103, a preset fixed length FL will be described first.

  The fixed length FL shown here can be said to be a parameter for controlling the maximum code length in block units. The encoded data of the block in the present embodiment allows a mixture of pixel data that is retained without being compressed and pixel data that is less than a fixed length and is represented by a variable length code. Then, the “fixed length FL” is defined to indicate the upper limit of the code length of the pixel data to be variable length encoded in the block.

  In the present embodiment, the fixed length FL is assumed to have the same value for all blocks. However, the fixed length FL is not limited to this and may be changed in units of blocks. However, it is necessary to set a fixed length FL for each block so as not to exceed the target code amount for the entire image. The fixed length FL may be input by a user from outside the apparatus (not shown), or a value may be set in the apparatus in advance.

  FIG. 2 is a flowchart illustrating processing contents from the encoding unit 100 to the code amount control unit 106 in the embodiment. Hereinafter, the processing procedure of the encoding unit 100 and the code amount control unit 106 will be described with reference to FIG. The figure shows a raster scan order encoding process in units of blocks. Therefore, the first block is block B (0, 0) in the upper left corner of the image data. The following is a description of the block of interest B (i, j) in the encoding process. In addition, prior to the processing of the first block B (0, 0), the variable SD (detailed later) is cleared to zero.

  In step S201, the block encoding unit 103 of the encoding unit 100 inputs the block of interest B (i, j) from the image data to be encoded, and searches for the minimum pixel Pmin and the maximum pixel Pmax therein. As shown in FIG. 4A, the pixels in the block are ranked from 0 to n in a predetermined scan order (for example, raster scan), and the pixel values are sequentially set to P0, P1,. Express. In the embodiment, since 1 block is 4 × 4 pixels, n = 15.

  FIG. 4B shows a specific example of pixel values of the pixels P1 to P15. In the case of the figure, the pixel position (order) of the minimum pixel Pmin is 0, and the pixel value is P0 (= 12). Further, in the same figure, the maximum Pmax pixel position is 7, and the pixel value is P7 (= 298). Incidentally, when there are a plurality of pixels having the minimum pixel value, the pixel that appears at the beginning of the scan is defined as Pmin. The same applies to the maximum pixel value Pmax. Further, each pixel other than the minimum maximum pixel is expressed as OPm as shown in FIG. In the embodiment, since 16 pixels are included in one block, and two of them are the minimum pixel Pmin and the maximum pixel Pmax, the number of pixels OPm is 14 (the value m can take is not 1) Is).

  Next, in step S202, the block encoding unit 103 encodes the minimum pixel Pmin and the maximum pixel Pmax of the block of interest B (i, j). In the present embodiment, the pixel position and pixel value of the minimum pixel Pmin, and the pixel position and pixel value of the maximum pixel Pmax are held as fixed-length code data. In the embodiment, since one block has 4 × 4 pixels (= 16 pixels), the “pixel position” can be expressed by 4 bits. Here, natural binary representation is used, but another binary representation such as alternating binary may be used as long as it can be uniquely decoded.

  In the example of FIG. 4B, since the position information of Pmin is 0, the position is “0000”, and the position information of Pmax is 7, so the position is expressed as “0111” (both are expressed in binary). The pixel value is 14 bits with no compression. However, although the pixel values of the maximum pixel Pmin and the maximum pixel Pmax are uncompressed, the minimum pixel value and the maximum pixel value can be set with less bits than the number of bits of the input pixel if some deterioration may be allowed. You can express it. The block encoding unit 103 stores the encoded data of the minimum pixel Pmin and the maximum pixel Pmax obtained from the block of interest B (i, j) in the buffer 105 shown in FIG.

  In step S203, the block encoding unit 103 calculates the block range BR of the target block B (i, j) to be encoded. The block range BR is defined as indicating the bit range of the difference value between the minimum pixel Pmin and the maximum pixel Pmax in the block. That is, the block encoding unit 103 obtains the minimum number of bits representing the difference value DIFF obtained by subtracting the value of the minimum pixel Pmin from the maximum pixel Pmax.

  For example, in the case of FIG. 4B, since Pmin = 12, Pmax = 298, the difference value DIFF is “286”. The minimum number of bits representing the value “286” is “9”. This means that the pixel OPm excluding the minimum pixel Pmin and the maximum pixel Pmax of the block of interest B (i, j) can be expressed by a 9-bit difference value with respect to the minimum pixel Pmin (or the maximum pixel Pmax). In the embodiment, since one pixel is 14 bits and the fixed length FL is 7, the maximum value of the block range BR is “14”.

  The relationship between the block range BR and the fixed length FL will be described below with reference to FIG. Reference numerals 61 to 63 in FIG. 6 indicate examples of three blocks having different characteristics. It is assumed that the maximum difference DIFF between the pixels in the block increases in the order of the blocks 61, 62, and 63. Here, it is assumed that the block range BR of the block 61 is “2”, the block range BR of the block 62 is “7”, and the block range of the block 63 is “14”. Now, assuming that the fixed length FL = 7, the relationship between the block range BR of each of the blocks 61, 62, and 63 and the fixed length FL is the reference numerals 61 ', 62', and 63 'shown at the bottom of the figure.

  Focusing on the block 61, the pixel OPm excluding the minimum pixel Pmin and the maximum pixel Pmax in the block 61 can be expressed by a 2-bit difference with respect to the minimum pixel Pmin. Therefore, 5 bits of the fixed length FL = 7 are surplus bits, and it is not necessary to generate a code amount correspondingly. This is because there are 14 pixels OPm in total. In the case of the block 61, this means that it can be expressed by encoded data that is 5 × 14 bits less than the target code amount of the pixel OPm of the block 61.

  Focusing on the block 62, since the block range BR and the fixed length FL are the same, the pixel OPm in the block 62 is represented by a difference by a fixed length FL bit with respect to the minimum pixel Pmin. That is, in the case of the block 62, the pixel OPm can be expressed by encoded data of the target code amount.

