JP4343862B2 - Image encoding apparatus, image decoding apparatus, control method thereof, computer program, and computer-readable storage medium - Google Patents

Description

  The present invention relates to a technique for lossless encoding / decoding of an image.

  2. Description of the Related Art Conventionally, there is a technique based on predictive coding as one method for configuring an image processing apparatus that performs lossless encoding and decoding of an image. The predictive encoding device uses a series conversion unit that converts image data into a prediction error by predictive conversion, and a prediction error output from the series conversion unit by using entropy coding such as Huffman coding. Generally, it is configured by an entropy encoding unit that converts data into a small amount of encoded data.

  In JPEG (ITU-T T.81 | ISO / IEC10918-1) recommended as an international standard still image coding method by ISO and ITU-T, a lossless coding method based on predictive coding is used as an independent function. It is prescribed. Hereinafter, this is referred to as a JPEG lossless encoding mode. In the JPEG lossless encoding mode, seven prediction formulas are defined as methods for predicting the value of the pixel of interest from neighboring pixels, and the prediction method can be selected according to the image.

  FIG. 2 shows a block diagram of a conventional image processing apparatus. The figure is an example of an apparatus for reversibly compressing an image using the above-described JPEG lossless encoding mode. In the figure, 201 is a buffer, 202 is a component value prediction unit, 203 is a subtracter, 204 is a Huffman table memory, 205 is a prediction error encoding unit, 209 is a code string forming unit, and 206, 207 and 208 are It is a signal line.

  The Huffman table memory 204 stores a Huffman table used in the prediction error encoding unit 205. Here, it is assumed that the Huffman table shown in FIG. 6 is stored.

  The flow of processing for encoding a color image in which each RGB component is represented by 8 bits by the conventional image processing apparatus will be described with reference to FIG.

  First, a prediction selection signal m for selecting a prediction method used in the component value prediction unit 202 is input from the signal line 208. The prediction selection signal m is an integer value from 0 to 7, and a different prediction formula is associated with each. The correspondence between the prediction formula used and the prediction selection signal m is as shown in FIG. Note that when the prediction selection signal m is 0, there is no prediction formula, but this means that prediction conversion is not performed and each component is encoded as it is. The symbols p, a, b, and c in the figure will be described later.

  Next, image data is sequentially input from the signal line 206. Assume that the pixel data is input in raster scan order, and component data is input in the order of R, G, and B in each pixel. R, G, and B are defined as component numbers 0, 1, and 2, respectively, and the component number C of the pixel at the horizontal right pixel position x and the vertical lower pixel position y with the upper left corner of the image as coordinates (0, 0). Is represented as P (x, y, C). For example, if the pixel at the position (x, y) = (3,4) has a value of (R, G, B) = (255, 128, 0), P (3,4, 0) = 255, P (3,4,1) = 128 and P (3,4,2) = 0.

  The buffer 201 stores image data input from the signal line 206 for two lines.

  The component value predicting unit 202 uses the same component value a = P (x−1, y, C) of the immediately preceding pixel for the component value x = P (x, y, C) of the pixel of interest from the buffer 201 one line before. The value b = P (x, y−1, C) of the same component of the pixel and the value c = P (x−1, y−1, C) of the same component of the upper left pixel are extracted, and the prediction method is selected. A predicted value p is generated according to the signal m. FIG. 3 shows the positional relationship between the component value x of the pixel of interest and a, b, and c. Note that when the pixel of interest is located at the end of the image (the uppermost line, the left end), a, b, and c do not exist. Therefore, the processing is performed assuming that a, b, and c are “0”. Since the input order of the image data is as described above, the positions of the pixels a, b, and c are already encoded pixel positions.

  The subtracter 203 obtains a difference value between the component value x to be encoded and the predicted value p, and outputs it as a prediction error e.

  The prediction error encoding unit 205 first classifies the prediction error e input from the subtractor 203 into a plurality of groups, and generates a group number SSSS and additional bits having a bit length determined for each group. FIG. 5 shows the relationship between the prediction error e and the group number SSSS. The additional bits are information for specifying the prediction error within the group, and the bit length is given by the group number SSSS. When SSSS = 16, the bit length is exceptionally 0 (however, when the accuracy of each component is 8 bits, SSSS = 16 does not occur). If the prediction error e is positive, the lower SSSS bits of the prediction error e are additional bits, and if it is negative, the lower SSSS bits of e-1 are additional bits. The most significant bit of the additional bits is 1 if the prediction error e is positive, and 0 if the prediction error e is negative. The encoding process first refers to the Huffman table stored in the memory 204 and outputs encoded data corresponding to the group number SSSS. If the SSSS is not 0 or 16, then the bit determined by the group number This is done in the order of outputting additional bits in length.

  The code string forming unit 209 includes the prediction selection signal m input via the signal line 208, additional information such as the number of pixels in the horizontal / vertical direction of the image, the number of components constituting the pixel, and the accuracy of each component, From the encoded data output from the prediction error encoding unit 205, a code string having a format compliant with the JPEG standard is formed and output to the signal line 207.

  When the image processing apparatus according to the conventional method described above encodes image data having a biased frequency distribution of the luminance of each component, such as a CG image or a limited color image, the prediction error after the above-described series conversion is reduced. The frequency of occurrence may also be biased towards some specific prediction error value.

  In such a case, although a short code length is assigned as a code length by entropy coding, there may be a prediction error that hardly or not occurs, and the compression rate does not improve.