  On the other hand, in the case of the block 63, the block range BR = 14. That is, the block range BR is larger than the fixed length FL. Therefore, the block encoding unit 103 generates encoded data of upper FL bits that are dominant as the pixel value of the pixel OPm. Then, the block encoding unit 103 supplies the low-order BP (= BR-FL) bits (exactly BP × 14 bits) exceeding the FL bits to the overbit encoding unit 104 and encodes them (described later). ).

  As will be apparent from the description below, comparing BR and FL determines whether the block encoding unit 103 can generate lossless encoded data having a target code amount or less for the block of interest. Is equivalent to

  Returning to the flowchart of FIG. In step S204, the block encoding unit 103 compares the block range BR of the block of interest B (i, j) calculated in step S203 with “0”. When the value of the block range BR is larger than 0 (Yes), the block encoding unit 103 shifts the process to step S205 and encodes the pixels OPm other than the minimum / maximum pixels. On the other hand, when the value of the block range BR is 0 (No), it means that all the pixels in the block of interest have the same pixel value. Therefore, the block encoding unit 103 does not need to encode the pixels OPm other than the minimum pixel Pmin and the maximum pixel Pmax. Therefore, the block encoding unit 103 skips the encoding process (steps S205 to S206) of the pixel OPm, and proceeds to step S207.

  In step S205, the block encoding unit 103 encodes pixels OPm (= OP1, OP2,..., OP14) other than the minimum pixel Pmin and the maximum pixel Pmix with a fixed length FL or less. The following is a description of the encoding.

  When the block range BR of the target block B (i, j) to be encoded is equal to or less than the fixed length FL, the block encoding unit 103 subtracts the pixel value of the minimum pixel Pmin from the value of the pixel OPm (OP1 to OP14). The value Qm (Q1 to Q14) is output as encoded data. Of course, each value Qm is the number of bits represented by the block range BR.

  On the other hand, when the block range BR of the block of interest is larger than the fixed length FL, the block encoding unit 103 subtracts the value Qm (Q1 to Q14) from the value of the pixel OPm (OP1 to OP14) by the pixel value Pmin of the minimum pixel. ) Is generated with an accuracy of the bit number BR. The lower bits are discarded so that this Qm can be expressed by a fixed length FL, and encoded data of a fixed length FL of each pixel OPm is generated. In other words, the block encoding unit 103 outputs the upper FL bits of the value Qm calculated with the accuracy of the bit number BR as encoded data of the pixel of interest OPm. Here, for easy understanding, a specific example in which the block of interest is the block shown in FIG.

  In the present embodiment, the fixed length FL = 7. The block range BR of the block shown in FIG. 4B is “9”. That is, the number of over bits is “2”. Therefore, the block encoding unit 103 generates the upper 7 bits of Qm (9 bits) obtained by subtracting the pixel value Pmin of the minimum pixel from the pixel OPm as the pixel OPm in the block of interest as fixed-length encoded data. . That is, the block encoding unit 103 generates the remaining 7 bits as fixed-length encoded data by truncating the lower 2 bits of Qm (9 bits).

  For example, the value of the pixel OP1 in FIG. 4B is “64”. Since the pixel value Pmin of the minimum pixel is “12”, the value Q1 representing the difference of the pixel OP1 with respect to the pixel value of the minimum pixel is “52”. Therefore, the block encoding unit 103 shifts the value Q1 to the right by 2 bits to generate “13” (“0001101” in natural binary notation). For other Q2 to Q14, the difference from the pixel value of the minimum pixel can be shifted to the right by 2 bits to generate encoded data.

  Here, each value of the encoded data of the pixel OPm excluding the minimum pixel Pmin and the maximum pixel Pmax of the block shown in FIG. 4B is as shown in FIG.

  In step S206, the block encoding unit 103 compares the block range BR of the block of interest B (i, j) with a preset fixed length FL. As a result of the comparison, if BR is equal to or less than FL (Yes), the process proceeds to step S207. On the other hand, when BR is a value larger than FL (No), the block coding unit 103 shifts the process to step S208.

  In step S207, the block encoding unit 103 updates the surplus data capacity SD from the first block B (0, 0) to the target block B (i, j). The surplus data capacity SD is an index value that represents an unused capacity that occurs when the block range BR is smaller than the fixed length FL. The surplus data capacity of the block of interest B (i, j) can be calculated as FL-BR. In the case of the block 61 in FIG. 6, the surplus data capacity has a value of FL−BR = 7−2 = 5.

  Since the block of interest B (i, j) processed in step S207 has a block range BR ≦ fixed length FL, it can be said that it is a block that can be losslessly encoded with a fixed length FL or less. Further, in the case of BR <FL, it means a block that can be losslessly encoded without using all of FL. Therefore, when there is a block whose surplus data capacity SD is 0 or more, encoded data of another block that exceeds the fixed length FL can be allocated.

  The total surplus data capacity SD from the first block B (0, 0) to the target block B (i, j) in the target image to be encoded is hereinafter referred to as a pool capacity SD. This pool capacity SD is calculated by the bit length FL-BR when the block range BR is equal to or less than the fixed length FL among all the blocks from the first block B (0, 0) to the target block B (i, j). Is the sum of the values.

  In the present embodiment, it is only necessary to know how many overbits there are in one block. Therefore, the pool capacity SD as described above is defined, but the present invention is not limited to this. For example, the surplus data capacity for an actual block has a surplus of (FL-BR) bits for 14 pixels other than the minimum pixel and the maximum pixel. Therefore, the surplus data amount of one block is exactly (FL-BR) × 14 bits. Therefore, the total of the surplus data capacity from the first block B (0, 0) to the target block B (i, j) may be used as the pool capacity SD.

  In step S208, the overbit encoding unit 104 encodes bit data exceeding the fixed length FL (referred to as overbits) for the target block for which BR> FL. The encoding process of the overbit encoding unit 104 will be described with reference to the flowchart of FIG. This figure is a flowchart showing the flow of the encoding process of the overbit encoding unit 103 of the embodiment.

  First, in step S301, the overbit encoding unit 103 inputs an encoding target block. Note that this encoding target block is a block having a block range BR larger than the fixed length FL. In the present embodiment, the block input in step S301 is overblock OB, and the total number of overblocks after the start of encoding of the image of interest is represented as bn.