Conventionally with respect to such a problem, a method of determining whether or not the frequency distribution of occurrence of a prediction error is discrete, and changing the encoded data corresponding to the prediction error according to the determination result (for example, a patent) And a method for determining whether or not the encoding target image is composed of discrete pixel values and correcting the predicted value according to the determination result (see, for example, Patent Document 2). ing. In addition, a method is also known in which matching information with neighboring pixels is encoded in encoding for each pixel (for example, Patent Document 3). Moreover, a method for switching a plurality of predictive encoding methods is known (Patent Document 4).
JP-A-10-004551 Japanese Patent Laid-Open No. 10-004557 JP-A-10-336458 Japanese Patent Laid-Open No. 9-224253

  In Patent Document 1 or 2, it is necessary to grasp the occurrence status of the component value of each pixel or the occurrence status of the prediction error, which increases the processing load. In addition, when each component is predictively encoded using a block code such as a Huffman code, at least one bit is required for each component. Further, the predictive coding technique can be expected to have a high compression rate for natural images (for example, images obtained by capturing the natural world with a digital camera), but in terms of coding efficiency for image data with low entropy such as characters, line drawings, and CG images. But there is still room for improvement. This also applies to Patent Documents 3 and 4.

  The present invention has been made in view of the above-mentioned problems, and can efficiently code not only natural images such as photographic images but also computer graphics with a relatively limited number of colors and document images. It is an object of the present invention to provide a technique for decoding and a decoding technique thereof.

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 that inputs image data in which one pixel is composed of a plurality of color component data and performs lossless encoding,
For each of a plurality of color component data constituting the input pixel-of-interest data, a prediction code that obtains a prediction value by referring to the color component data of an encoded pixel in the vicinity of the pixel of interest and generates predictive encoded data And
Color number counting means for counting the number of colors represented by a plurality of pixels at the encoded pixel position in the vicinity of the target pixel and generating color number information;
Wherein the target pixel data, wherein there in the vicinity of the target pixel is compared with the plurality of pixel data each other in the encoded pixel position, whether neighboring pixels of the same color as the target pixel is present, and the presence a vector information generating means for generating a vector information specifying the relative positions of the corresponding neighboring pixels for the target pixel in the case of,
Vector information encoding means for encoding the vector information;
Based on the color number information obtained from the color number counting means and the vector information obtained by the vector information generation means, output from the encoded data obtained by the predictive coding means and the vector information coding means. Encoded data generation means for generating encoded data for use,
The encoded data generation means includes
Number of colors represented by the color number information is equal to or less than a predetermined threshold value, and, if the vector information indicates the presence of a neighboring pixel having the same color as the target pixel, the generated vector information encoding means encodes data and then output as encoded data of the target pixel,
In the counted number of colors than the predetermined threshold value by the number of colors counting means, and, if the vector information indicates absence of neighboring pixels having the same color as the target pixel, generated by the vector information encoding means predictive coded data and coded data said generated by predictive coding means, and output as coded data of the target pixel,
If the number of colors that are counted in the number of colors counting means exceeds the predetermined threshold, the predictive coding data generated by said prediction encoding means, and outputs the encoded data of the target pixel.

  According to the configuration of the present invention, the coding efficiency for image data such as a natural image is substantially the same as that of an image data such as a CG image or a text document while maintaining the conventional efficiency. It becomes possible to perform lossless encoding at a further compression rate.

  Further, according to another invention, it is possible to perform lossless decoding from the encoded data as described above and to completely restore the original image.

  Embodiments according to the present invention will be described below in detail with reference to the accompanying drawings.

<First Embodiment>
FIG. 1 is a block diagram of an image processing apparatus according to the embodiment.

  As shown in FIG. 1, the image processing apparatus according to the present embodiment includes a neighborhood match determination unit 101, a match pixel position encoding unit 102, a code generation unit 103, a code formation unit 104, a buffer 201, a component value prediction unit 202, A subtractor 203, a Huffman table memory 204, a prediction error encoding unit 205, and a pixel match detection unit 107 are provided. In the figure, reference numerals 206, 207, 105, 106 and 108 denote signal lines. It should be noted that blocks that perform the same operations as the processing blocks of the conventional image processing apparatus of FIG.

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

  The 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 configured by pixel data representing a luminance value in the range of 0 to 255 in 8 bits. Shall be. The image data to be encoded has a size of W pixels in the horizontal direction and H pixels in the vertical direction. Therefore, the input image data has a data amount of W × H × 3 bytes. The horizontal right direction is defined as the positive direction of the X coordinate, and the vertical downward direction is defined as the positive direction of the Y coordinate. The input order of the input image data is a raster order, and each pixel is configured by arranging data in the order of R, G, and B.

  Now, the encoding target image data is sequentially input from the signal line 206. The supply source of the image data to be input may be an image scanner or image data stored in a predetermined storage device, and the type of input source is not limited.

  As described above, the input order of the pixel data is the raster scan order, and the component data is input in the order of R, G, and B for each pixel. As for the luminance value of each component of the pixel, R, G, and B are expressed by component number C, C = 0 indicates R, C = 1 indicates G, and C = 2 indicates B. Further, with the upper left corner of the input image as the coordinates (0, 0), the value of the component number C (luminance in the embodiment) of the pixel at the horizontal pixel position x and the vertical pixel position y is set to P (x, y, C). ). Accordingly, the values of the R, G, and B components of the pixel at the horizontal pixel position “10” and the vertical pixel position “25” are P (10, 25, 0), P (10, 25, 1), P, respectively. It is expressed as (10, 25, 2). When pixel values are expressed as a group of components, they are expressed as a vector X having R, G, and B component values as elements. For example, a pixel whose R, G, and B component values are r, g, and b is described as X = (r, g, b).

  The image processing apparatus outputs a corresponding code from the code string forming unit 104 for each pixel of image data sequentially input from the signal line 206.

  The buffer 201 has an area for storing image data for two lines, sequentially stores pixel data input from the signal line 206, and stores and holds pixel data for two lines. As described above, in the image data, component values are input in the order of raster for each pixel and in the order of R, G, and B for each pixel. From the buffer 201, the same component values of the three pixels a, b, and c around the pixel of interest x in FIG. 3 are read and output to the neighborhood match determination unit 101 and the component value prediction unit 202. When the component value of interest is P (x, y, C), a = P (x−1, y, C), b = P (x, y−1, C), c = P (x−1) , Y-1, C). When a, b, and c are outside the image at the first line of the image or at the beginning / end of each line, it is assumed that the encoding side and the decoding side have a common value. In this embodiment, it is assumed that a, b, and c = 0. Note that the pixel of interest x is the encoding target position, and the positions of the surrounding pixels a, b, and c are already encoded positions.

  Now, the neighborhood match determining unit 101 compares the component value x = P (x, y, C) of the pixel of interest with a, b, c read from the buffer 201, and a neighborhood having the same pixel value as the pixel of interest. Information indicating the presence / absence of a pixel (hereinafter referred to as matching pixel presence information) 105 is provided in the subsequent code generation unit 103, and information indicating the relative direction of a neighboring pixel matching the pixel of interest (hereinafter referred to as vector information) 106 is determined as subsequent matching It outputs to the pixel position encoding part 102, respectively.

  Here, when the coincidence pixel presence information 105 is generated, the meaning that two pixels are the same means that there are two pixels X1 = (r1, g1, b1) and X2 = (r2, g2, b2). This means that all component values are the same, such as r1 = r2, g1 = g2, and b1 = b2. The matching pixel presence information 105 is “0” when one of the neighboring pixels a, b, and c of the target pixel matches the target pixel, and “1” when none exists. Become.

  The vector information 106 is as follows.

When the pixel value of interest (color) is X, the previous pixel value is Xa, the x coordinate is the same, the pixel value of the previous line is Xb, and the pixel value diagonally upper left is Xc.
When X = Xa (Xb and Xc do not matter), the vector information 106 is “0”,
When X ≠ Xa and X = Xb (Xc does not matter), the vector information 106 is “1”,
When X ≠ Xa, X ≠ Xb, and X = Xc, the vector information 106 is “2”,
In other cases, the vector information 106 is “3”.

When the position of the pixel of interest for obtaining the matching pixel presence information 105 and the vector information 106 is (x, y) and X, Xa, Xb, and Xc are expressed using this, the following information may be used. It is clear.
X = (P (x, y, 0), P (x, y, 1), P (x, y, 2))
Xa = (P (x-1, y, 0), P (x-1, y, 1), P (x-1, y, 2))
Xb = (P (x, y-1, 0), P (x, y-1, 1), P (x, y-1, 2))
Xc = (P (x-1, y-1,0), P (x-1, y-1,1), P (x-1, y-1,2))
As described above, when the vector information 106 is “3”, the matching pixel presence information 105 is always “1”. Conversely, when the matching pixel presence information 105 is “1”, the vector information 106 is always “3”. Therefore, the matching pixel presence information 105 is obtained from the vector information 106, but is shown as a separate signal for convenience in the embodiment.

  The matching pixel position encoding unit 102 encodes the vector information 106 from the neighborhood matching determination unit 101, and outputs the resulting code word to the code generation unit 103. In the present embodiment, a code word is generated from the vector information 106 using the code table shown in FIG. 7 (vector information is encoded).

  The pixel coincidence detection unit 107 compares the three pixel values a, b, and c around the pixel of interest read from the buffer 201, and sets the same pixel value among the three surrounding pixels (Xa, Xb, Xc). It is checked whether or not a pair exists. If a pair having the same pixel value is detected, “1” is output as peripheral pixel state information 108, and if not detected, “0” is output.

  That is, the pixel coincidence detection unit 107 sets “1” when “Xa = Xb”, “Xa = Xc”, or “Xb = Xc” between the three pixels Xa, Xb, and Xc around the pixel of interest X. In other cases, peripheral pixel state information 108 of “0” is output. Note that the presence of at least one pair of identical pixel values in the three surrounding pixels is equivalent to the number of colors included between the three surrounding pixels being two or less. That is, the above can be considered as a determination result of whether or not the number of types of colors existing in the three neighboring pixels of the target pixel is two or less.

  For this reason, when encoding the first line of the image, at least the neighboring pixels b and c are outside the image area and are treated as “0”. That is, since at least Xb = Xc, the peripheral pixel state information 108 when the first line is encoded is “1”.

  On the other hand, each component of the pixel is determined in accordance with the prediction selection signal m input from the signal line 208 by the component value prediction unit 202, the subtractor 203, and the prediction error encoding unit 205 in the same manner as the conventional method described above. The predictive encoding process for each component is performed. Note that the prediction selection signal m is a signal that can be arbitrarily determined by setting of the apparatus, and is fixed at least during encoding of all image data to be encoded.

The code generation unit 103 temporarily stores the code word of the vector information output from the matching pixel position encoding unit 102 and the prediction encoded data of each component of the target pixel output from the prediction error encoding unit 205. Then, the code generation unit 103 forms code data for the pixel of interest according to the matching pixel presence information 105 and the surrounding pixel state information 108 and outputs the code data. Specifically, it is as follows.
i) When the surrounding pixel state information 108 is “0”, that is, when there is no pair of the same color between the three pixels Xa, Xb, and Xc around the pixel of interest X, the prediction error encoding unit 205 Select and output prediction encoded data of each component to be output. This is a process corresponding to normal encoding.
ii) The peripheral pixel state information 108 is “1” (when there is at least one pair of the same color among the three pixels Xa, Xb, and Xc around the pixel of interest X), and the matching pixel presence information 105 is In the case of “0” (when the color of the pixel of interest is the same as any of the surrounding three pixels Xa, Xb, and Xc), vector information output from the coincident pixel position encoding unit 102 (in this case, always one of 0 to 2) ) Is selected and output.
iii) When the surrounding pixel state information 108 is “1” and the matching pixel presence information 105 is “1” (when the color of the pixel of interest does not match any of the surrounding three pixels Xa, Xb, Xc), The encoded data of the vector information (always “3”) output from the coincidence pixel position encoding unit 102 is output, and then the prediction encoded data of each component output from the prediction error encoding unit 205 is output. .

  FIGS. 8A to 8C show code configurations of the above three encoded data. FIG. 5A shows the code data of the pixel of interest when the condition i, that is, the peripheral pixel state information 108 is “0”. FIG. 8B shows code data of the pixel of interest when the peripheral pixel state information 108 of the condition ii is “1” and the matching pixel presence information 105 is “0”, and FIG. This is code data of the pixel of interest when the peripheral pixel state information 108 of the condition iii is “1” and the matching pixel presence information 105 is “1”.