Subsequently, in step S302, the overbit encoding unit 103 calculates the overflowing bit number BP (hereinafter referred to as the overbit number BP) of the processing target block OB by the following equation (1).
BP = BR-FL (1)
Then, the overbit encoding unit 103 initializes the counter C with zero in step S303. The possible range of the counter C is 0 to “BP-1”. Then, the overbit encoding unit 103 refers to the counter C, and repeats the processes of steps S304 to S306 described below for the overblock OB to be processed by the number of overbits BP.

  The processing in steps S304 to S306 will be briefly described. The lower BP bits (BP = BR-FL) of the value Qm (Q1 to Q14) of the pixel OPm in the overblock are divided and stored in one or more bit layers. It is processing to do.

  Of the lower BP bits in the value Qm, the bit string of the highest bit (14 bits in the embodiment) is stored and managed in the additional data EL (0). Of the lower BP bits, the bit string of the second highest bit is stored and managed as additional data EL (1). The same applies hereinafter.

  In other words, the bit string of the FL + 1th bit (the 8th bit from the most significant bit) from the most significant bit position of the value Qm is stored and managed in the additional data EL (0). It can also be said that the FL + 2 bit (9th bit from the most significant bit) bit string from the most significant bit position of the value Qm is stored and managed in the additional data EL (1).

  Thus, by storing and managing the overflowed bits of each overblock as a bit string layer as additional data EL (C), the reproduction accuracy of the pixel value when each overblock is decoded is improved. If all overblocks of the target image to be encoded can be held as all additional data EL (C), the target image can be completely decoded. That is, lossless encoding can be realized. It is assumed that the additional data EL () is initialized in a state where nothing is stored prior to the encoding of the image of interest.

  In step S304, overbit encoding section 104 updates additional data EL (C).

In the case of FIG. 5, it is assumed that C = 0. In this case, the overbit encoding unit 104 additionally stores the bit string (14 bits) “010111110011111” positioned in the upper two of the lower two bits of Qm in the additional data EL (0).
EL (0) ← EL (0) + “010111110011111”
As a result, a bit string of 1 bit out of 2 bits rounded down in the above-described fixed length encoding process is held in the additional data EL (0).

  When the process of step S304 is completed, the overbit encoding unit 104 advances the process to step S305, and increments the counter C by “1”. In step S306, the overbit encoding unit 104 compares the counter C with the overbit number BP of the target overblock. If C is smaller than BP as a result of the comparison (YES), the process returns to step S304.

  Here, in the case of FIG. 5, it is assumed that the counter C is “1”. Since BP = 2, it is determined that C <BP, and the overbit encoding unit 104 returns the process to step S304.

Then, overbit encoding section 104 additionally stores the second highest bit string (14 bits) “0101010010001111” in the lower 2 bits of Qm in additional data EL (1) in the second step S304. .
EL (1) ← EL (1) + “010100100001111”
Then, in step S305, the counter C is incremented by “1” to become “2”, and the inequality in step S306 is not satisfied, so this processing ends.

  As described above, when the block of interest is FIG. 4B, BP = 2, so the additional data to be updated is EL (0), (1), and the other EL (2) to (6) ) Is not updated. The additional data EL (0) to (6) is updated by performing the above process every time an overblock occurs.

  FIG. 7 is a diagram illustrating the relationship between the overblock OBn and the overbit encoded data. For each of the overblocks OB0 to OB7, BP is 1, 2, 3, 4, 5, 5, 4 in order. The numbers in the squares in the figure indicate how many bits from the most significant bit of the value Q of each overblock are stored. For example, the additional data EL (0) indicates that the bit string of the eighth bit from the most significant bit of the value Q in the overblock is stored. Further, the additional data EL (1) indicates that the 9th bit string from the most significant bit of the value Q in the overblock is stored. However, this storage method is not limited to this. For example, overbit encoded data may be stored following the fixed length encoded data of each block. In the present embodiment, the method of managing overbit encoded data in units of layers has been described, but it may be managed for each layer for each block. In this case, the method is not limited to the above method, and it is sufficient to add the data in order from the additional data with high pixel value reproduction accuracy.

  The rightmost LD () in FIG. 7 indicates the amount of data stored in the additional data EL (). For example, LD (0) = 7 indicates that the additional data EL (0) stores 7 overblock bit strings (= 14 × 7 bits).

  With the above method, the overbit coding unit 104 generates overblock overbit coded data.

  Returning to the flowchart of FIG. When the process of step S207 or step S208 ends, the encoding unit 100 advances the process to step S209. In step S209, the encoding unit 100 determines whether or not the target block to be encoded is the last block. If the target block is the last block (Yes), the process proceeds to S210. If the block is not the last block (No), the encoding unit 100 returns the process to step S201 and performs the encoding process for the next block.

  Step S210 and subsequent steps are processing of the code amount control unit 106. At this point, note that the buffer 105 stores the encoded data generated by the block encoding unit 103 for the target image to be encoded and the encoded data generated by the overbit encoding unit 104. I want to be.

  In addition, although the description is around, the encoded data of one block generated by the block encoding unit 103 is encoded data of the minimum pixel and its position, the maximum pixel and its position, and the pixel OPm. Among these, the minimum pixel and its position, and the position of the maximum pixel have a fixed length of 14 + 4 + 14 + 4 = 36 bits. The pixel OPm does not exceed the fixed length FL (= 7), and the number of pixels OPm is 14.

Therefore, the block target code amount Bmax for one block is 134 bits as shown in the following equation.
Bmax = 14 + 4 + 14 + 4 + 7 × 14 = 134
When the total number of blocks included in the target image is B _N , the target code amount T of the target image in the embodiment is
T = 134 × B _N
And Since Bmax is the maximum encoding size of one block generated by the block encoding unit 103, it is hereinafter referred to as a block maximum code amount.

  In step S210, the code amount control unit 106 initializes the counter C to zero. The counter C is a variable for specifying the additional data EL ().