  The points to be noted in FIGS. 8A to 8C are the encoded data in FIG. 8B. The condition for generating the encoded data is that the number of colors in the vicinity of the target pixel is equal to or less than the predetermined number and the color of the target pixel X matches any of the neighboring pixels Xa, Xb, and Xc. It is. Compared with FIG. 7B, FIG. 7A or FIG. 9C, the encoded data for each color component is no longer necessary, and the code length can be expressed by 3 bits at the maximum as shown in FIG. This means that further compression efficiency is desired. This will be described in more detail below.

  In general, it is known that in a natural image (an image (especially a landscape image) taken by a digital camera), neighboring pixels tend to have substantially the same color. However, this does not mean that the colors of neighboring pixels completely match each other, but rather has a high probability of having a minute difference. Hereinafter, an example will be described based on such points.

Assume that the R component prediction error is in the range of −31 to −16 or 16 to 31. In this case, the prediction error group number SSSS is “5” with reference to FIG. According to FIG. 6, the code word of the group number “5” is “110 (B)” (B means a binary number), and the additional bit length is found to be “5” from FIG. The prediction encoded data of the R component is
“110XXXX (B)”
(X is a bit of 0 or 1).

  On the other hand, referring to FIG. 7, the code word when the vector information 105 is “2” is “110 (B)” (B means a binary number), and the first 3 bits of the R component encoded data and It will be the same. That is, in the processing on the decoding side, as long as only this code word is referred to, “110 (B)” means a code word of vector information “2” or a code word of group number SSSS “5”. I can not distinguish.

  However, when decoding, it should be noted that neighboring pixels a, b, and c of the pixel of interest have already been decoded. That is, when decoding the pixel of interest, it can be easily detected whether at least one pair of the same color exists between the neighboring pixels a, b, and c. When the same color pair exists in the neighboring pixels a, b, and c (when the number of colors is 2 or less), it is promised that the encoded data is one of FIGS. It turns out that 110 "is a code word of vector information. On the other hand, when there is no pair of the same color in the neighboring pixels a, b, and c (the number of colors is 3), the encoded data is encoded by prediction error encoding as shown in FIG. It turns out that it consists only of data. The same applies to other code words of vector information.

  On the other hand, the encoded data of FIG. 8C is the same as the encoded error data of FIG. 8A in that the code word “111 (B)” with vector information “3” is added prior to the prediction error encoded data. It can be said that the data is more redundant than the encoded data. That is, when predictive encoded data is employed, it is desirable to reduce the probability that a code word as shown in FIG. In this respect, according to the present embodiment, the condition for generating the encoded data in FIG. 5C is that at least a pair of the same color exists between the neighboring pixels a, b, c (the number of colors is two colors). The following). In other words, when the neighboring pixels a, b, and c are all different colors (number of colors is 3), the encoded data of FIG. Inevitably, the probability that the encoded data shown in FIG.

  In the case of computer graphics, painting is usually performed with colors artificially generated on a computer. Therefore, the number of colors used when an image imitating a natural image is generated by computer graphics is smaller than that of a natural image. In other words, since the proportion of colors that completely match between neighboring pixels is significantly higher than that of a natural image, the probability that the encoded data of FIG. 5B is generated is increased, which is the conventional lossless encoding. A compression rate that is several stages better than predictive coding can be expected. In particular, in the case of a text document image composed of ordinary characters, there are only two colors (black and white), so the probability of generating encoded data as shown in FIG. 8A is increased. be able to.

  On the other hand, a large amount of encoded data shown in FIG. 8A is generated in an image captured by a digital camera, and the same code amount as that of the conventional image can be maintained for a natural image.

  Returning to the description of FIG. In the code string forming unit 104, the prediction selection signal m input via the signal line 208, the number of pixels in the horizontal / vertical direction of the image, the number of components constituting the pixel, the accuracy of each component, and the code A code string in a predetermined format is formed from the encoded data output from the generation unit 103 and is output via the signal line 207. The output destination may be a storage device or a communication line. When output to a storage device, it will be stored as a file.

  FIG. 9 shows an example of the configuration of the code string output from the code string forming unit 104. Additional information (including the prediction selection signal m and the horizontal and vertical pixel number information of the image) is stored in the head portion of the head, and then the encoded data of each pixel in the raster scan order is arranged. Each pixel encoded data in the figure has one of the structures shown in FIGS. 8A, 8B, and 8C.

  As described above, in the image processing apparatus according to the present embodiment, the encoded pixels Xa, Xb, and Xc around the encoding target pixel X are checked, and the same color is obtained between Xa, Xb, and Xc. When a pair exists (when there are two colors or less), a vector that specifies the match / mismatch between the pixel X and the pixels Xa, Xb, and Xc, and, if they match, the relative position of the matching neighboring pixels Encode information. If the same color does not exist between the pixel of interest X and Xa, Xb, and Xc, then by generating predictive encoded data for each component of the pixel of interest, characters, line drawings, CG images, etc. The reversible compression performance of image data that is likely to have pixels of the same color around the pixel of interest can be greatly improved.

  If there is no pair of the same color among Xa, Xb, and Xc, each component of the pixel of interest is predictively encoded without encoding the vector information. As a result, addition of encoded data of vector information can be avoided even when encoding image data such as a natural image that has a low probability of complete matching with surrounding pixel values. It becomes possible to maintain a code amount equivalent to the conventional predictive coding technique.

<Modification of First Embodiment>
The above embodiment may be realized by software. Therefore, in this modification, an example realized by software will be described.

  When implemented by software, the processing corresponding to the various components represented by the proximity match determination unit 101 shown in FIG. 1 is implemented by functions, subroutines, etc. on the software. The names of the processing units in FIG. 11 are used as they are. Note that the CPU 1401 secures the buffer 1201 and the code buffer 1103 in the RAM 1402 before starting the processing.

  FIG. 14 is a diagram showing a basic configuration of an apparatus when realized by software.

  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 and decoding processing described later.

  A RAM 1402 stores 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 a work area when the CPU 1401 executes various processes. Used as. The buffer 1201, the Huffman table 204, etc. shown in FIG. 1 are also secured in the RAM 1402.

  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 pointing devices such as a keyboard and a mouse (registered trademark), and can input various instructions 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 includes an OS, a program for image encoding and decoding processing described later, image data to be encoded, and code of the image to be decoded. And the like, and these programs and data are loaded into a predetermined area on the RAM 1402 under the control of the CPU 1401.

  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 and decoding processing described later, image data to be encoded, encoded data of an image to be decoded, and the like may be recorded in this storage medium. 1408 loads these programs and data into a predetermined area on the RAM 1402 under the control of the CPU 1401.

  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, image data to be encoded, encoded data of an image to be decoded, and the like can be input to the RAM 1402, the external storage device 1407, or the storage medium drive 1408 of this apparatus. A bus 1410 connects the above-described units.

  FIG. 10 is a flowchart showing the flow of the encoding process of the image encoding application by the image processing apparatus according to this modification. Note that the program according to the figure is loaded in the RAM 1402, and the CPU 1401 can execute the process according to the flowchart shown in FIG. The processing corresponding to the various processing units shown in FIG. 1 is realized by functions, subroutines, etc. on the software. In this modification, the names of the processing units in FIG. 1 are used as they are for convenience. It shall be. Hereinafter, the overall flow of the image encoding process performed by the image processing apparatus according to the present modification will be described with reference to FIG.

  First, the prediction selection signal m is input from the outside of the apparatus via the signal line 208 (step S1000). Subsequently, the code string forming unit 104 generates and outputs a header including additional information of the encoding target image, such as the prediction selection signal m (step S1001).

  Next, a counter y that holds the vertical position of the pixel of interest is set to 0 (step S1002), and a counter x that holds the horizontal position of the pixel of interest is set to 0 (step S1003). Similarly, the counter C holding the component number of interest is set to 0 (step 1004).

  The component value P (x, y, C) of the target pixel is input from the signal line 206 and stored in the buffer 1201. The component value P (x, y, C) is predictively encoded by the component value predicting unit 202, the subtractor 203, and the prediction error encoding unit 205, and at the same time, the neighborhood match determining unit 101 calculates the same component value of the surrounding pixels. Comparison is performed (step S1005).

  1 is added to the component number C of interest (step S1006). If C <3 as compared with 3, which is the number of color components of the RGB image, the process returns to step S1005, and the next component is similarly processed. If not, the process proceeds to step S1014.

  In step S1014, the pixel match detection unit 107 determines whether or not the same pixel value exists between the three neighboring pixels. If it exists (YES), the process proceeds to step S1008. If not (NO), the process proceeds to step S1009. Move on to In step S1008, the vector information 105 obtained by the neighborhood match determination unit 101 is coded by the match pixel position coding unit 102.

  In step S1009, according to the matching pixel presence information 105 obtained by the neighborhood match determination unit 101, the surrounding pixel state information 108 from the pixel match detection unit 107, and the encoded data of the vector information from the match pixel position coding unit 102 A code for the pixel of interest is formed and output via the code string forming unit 104 (step S1009).

  Thereafter, 1 is added to the counter x that holds the horizontal position of the pixel of interest (step S1010). If x <W compared with the horizontal pixel count W of the image, the process returns to step S1004, and the next step The pixel is encoded. If it is determined that x = W, 1 is added to the counter y that holds the vertical position of the pixel of interest (step S1012). Then, it is determined whether or not y <H by comparing the y coordinate of the target pixel with the number of vertical pixels H (step S1013). If y <H, the encoding process is incomplete, so the process returns to step S1003, and the same process is performed for the pixels on the next line. When y = H, the encoding process for one image has been completed, and thus the present encoding process ends.

  As described above, even with software, it is possible to achieve the same operational effects as the first embodiment.

<Second Embodiment>
Next, an example in which the encoded data generated in the first embodiment (or a modification thereof) is decoded will be described as a second embodiment. The decoding process may basically be executed in the reverse procedure of the encoding process in the image processing apparatus of the first embodiment.

  FIG. 11 is a block diagram of an image processing apparatus that decodes encoded image data according to the second embodiment.

  As shown in FIG. 11, the image processing apparatus according to the present embodiment includes a code buffer 1101, a header analysis unit 1102, a matching pixel position decoding unit 1103, a prediction error decoding unit 1104, a selector 1105, a switch 1106, a buffer 1201, a component value. A prediction unit 1202, an adder 1111, and a pixel match detection unit 1107 are provided. In the figure, reference numerals 1107, 1108, 1109, 1110 and 1112 denote signal lines.

Hereinafter, processing performed by the image processing apparatus according to the present embodiment will be described with reference to FIG.
The encoded data to be decoded is input to the code buffer 1101 of the image processing apparatus via the signal line 1107.

  The header analysis unit 1102 analyzes the header portion of the encoded data stored in the code buffer 1101, and predicts the selection signal m, the number of pixels in the horizontal / vertical direction of the image, the number of components constituting the pixel, Get additional information such as accuracy. The additional information is used in the decoding process. In particular, the prediction selection signal m is output to the component value prediction unit 202.

  Finally, pixel data obtained by decoding is output from the switch 1106. This decoded pixel data is also stored in the buffer 1201 (having a capacity for storing decoded pixel data for two lines).

  Similar to the pixel match detection unit 107 of the first embodiment, the pixel match detection unit 1107 checks whether there is a pair of the same color in the three pixels (Xa, Xb, Xc) around the pixel of interest. The result is output to the matching pixel position decoding unit 1103 as the surrounding pixel state information 1112. In order to make it the same as the encoding process, when the pixel of interest to be decoded is located at the head line of the image, it is handled as Xb = Xc = 0. Further, when the pixel of interest is at the head position of each line, it is handled as Xa = 0.