  Subsequently, in step S211, the code amount control unit 106 determines whether or not the data amount LD (C) of the additional data EL (C) is equal to or less than the pool capacity SD. The data amount LD (C) of the additional data EL (C) is a value representing the number of blocks extracted from the number of blocks stored in the additional data EL (C) as described above with reference to FIG. Is shown. Referring again to FIG. 7, “LD (0) = 7” indicates that “additional data EL (0) contains 7 overblock bit strings”. Similarly, “LD (1) = 6” indicates that “additional data EL (1) stores six overblock bit strings”.

  If LD (C) ≦ SD (Yes), it means that even if the additional data EL (C) is added to the block encoded data, the total code amount is equal to or less than the target code amount T. Therefore, the code amount control unit 106 determines that the additional data EL (C) is included in the encoded data. In step S212, the code amount control unit 106 subtracts LD (C) from the pool capacity SD. Then, in step S213, the code amount control unit 106 increases the counter C by “1”.

  Thereafter, the code amount control unit 106 proceeds to step S215, and compares the counter C with “N-FL”. Here, N is the number of bits represented by one pixel of the image of interest, and is “14” in the embodiment.

  Therefore, if it is determined in step S215 that C <“N-FL”, the process returns to step S211. In this way, the code amount control unit 106 repeats the processes in steps S211 to S213 until the determination in step S211 or step S215 is No.

  Now, in step S211, the code amount control unit 106 determines that LD (C) ≦ SD is not satisfied. This means that if the additional data EL (C) is output as encoded data, the target code amount T is exceeded. Therefore, the code amount control unit 106 advances the processing to step S214 and discards the additional data EL (C) and subsequent data. That is, the code amount control unit 106 determines the additional data EL (0) to EL (C-1) as valid data. The generated encoded data is irreversible encoded data. However, since the discarded data is a low-order bit used when reproducing the pixel value, the influence on the image quality can be reduced. Hereinafter, the number of effective additional data layers is defined as Anum. The code amount control unit 106 converts the “encoded data generated by the block encoding unit 103” and the “additional data EL (0) to (Anum-1) generated by the overbit encoding unit 104” into the target image (attention (Color component image) is determined as encoded data and supplied to the code string forming unit 107.

  As a result, it is promised that the total code amount of the image of interest is equal to or less than the target code amount T. The code string forming unit 107 generates a header including information necessary for image decoding (including the Anum of each color component), and subsequently converts the encoded data for each color component into a preset format. Output to the output unit 108. The output unit 108 performs processing for storing the file as a file in a storage medium (not shown).

  On the other hand, it is assumed that the code amount control unit 106 determines in step S215 that C <“N−FL” is not satisfied. This means that even if the encoded data generated by the block encoding unit 103 and the encoded data represented by all the additional data EL () are added, the target code amount T or less is obtained. Therefore, the code amount control unit 106 determines the total number of layers of the additional data EL () generated by the overbit encoding unit 104 as Anum. The code amount control unit 106 supplies the encoded data generated by the block encoding unit 103 and the additional data EL (0) to (Anum-1) stored in the buffer 105 to the code string forming unit 107. . Here, the point of interest means that the encoded data delivered to the code string forming unit 107 when No in S215 is encoded data that can completely restore the image of interest. In other words, it means that the lossless encoded data having the target code amount or less has been successfully generated.

  As described above, according to the above embodiment, the maximum code amount per block is 134 bits. In the case of no compression, the data amount of one block is 16 pixels × 14 bits = 224 bits. Therefore, according to the above embodiment, it means that it becomes possible to generate a code amount of 5/8 or less with respect to the original image.

  FIG. 9 is a diagram showing the configuration of the output code string of this embodiment. 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 block, the height, etc. Additional information is attached as a header. Following the header is encoded data of color components. The encoded data of one color component is added with encoded data of a fixed length or less of each block in the raster scan order. Thereafter, a code string is formed with a configuration in which additional data EL (0) to EL (Anum-1) of overbit encoded data is added.

  As shown in FIG. 9, in the present embodiment, the encoded data of one block is arranged with the minimum pixel position, the minimum pixel value, the maximum pixel position, and the maximum pixel value from the top. Subsequently, encoded data having a fixed length or less is arranged for the pixels OPm other than the minimum maximum pixel. At this time, if the difference between the minimum pixel and the maximum pixel is 0, it means that all the pixels in the block have the same value, and therefore encoded data of other pixel values OPm is not generated or added. . Following the encoded data having a fixed length or less, overbit encoded data (additional data EL ()) is added. In the present embodiment, the overbit encoded data has encoded data from additional data EL (0) to additional data EL (Anum-1). According to this configuration, the block range of all blocks in the image can be specified when the decoding apparatus decodes the encoded data having a fixed length or less. Therefore, the decoding apparatus can specify the position of the overblock and the corresponding data position of the overbit encoded data. Therefore, additional data can be decoded without requiring additional information. However, the code string is not limited to this. For example, fixed-length encoded data and overbit encoded data may be held in units of blocks. In this case, decoding can be performed for each block, but it is not possible to specify how many bits of additional data are present when the blocks are sequentially decoded. Therefore, it is necessary to add information for each overblock so that the added data amount can be known.

  In the above-described embodiment, it has been described that the encoded data of one block includes information indicating the values and positions of the minimum pixel Pmin and the maximum pixel Pmax in the block without encoding. However, this does not limit the present invention. For example, if the block range BR of the processing target block is known, either the minimum pixel Pmin or the maximum pixel Pmax is determined as the representative pixel value, and the pixel value and position of the representative pixel are included in the encoded data. You may do it. In the case of the embodiment, the block range BR can take values from 0 to 14, so that 4 bits are sufficient. In this case, the encoded data of one block is encoded data of the pixel value of the representative pixel (either the minimum pixel Pmin or the maximum pixel Pmax), the pixel position, the block range BR, and the fixed length of the pixel OP. .