  When the surrounding pixel state information 1112 is “1”, that is, when there is a pair of at least the same color between the decoded pixel data Xa, Xb, and Xc around the pixel X to be decoded, the input encoded data structure Is one of FIGS. 8B and 8C. Therefore, the coincidence pixel position decoding unit 1103 promises that the encoded data to be decoded is encoded data of vector information code, and therefore decodes it as vector information 1109 with reference to the table shown in FIG. Output to the selector 1105. At this time, if the decoded vector information is “3”, the signal 1108 for controlling the prediction error decoding unit 1104 and the switch 1106 is set to “1” via the signal line 1108 and output otherwise. Outputs “0”.

  On the other hand, when the surrounding pixel state information 1112 is “0”, the encoded data is the encoded data of FIG. Therefore, the matching pixel position decoding unit 1103 outputs “1” as the signal line 1108 without performing the decoding process.

  When the control signal 1108 from the coincidence pixel position decoding unit 1103 is “1”, the prediction error decoding unit 1104 refers to the Huffman table stored in the Huffman table memory 204 and determines each component of the pixel of interest. Decode the prediction error.

  The component value prediction unit 202 generates and outputs a prediction value p for the component value of interest using the prediction formula selected by the prediction method selection signal m, similarly to the case of encoding.

  The adder 1111 adds the prediction error e decoded by the prediction error decoding unit 1104 to the prediction value p output from the component value prediction unit 202 to restore and output the component value of interest.

  The selector 1105 selects and outputs one of the component values of the surrounding pixels Xa, Xb, and Xc of the pixel of interest X read from the buffer 201 in accordance with the vector information 1109 from the coincident pixel position decoding unit 1103. More specifically, when the vector information is “0”, the selector 1105 selects each component value of Xa. Further, when it is “1”, each component value of Xb is selected and output when it is “2”. If the vector information is “3”, the output from the selector 1105 is invalid, but anything may be output. This is because the switch 1106 does not select the data.

  When the signal line 1108 from the coincidence pixel position decoding unit 1103 is “1”, the switch 1106 selects and outputs pixel data obtained by predictive decoding, that is, the input terminal a. When the signal 1108 is “0”, the pixel data generated according to the input terminal b, that is, the vector information is selected and output.

  Although the value of each component of each pixel decoded from the signal line 1110 is output, as described above, the pixel data for two lines of the decoding result is also stored in the buffer 1201 and the subsequent pixels are decoded. Is used as a surrounding pixel value.

  Through the above processing, the original image data can be completely restored (reversible decoding) from the encoded data.

<Modification of Second Embodiment>
The second embodiment may be realized by software. An example of this will be described below. Since the apparatus configuration when realized by software is the same as that of FIG. 14 referring to the modification of the first embodiment, description thereof is omitted, and the processing procedure of the decryption application executed on the apparatus is shown in the flowchart of FIG. It explains according to. The processing corresponding to the various processing units shown in FIG. 11 is realized by functions, subroutines, etc. on the software, but in this modification as well, the names of the processing units in FIG. 11 are used as they are for convenience. It shall be. And the processing content of each part shall perform the process demonstrated in the said 2nd Embodiment. As a matter of course, the CPU 1401 secures the buffer 1201 and the code buffer 1103 in the RAM 1402 before starting the processing.

  The encoded data to be decoded is input from the signal line 1107 to the code buffer, and the header analysis unit 1102 analyzes the header of the encoded data and adds additional information (prediction selection signal m, image horizontal direction / The number of pixels in the vertical direction, the number of components constituting the pixels, the accuracy of each component, and the like are extracted (step S1201).

  Next, a counter y that holds the vertical position of the pixel of interest is set to 0 (step S1202), and a counter x that holds the horizontal position is set to 0 (step S1203). Further, the counter C holding the component number of interest is set to 0 (step S1204).

  In the decoding of each pixel, first, the pixel match detection unit 107 determines whether there is a pair of the same color among the three neighboring pixels. If such a pair exists (YES), the process proceeds to step S1205. If not (NO), the process proceeds to step S1216.

  In step S1205, the coincidence pixel position decoding unit 1103 decodes the vector information and outputs it from the signal line 1109, and outputs a control signal 1108 for controlling the prediction error decoding unit 1104 and the switch 1106. In step S1206, it is determined whether the vector information obtained by decoding is “3”. If this determination is YES, the process proceeds to step S1207, and if NO, the process proceeds to step 1208.

  If it is determined Yes in step S1206 (note that the signal line 1108 is “1” at this time), the switch 1106 is caused to select the terminal a. As a result, the component value P (x, y, C) of interest is decoded and output by the component value prediction unit 202, the prediction error decoding unit 1104, and the adder 1111 (step S1207). On the other hand, if the determination in step S 1206 is No (note that the signal line 1108 is “0” at this time), the terminal b is connected to the switch 1106. As a result, the selector 1105 selects any one component value of peripheral pixels read from the buffer 201 based on the neighborhood match information decoded by the match pixel position decoding unit 1103, and the component value of the decoding result of the pixel of interest is selected. Output as P (x, y, C) (step S1208).

  The component value P (x, y, C) output in step S1207 or step S1208 is output from the signal line 1110 to the outside of the apparatus and stored in the buffer 201.

  In step S1209, 1 is added to the counter C holding the component number of interest, and compared with 3, which is the number of color components of the RGB image. If C <3, the process returns to step S1206, the same process is performed for the next component, and if not, the process proceeds to step S1211.

  On the other hand, when the process proceeds to step S1216 due to the branch of step S1215, the detection result output from the pixel match detection unit 107 is “0”, and the control signal output from the match pixel position decoding unit 1103 is “1”. The switch 1106 is connected to the terminal a, and the component value P (x, y, 0) of interest is decoded by the component value prediction unit 202, the prediction error decoding unit 1104, and the adder 1111 (step S1216).

  Subsequently, similarly, the component value prediction unit 202, the prediction error decoding unit 1104, and the adder 1111 decode the component value P (x, y, 1) (step S1217), and further the component value P (x, y, 2) is obtained. Decoding is performed (step S1218).

  In step S1211, 1 is added to the counter x that holds the horizontal position of the pixel of interest, x is compared with W in step S1212, and it is determined whether or not the decoded pixel of interest is a pixel at the end position of one line. judge. If x <W, the decoding process for one line is incomplete, so the process returns to step S1204 to perform the decoding process for the next pixel. If x = W, the process advances to step S1213 to add 1 to the counter y that holds the vertical position of the pixel of interest.

  In step S1214, the vertical coordinate y of the pixel of interest is compared with the number H of pixels in the vertical direction of the image. If y <H, the process returns to step S1203, and the same process is performed for each pixel of the next line. If y = H, the decoding process for the entire image has been completed. The decryption process ends.

  As a result, lossless decoding is possible as in the second embodiment.

<Third Embodiment>
In the said 1st Embodiment and its modification, it was set as the structure which encodes vector information using a single code table. Although this configuration is simple, redundancy remains. In addition, when the number of surrounding colors is small, there is a high possibility that either the target pixel value and the surrounding pixel value are the same, and conversely, when the number of colors is large, the pixel value tends to be low. Therefore, as a third embodiment, by switching whether to adaptively encode vector information according to the number of colors of surrounding pixels, and further changing the encoding method of vector information according to the number of colors, A technique aimed at further improving the compression efficiency by removing the code redundancy will be described.

  FIG. 13 shows a block diagram of an image processing apparatus that performs image coding in the third embodiment.

  As shown in FIG. 13, the image processing apparatus for encoding image data according to the third embodiment includes a color number determination unit 1301, a neighborhood match determination unit 1302, a match pixel position encoding unit 1303, a prediction error code. A conversion unit 1304, a code generation unit 1305, a buffer 201, a component value prediction unit 202, and a subtractor 203. In the figure, reference numerals 105, 106, 206, and 207 denote signal lines. Note that processing units that perform the same operations as the processing units of the image processing apparatus according to the first embodiment are given the same reference numerals, and descriptions thereof are omitted.

Hereinafter, an image encoding process performed by the image processing apparatus according to the third embodiment will be described with reference to FIG.
The image data to be encoded by the image processing apparatus according to the present embodiment is the same as that of the first embodiment. That is, it is RGB color image data, and is composed of pixel data expressing luminance values in the range of 0 to 255 with 8 bits of each component (color component). The image data to be encoded is arranged in dot order, that is, in the raster scan order, and the pixels are arranged in the order of R, G, and B. The image is assumed to be composed of horizontal W pixels and vertical H pixels.

  In the image processing apparatus according to the first embodiment, the prediction selection signal m for selecting the prediction method used by the component value prediction unit 202 is input from the outside of the apparatus, and the prediction method is switched by this. In the image processing apparatus according to the embodiment, the prediction selection signal m set in the component value prediction unit 202 is assumed to be fixed at “4”.

For each component of the pixel X of interest, the color number determination unit 1301 reads the same component values of the surrounding four pixels indicated by a, b, c, and d in FIG. 3 from the buffer 201 to obtain pixels Xa, Xb, Xc, and Xd. . Assuming that the position of the pixel of interest is (x, y), Xa, Xb, and Xc are as described above. The pixel Xd is as follows.
Xd = (P (x + 1, y-1, 0), P (x + 1, y-1, 1), P (x + 1, y-1, 2))

  The color number determination unit 1301 detects how many kinds of colors exist in the set of the pixel data Xa, Xb, Xc, and Xd, and outputs the color number information Nc. Specifically, six combinations that extract two pixels from four pixels, that is, whether Xa and Xb, Xa and Xc, Xa and Xd, Xb and Xc, Xb and Xd, Xc and Xd, are matched Determine whether or not, and count the number of matching pairs. That is, if the number of matching pairs is 0, it is determined that the color is 4 colors, 1 is 3 colors, 2 or 3 is 2 colors, and 6 is 1 color, and the number of colors is the color number information Nc. Output.

  The neighborhood coincidence determination unit 1302 generates coincidence pixel presence information 105 and vector information 106 that indicate whether or not a pixel having the same color as the pixel of interest X exists in the surrounding four pixels (Xa, Xb, Xc, and Xd). And output.

  In the neighborhood coincidence determination unit 101 in the first embodiment described above, the surrounding three pixels are referred to. However, in the third embodiment, it is noted that the comparison with the surrounding four pixels including the pixel data Xd is performed. I want.

Further, the vector information 106 depending on the color number information Nc output from the neighborhood match determination unit 1302 and the color number determination unit 1301 in the third embodiment is generated and output. The processing of the neighborhood match determination unit 1302 is divided as follows for the number of colors Nc (1 to 4).
i). When the number-of-colors information Nc is 4, that is, when Xa, Xb, Xc, and Xd are all different colors, the neighborhood coincidence determination unit 1302 performs vector information regardless of coincidence mismatch in the neighboring pixels Xa to Xd. “4” is output as 106.
ii) When the number of colors Nc is 1, that is, when the neighboring pixels Xa to Xd are all pixel values (same color), the pixel of interest X and Xa are compared, and if X = Xa, “0”, X If ≠ Xa, “1” is output as the vector information 106.
iii) When the number of colors Nc is 2, that is, when there are two types of colors in the neighboring pixels Xa to Xd, the first pixel value (color) X1 and the second pixel value in the order of Xa, Xb, Xc, Xd The pixel value X2 is obtained, and “0” is output as the vector information 106 when the pixel of interest X = X1, “1” when X = X2, and “2” when X ≠ X1 and X ≠ X2. Here, the first and second pixel values are defined. When viewed in the order of Xa, Xb, Xc, and Xd, the first pixel X1 is unconditionally Xa. The second pixel value is any one of Xb, Xc, and Xd having a value (color) different from Xa. For example, when Xa = Xb = Xc and Xd does not match any of Xa to Xc, the first pixel value X1 is Xa and the second pixel value X2 is Xd.
iv). When the number of colors Nc is 3, that is, when three types of colors exist in the neighboring pixels Xa to Xd, the first pixel value X1, the second pixel value X2, and the like in the order of Xa, Xb, Xc, Xd, The third pixel value X3 is obtained. If X = X1, “0”, X = X2 is “1”, X = X3 is “2”, X ≠ X1, X ≠ X2, and X ≠ X3 are “ 3 ″ is output as the vector information 106.