Therefore, the block maximum code amount Bmax is
Bmax = 14 + 4 + 4 + 7 × 15 = 127 bits, and 7 bits can be reduced per block as compared with the above embodiment. In the above, the reason why the fixed length FL (= 7) is multiplied by “15” instead of “14” is that the maximum pixel Pmax (or the minimum pixel Pmin) is an encoding target having a fixed length or less. .

  With the configuration as described above, the image encoding apparatus of the present embodiment can generate encoded data with a high image quality and less than or equal to the target code amount.

  In the above embodiment, an example in which the block size of the encoding unit is 4 × 4 pixels, one pixel is 14 bits, and the fixed length FL = 7 has been described, but the present invention is limited by these numerical values. These values are not intended, and these numerical values may be appropriately set by the user.

[Second Embodiment]
In the first embodiment described above, fixed-length encoding is performed on the entire surface of the image while generating a code having a fixed length or less and a code exceeding the fixed length in units of 4 × 4 pixels, that is, 16 pixels. An example of generating encoded data equal to or less than the target code amount T) has been described. In the case where there is a limit to the buffer capacity that is temporarily held in order to form an output code string, a configuration may be adopted in which fixed-length encoding is performed not on the entire image but on a limited area. In the second embodiment, an example will be described in which an image is divided into small areas composed of a plurality of blocks, and encoding is performed with a fixed length or less in units of the small areas. Hereinafter, an area composed of a plurality of blocks to be encoded with a fixed length or less is expressed as “TR”.

  The processing of the second embodiment will be briefly described with a focus on differences in operation from the first embodiment. Also in the second embodiment, it is assumed that the image to be encoded is an image composed of pixels of a Bayer array, and each component of R, G0, G1, and B is composed of 14 bits. . However, the encoding target image may be an image in which one pixel is composed of a plurality of components such as R, G, and B. As in the first embodiment, the input unit 101 generates monochrome image data including only pixels of one color component from input image data and supplies the monochrome image data to the encoding unit 100. Therefore, the encoding unit 100 encodes a monochrome image in which one pixel is represented by 14 bits.

  FIG. 11 is a flowchart showing the flow of the encoding process in the present embodiment. Hereinafter, the processing from the encoding unit 100 to the code amount control unit 106 will be described with reference to the flowchart shown in FIG.

  First, in step S1101, the encoding unit 100 sets the number of blocks bn included in the region TR to be fixed-length processed so as to be within the specified buffer capacity. It is assumed that the number of blocks bn included in TR takes at least one value. Further, in this embodiment, to simplify the description, it is assumed that the total number of blocks for the entire image is an integral multiple of the number of blocks bn in TR.

In step S1102, encoding unit 100 calculates number RM of regions TR according to the following equation.
RM = total number of blocks in image / bn
In addition, the encoding unit 100 initializes a counter k for counting the number of processed regions TR with 0.

  In step S1103, encoding unit 100 encodes region of interest TR. This encoding process is the same as steps S201 to S209 in FIG. The difference is that the processing target in FIG. 2 is the region TR, and whether or not the block of interest in step S209 in the flowchart shown in FIG. 2 is the last block in the region. That is, the final block here means the bn-th block in the region TR. If it is determined that the encoding process (step S1103) has been completed for all the blocks in the region TR (YES in FIG. 2, step S209), the process proceeds to step S1104.

  In step S1104, the code amount control processing by the code amount control unit 106 is performed for the region TR to be processed. This code amount control process is substantially the same as steps S210 to S215 in FIG. That is, the code amount control unit 106 compares the pool capacity SD calculated from the fixed-length encoding result for the number of blocks bn and the overbit encoded data amount, and the processing target TR is equal to or less than the target code amount. The code amount control is performed so that the code amount becomes. In the present embodiment, the target code amount for TR is set to a value obtained by dividing the target code amount of the entire image by the number of TR regions k. The target code amount for TR is not limited to this, and may be changed for each region depending on the feature amount of the image. However, it is necessary to set so as not to exceed the target code amount for the entire image. In addition, in this embodiment, the same number of blocks is assigned to all TRs in the present embodiment. However, the number of blocks is not limited to this, and the number of blocks may be changed for each region.

  In step S1105, the encoding unit 100 compares the counter k with the number of regions RM. If k <RM (YES), in order to repeat the encoding process of region TR, in step S1106, counter k is incremented, and the next region TR is set as the processing target. On the other hand, when the encoding process for the number of RMs after completion is performed (No), the process ends.

  When the encoding process and the code amount control process are completed for all regions TR, the code string forming unit 107 forms a code string. FIG. 10 is a diagram illustrating a configuration of an output code string in the second embodiment. Similar to the first embodiment, a header including image information necessary for decoding and the subsequent encoded data of each color component are arranged at the head of the code string. The encoded data of one color component has fixed length encoded data and overbit encoded data for each region TR. However, the code string is not limited to this. In the present embodiment, since encoding is performed using a predetermined plurality of blocks in the image as a fixed length unit, a code string may be formed and output each time the region TR to be processed is encoded. In this case, the header information may be output separately from the encoded data, or may be added as a header to the encoded data of each TR. By outputting the encoded data of each TR individually, it is possible to output the encoded data with a low delay compared to the case where the fixed length encoding is performed on the entire image.

  In the present embodiment, the method of setting the fixed length unit to a plurality of predetermined blocks according to the buffer capacity has been described, but the present invention is not limited to this. The code amount may be controlled by managing encoded data having a fixed length or less and overbit encoded data so that all predetermined buffer capacities can be used. However, in that case, additional information is required for the code amount control unit, and control during decoding may be complicated.

  As described above, by performing the processing control described above, it is possible to output fixed-length encoded data with low delay compared to the case where fixed-length encoding is performed on the entire image by performing fixed length in a predetermined unit. .

[Third Embodiment]
In the first embodiment and the second embodiment, when the code string is formed, all the fixed-length encoded data in the area of the code amount control unit are arranged as a code string, and at the end of the fixed-length encoded data. In this example, the code string is formed by adding overbit encoded data. In this case, when the fixed-length encoded data and the overbit encoded data are arranged for the code amount control unit, the fixed-length data group is completed and transmission of the encoded data can be started. In other words, the encoded data cannot be transmitted unless the fixed-length encoded data and the overbit encoded data are prepared.

  In the present embodiment, a method for forming a code string that enables transmission of encoded data at shorter processing times will be described.

  Since the encoding process itself is the same process as in the first and second embodiments, the description thereof is omitted. The main process in the present embodiment is in the code string forming unit 107. An example of a code string generated by the code string forming unit 107 is shown in FIG. 12, and a method for forming the code string will be described below.

  First, in the present embodiment, a unit of fixed-length encoded data for transmitting encoded data at a fixed timing is referred to as a fixed-length packet. The data length of the fixed-length packet may be set to a value equal to or larger than the upper limit of the code length of the fixed-length encoded data.

  In the third embodiment, the code string forming unit 107 generates a fixed-length packet for the fixed-length encoded data of each block generated for the region TR0 of the code amount control unit to be processed first, Transmission is performed via the output unit 108. FIG. 12 is an example of an output code string. Here, the head block B (0, 0) is a block in which an overbit is generated, and the subsequent B (1, 0) is a block encoded with a fixed length or less without an overbit. As shown in FIG. 12, the code string forming unit 107 generates a fixed-length packet as it is from the encoded data of the head block B (0, 0) of the image and transmits it through the output unit 108. Subsequently, since the encoded data of block B (1, 0) is shorter than the data length of the fixed-length packet, the code string forming unit 107 adds dummy data so as to be the data length of the fixed-length packet. Yes. The data is transmitted via the output unit 108. Note that the overbit encoded data in the region TR0, generated in parallel with the fixed length encoded data in the region TR0, is stored in the buffer 105 during the generation of the encoded packet. It is not output at the stage.

  Subsequently, fixed length coding of the region TR1 is performed. After generating fixed length encoded data for the encoding target block B (i, j) in TR1, in order to generate a fixed length packet, the fixed length encoded data of the target block is overwritten with the previous region TR0. Encoded data for an additional data length is added from the bit encoded data. At the time of decoding, if the data length of the fixed-length packet is known, the data length added to the fixed-length packet to be decoded can be specified when the fixed-length encoded data of B (i, j) is decoded. Transmission is started when a fixed-length packet for the encoding target block is generated. If the data length of the fixed-length encoded data of the encoding target block B (i, j) and the additional data of the area TR0 is less than the data length of the fixed-length packet, dummy data is added. Then, it may be adjusted so as to be a fixed-length packet. In parallel, the overbit encoded data of the region TR1 is generated and stored.

  As for the overbit encoded data of the code amount control region TR (RM-1) to be processed last in the image, a fixed-length packet may be generated and transmitted so that all data can be added. In addition, it is assumed that overbit encoded data corresponding to the number of fixed-length packets stored in advance may be added and transmitted.

  As described above, by performing the processing control described above, it is possible to generate a fixed-length packet for each block encoding while realizing code amount control in a wider range than performing fixed-length encoding in units of blocks. The encoded data can be transmitted at the timing.

[Fourth Embodiment]
In the embodiments described so far, the method of performing encoding by setting one fixed length FL for an image has been described. The fixed length FL may acquire information (hereinafter referred to as recognition information) representing characteristics for each region of the image, and set a fixed length FL value independently for each block based on the information. In the fourth embodiment, a method of setting a fixed length FL for each block using the complexity of the encoding target image as recognition information will be described. It is assumed that the encoding target image data is a monochrome image and is represented by 14 bits per pixel, for example.

  FIG. 13 is a block diagram illustrating a functional configuration of the image processing apparatus according to the fourth embodiment. As shown in FIG. 13, the apparatus according to the fourth embodiment has a configuration in which a recognition information acquisition unit 1301 and a fixed length setting unit 1302 are added to the functional configuration of FIG. Therefore, hereinafter, the flow of processing of the recognition information acquisition unit 1301 and the fixed length setting unit 1302 will be mainly described below.

  In the present embodiment, the recognition information acquisition unit 1301 performs processing in two passes: recognition information acquisition processing from an encoding target image and encoding processing.

  FIG. 14 shows a processing flow in the case of setting the fixed length FL based on the second pass encoding process, that is, the recognition information acquired in the first pass.

  First, in step S1401, the input unit 101 inputs image data to be encoded. In the processing flow shown in FIG. 14, the processing flow is such that the recognition information is acquired at a timing different from the image input. However, the present invention is not limited to this, and the recognition information may be input simultaneously with the image input.

  In step S1402, the block dividing unit 102 divides the input encoding target image into blocks.

  Subsequently, in step S1403, the recognition information acquisition unit 1301 acquires the recognition information (details will be described later).

  In step S1404, the fixed length setting unit 1302 determines the fixed length FL of the target block to be encoded from the acquired recognition information. Details of the method of setting the fixed length FL according to the recognition information will be described later.

  In step S1405, the encoding unit 100 encodes the block of interest using the fixed length FL set in step S1404. When the encoding process is completed for all the blocks, the code amount control unit 106 performs code amount control processing in step S1406, and the code string forming unit 107 forms a code string. The formed encoded data is output by the output unit 108 in step S1407, and the encoding process ends.

  Here, the fixed length setting method corresponding to the recognition information in step S1404 will be described below. In the fourth embodiment, the complexity of the image is input as recognition information. Here, the complexity of an image is information that can predict whether or not a large amount of code is generated by the encoding process described in the embodiment. As can be understood from the above description, the block encoding unit 103 searches for the minimum pixel and the maximum pixel in the block, and encodes pixels other than the minimum pixel and the maximum pixel with a fixed length FL or less. In this case, the larger the value of the difference DIFF between the minimum pixel and the maximum pixel, the more data that exceeds the fixed length FL. On the other hand, the smaller the difference value DIFF between the minimum pixel and the maximum pixel, the shorter the code length than the fixed length FL. From this characteristic, the magnitude of the difference in pixel value for each area of the image can be used as an index value representing complexity. For example, as a method for obtaining complexity, it is assumed that processing is performed on a reduced image of an encoding target image in order to easily realize recognition processing. As a generation method of the reduced image, the sub-sampled image may be generated by thinning out the pixel values at regular intervals, or may be generated by taking an average value of a plurality of pixels. In the present embodiment, since one block is 4 × 4 pixels, a method of acquiring complexity when a reduced image is generated by taking one pixel of 4 × 4 pixels will be described.

  As shown in FIG. 15, when the upper left pixel of 4 × 4 pixels is a pixel of a reduced image (hereinafter referred to as a reduced pixel), when the block is composed of 16 pixels of 4 × 4, The correct block range cannot be specified. Therefore, the difference between adjacent reduced pixels is calculated, and the number of bits necessary to express the difference value is set as the predicted value of the block range. The predicted value of the block range is acquired as the complexity Cp in units of blocks. For example, if the difference value between the reduced pixel of interest and the adjacent reduced pixel is 0, the complexity Cp is set to 0. When the pixel is 14 bits, the pixel value can take a value of 0 to 16383. Therefore, since the number of bits that can represent this is 0 to 14 bits, the complexity Cp may be 0 to 14. The calculation method of the complexity Cp and possible values are not limited to this, and it is only necessary to obtain data that can set a fixed length for each block by predicting the block range.

  The fixed length FL for each block can be set from the acquired complexity Cp and the target code amount for the entire image.

  It is assumed that the target code amount and the basic fixed length FL for the entire image are determined in advance. In this embodiment, the basic fixed length is expressed as FL, the fixed length set for each block is expressed as VFL, and the predetermined basic fixed length FL is 7. At this time, as in the examples shown in FIGS. 16A, 16B, and 16C, when the variation in the value of the complexity Cp for all the blocks in the image is small and almost constant, The fixed length VFL setting process is omitted, and the encoding process is performed using the basic fixed length FL as it is.

  On the other hand, when the variation of the complexity Cp value for each block is large, a fixed length VFL is set for each block. In the present embodiment, the VFL sets a value obtained by subtracting a certain number of bits from the complexity Cp. However, it is necessary to set the fixed length VFL of each block so that the upper limit is the data length when the basic fixed length FL is applied to all blocks. That is, the fixed length VFL for each block is set so as to be equal to or less than the target code amount for the entire image.

  FIGS. 17 (a) and 17 (b) show conceptual diagrams of changes in the set value of the fixed length VFL for each block. The fixed length VFL for each block is set so that the area of the shaded portion in FIG. 17B is equal to or smaller than the area of the shaded portion in FIG. Each block is encoded with the fixed length VFL for each block set by the above method.

  When the actual block range BR is larger than the predicted value at the time of encoding, correction processing such as increasing the priority of the overbit encoded data of the target block by the difference from the predicted value is performed. By doing so, it is possible to control the degree of image quality degradation with higher accuracy. In addition, by setting the priority of the overblock encoded data high even for a block having a small block range BR, there is a high possibility that losslessness can be achieved during code amount control.

  When the fourth embodiment is implemented, a block header is provided in the encoded data of the block, and the fixed length FL used by the block encoding unit 103 is described therein.

  As described above, according to the fourth embodiment, by setting the fixed length FL for each block variably, it is possible to keep the degree of image quality deterioration within the range of the code amount control unit constant. It is. Furthermore, when the encoded data of a fixed length or less is output to the outside of the apparatus sequentially for each block, the data to be retained during the encoding process is only the overbit encoded data, so the fixed length FL is constant. Compared to the case, the buffer capacity can be reduced.

[Fifth Embodiment]
In the fourth embodiment, the method has been described in which the complexity is acquired as the recognition information, and the encoding is performed by changing the fixed length for each block. In the fifth embodiment, a method for setting a fixed length when a region of interest (hereinafter also referred to as ROI) is acquired as recognition information will be described.

  The processing flow is the same as that of the fourth embodiment and is as shown in FIG. Since the fifth embodiment is characterized by the content of the recognition information to be acquired and the method of setting a fixed length according to the recognition information (step S1404 in FIG. 14), this point will be described below.

  In this embodiment, it is assumed that the position information and size of the ROI specified outside the apparatus are acquired as the recognition information. By acquiring the position information and the size of the ROI, it is possible to specify whether or not the ROI is included for each block in the image. FIG. 18 shows an example of the relationship between the ROI and the block including the ROI. A region surrounded by a thick frame in FIG. 18 is an ROI, and a region indicated by diagonal lines indicates a block including the ROI.

  When encoding a block including an ROI, the fixed length setting unit 1302 sets a predetermined fixed length HFL for ROI in the block encoding unit 103. On the other hand, when encoding a block that does not include an ROI, the fixed length setting unit 1302 sets a predetermined fixed length LFL other than the ROI in the block encoding unit 103. The values of the fixed length HFL for ROI and the fixed length LFL other than ROI are determined from the basic fixed length FL and the target code amount for the entire image. As in the fourth embodiment, it is necessary to determine the values of HFL and LFL so that the data length becomes the upper limit when the basic fixed length FL is applied to all blocks. The fixed length HFL for ROI is determined to be a predetermined value or more in advance according to the attribute of the image, the number of ROIs, and the like. From the relationship with the target code amount, the fixed length LFL for non-ROI is used. It is also possible to use a method for calculating. An example of the variation of the fixed length for each block is shown in FIG.

  With the above method, by setting a fixed length of a certain value or more for a block including the ROI, the code amount control can be performed while maintaining the ROI with high image quality.

  In addition, although the case where complexity and ROI were acquired as recognition information was demonstrated separately, you may set fixed length combining these two information. For example, although the fixed length VFL for each block is set as a base, the fixed length HFL may be set so that the block including the ROI has a data length equal to or greater than a predetermined value. Also in this case, the fixed length that varies for each block needs to be set so that the data length when the basic fixed length FL is applied to all the blocks becomes the upper limit.

  Information for specifying a block including the ROI may be stored in the header of the encoded data.

(Other examples)
The present invention supplies a program that realizes one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and one or more processors in a computer of the system or apparatus read and execute the program This process can be realized. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

  DESCRIPTION OF SYMBOLS 101 ... Input part, 102 ... Block division part, 103 ... Block encoding part, 104 ... Overbit encoding part, 105 ... Buffer, 106 ... Code amount control part, 107 ... Code sequence formation part, 108 ... Output part

Claims (12)

An image encoding device for encoding an image, comprising:
A dividing means for dividing the image data into blocks composed of a plurality of pixels;
When lossless encoded data is generated from the block of interest obtained by the dividing means and the code amount of the lossless encoded data exceeds the target code amount, the excess data is stored in the memory and the excess data is removed. When the lossless encoded data is generated as encoded data of the block of interest and the code amount of the lossless encoded data is less than or equal to the target code amount, the lossless encoded data is converted to the code of the block of interest. First encoding means for generating as encoded data;
Second encoding means for encoding the excess data stored in the memory;
Counting means for counting an excess data capacity with respect to the target code amount each time encoded data less than the target code amount is generated in the block encoding by the first encoding unit;
Encoded data corresponding to the surplus data capacity counted by the counting unit among the encoded data generated by the second encoding unit when the encoding by the first encoding unit is completed for the image data And an amount of code control means for adding to the encoded data obtained by the first encoding means and outputting the encoded data. The first encoding means includes:
Extraction means for extracting the minimum pixel Pmin having the minimum value in the block of interest, the maximum pixel Pmax having the maximum value, and the positions of the minimum pixel Pmin and the maximum pixel Pmax;
From the minimum pixel Pmin and the maximum pixel Pmax, BR representing the minimum number of bits for representing a possible range of pixel values in the block of interest is obtained, and other pixels excluding the minimum pixel Pmin and the maximum pixel Pmax are obtained. Calculating means for calculating the difference from the value of the minimum pixel Pmin or the maximum pixel Pmax with the accuracy of the number of bits indicated by the BR;
Comparing means for comparing the BR and FL representing a fixed length for achieving the target code amount;
When the comparison result of the comparison means indicates BR ≦ FL, the information obtained by the extraction means and the information obtained by the calculation means are generated as encoded data of the block of interest,
When the comparison result of the comparison means indicates BR> FL, information on the number of bits indicated by the higher order FL among the information obtained by the extraction means and the bit number BR indicating the difference obtained by the calculation means Generating means for generating the encoded data of the block of interest, and storing data of the lower number of bits “BR-FL” in the memory among the number of bits BR indicating the difference;
The image encoding apparatus according to claim 1, further comprising: The first encoding means includes:
Extraction means for extracting a representative pixel having a minimum or maximum value in the block of interest, a position of the representative pixel, and a BR representing a minimum number of bits representing a possible range of the value of the pixel in the block of interest;
For other pixels excluding the representative pixel, calculation means for calculating the difference from the value of the representative pixel with the accuracy of the number of bits indicated by the BR;
Comparing means for comparing the BR and FL representing a fixed length for achieving the target code amount;
When the comparison result of the comparison means indicates BR ≦ FL, the information obtained by the extraction means and the information obtained by the calculation means are generated as encoded data of the block of interest,
When the comparison result of the comparison means indicates BR> FL, information on the number of bits indicated by the higher order FL among the information obtained by the extraction means and the bit number BR indicating the difference obtained by the calculation means Generating means for generating the encoded data of the block of interest, and storing data of the lower number of bits “BR-FL” in the memory among the number of bits BR indicating the difference;
The image encoding apparatus according to claim 1, further comprising:   4. The image coding apparatus according to claim 2, wherein the counter unit counts the value of FL-BR when the relation BR <FL is satisfied as the surplus data capacity. The second encoding means stores the first encoding means in the memory from the higher order to the lower order among the data of the lower number of bits “BR-FL” of the number of bits BR indicating the difference. 5. The image encoding device according to claim 2, further comprising: a unit that stores the bit sequence in each bit position. The code amount control means preferentially adds the higher-order bit string stored in the second encoding means to the encoded data obtained by the first encoding means. The image encoding device described. Input means for inputting an image of a region of a preset size from the image data to be encoded;
The image coding apparatus according to claim 1, wherein the dividing unit divides the image of the region input by the input unit. The code amount control means includes:
When the data amount of the encoded data obtained when the target block in the input area is encoded by the first encoding means does not reach the target code amount, and surplus data capacity is generated,
Encoding corresponding to the surplus data capacity in the encoded data generated by the second encoding means for a block that has been encoded before the target block, including up to an area input before the area The image encoding apparatus according to claim 7, wherein the data is added to encoded data generated by the first encoded data of the block of interest. Acquisition means for acquiring information representing the complexity of each block of image data;
The image coding apparatus according to any one of claims 1 to 8, further comprising setting means for setting the target code amount for each block based on information obtained by the obtaining means. Information indicating ROI for the image represented by the image data;
Setting means for setting a first target code amount for a block including the ROI acquired by the acquiring means, and a second target code amount lower than the first target code amount for a block not including the ROI. The image encoding device according to claim 1, further comprising: A control method for an image encoding device for encoding an image, comprising:
A dividing step in which the dividing unit divides the image data into blocks composed of a plurality of pixels;
The first encoding means generates lossless encoded data from the block of interest obtained in the dividing step, and stores the excess data in the memory when the code amount of the lossless encoded data exceeds the target code amount. When the lossy encoded data obtained by removing the excess data is generated as the encoded data of the block of interest, and the code amount of the lossless encoded data is equal to or less than the target code amount, the lossless code A first encoding step of generating encoded data as encoded data of the block of interest;
A second encoding step, wherein a second encoding means encodes the excess data stored in the memory;
A counting step of counting surplus data capacity with respect to the target code amount each time encoded data less than the target code amount is generated in the block encoding in the first encoding step;
When the code amount control means finishes encoding the image data in the first encoding step, the surplus data capacity counted in the counting step among the encoded data generated in the second encoding step A coding amount control step of adding and outputting the coding data corresponding to the coding data obtained in the first coding step, and a control method for the image coding device.   The program for making the said computer perform each process which the method of Claim 11 has, when a computer reads and executes.

Images