  Further, the process when the neighborhood match determination unit 1302 outputs the matched pixel presence information 105 is as follows.

  The neighborhood match determination unit 1302 compares the information output as the vector information 106 with the color number Nc obtained from the color number determination unit 1301, and outputs “1” as the signal line 105 if both are equal. Otherwise, “0” is output. Therefore, when the number of colors Nc = 4, the matching pixel presence information 105 is always “1”. Further, this is a case where the number of colors Nc is other than 4, and if the target pixel X does not match any of the surrounding four pixels Xa, Xb, Xc, and Xd, “1”, and the target pixel X is at least any of the four surrounding pixels. If there is a match, “0” is output.

  The coincidence pixel position encoding unit 1303 encodes the vector 106 from the neighborhood coincidence determination unit 1302 and outputs the code word that is the encoding result to the code generation unit 1305. The matching pixel position encoding unit 1303 in the third embodiment switches between the code tables of FIGS. 15A to 15C depending on the number of colors Nc output from the number-of-colors determination unit 1301. When the color number Nc is 4, the code output from the coincidence pixel position encoding unit 1303 is not performed.

  In the code words in FIGS. 15A to 15C, for example, “1 (binary)” exists when Nc = 1, 2, 3. However, in the decoding apparatus (described later), even if the encoded data of the pixel of interest is this code, any of the codes (a) to (c) in FIG. Since it can be identified, it is promised to decode correctly.

  On the other hand, as in the first embodiment, each component of the pixel of interest is predictively encoded by the component value predicting unit 202, the subtracter 203, and the prediction error encoding unit 1304. In the present embodiment, since the prediction selection signal m is a fixed value 4, prediction values are obtained from the equation “a + b−c” for all component values.

  In the first embodiment, the predictive coding unit 205 performs Huffman coding using the code table of FIG. 6, but in this embodiment, an example using Golomb code will be described.

The Golomb code has a feature that a non-negative integer value can be encoded, and encoding can be performed using a plurality of probability models that differ depending on the parameter variable k. In addition, the Golomb code has an advantage that a code table is unnecessary because a code word can be derived from a symbol to be encoded and a parameter variable k. One form of Golomb code is adopted as a prediction error encoding method in JPEG-LS (ISO / IEC 14495-1 | ITU-T Recommendation T.87), which is an international standard recommendation from ISO and ITU-T. Here, the prediction error e output from the subtracter 203 is converted into a non-negative integer value (V) by the following equation, and then Golomb-coded with the selected k parameter.
When e ≧ 0, V = 2 × e
When e <0, V = −2 × e−1

  The procedure for Golomb encoding the non-negative integer value V to be encoded with the encoding parameter k is as follows.

  First, V is shifted right by k bits to obtain an integer value m. The code for V is composed of a combination of “1” (variable length portion) following m “0” and lower k bits (fixed length portion) of V. FIG. 16 shows an example of Golomb codes at k = 0, 1, and 2. The code configuration method described here is merely an example, and a uniquely decodable code can be configured even if the order of the fixed-length part and the variable-length part is reversed. Also, the code can be configured by reversing 0 and 1. Various methods such as a method of selecting an optimal k parameter in a predetermined unit and including it in the code string can be considered as a method of selecting the encoding parameter k. In this embodiment, encoding is performed by a method similar to JPEG-LS. Use a method of updating in the process. Hereinafter, a method for selecting the encoding parameter k will be described.

  The prediction error encoding unit 1304 includes a counter N that stores the number of encoded pixels, and a counter A [C] that stores the sum of absolute values of encoded prediction errors for each component (C is a component) Number, with 0-2). At the start of encoding, 1 is set in the counter N, and 2 is set in the counters A [0] to A [2]. For each component value to be encoded, N × 2 ^ k (x ^ y means x to the power of y) is determined to obtain the maximum k that does not exceed A [C], and the procedure described above is used. Then, the prediction error e is Golomb encoded and a code word is output. When the information 105 from the neighborhood match determination unit 1302 is “1” (when the same color as the pixel of interest X does not exist in the neighborhood pixels a, b, c, and d), A [ C] is updated by adding the absolute value | e | of the prediction error. In this case, the variable N is updated by adding 1 after the encoding process of all components. In order to limit A [C] and N within a predetermined range, a process of reducing A [C] and N to 1/2 at a timing when N reaches a certain value (for example, 32) is applied. good.

  In accordance with the matching pixel presence information 105, the code generation unit 1305 encodes the code for the pixel of interest from the codeword output from the matching pixel position encoding unit 1303 and the prediction encoded data of each component output from the prediction error encoding unit 1304. Generate data. When the matching pixel presence information 105 is “0”, only the code word output from the matching pixel position encoding unit 1303 is used as the encoded data of the pixel of interest as shown in FIG. On the other hand, when the matching pixel presence information 105 is “1”, each code word output from the prediction error encoding unit 1304 is added to the code word output from the matching pixel position encoding unit 1303 as shown in FIG. The predicted encoded data of the component value is connected to be encoded data of the pixel of interest. When the number of colors Nc detected by the number-of-colors determination unit 1301 is 4, there is no codeword output from the matching pixel position encoding unit 1303 (that is, the codeword length of the neighborhood matching information is 0. Please note that there is.

  In the code string forming unit 104, additional information such as the prediction selection signal m, the number of pixels in the horizontal / vertical direction of the image, the number of components constituting the pixel, the accuracy of each component, and the code output from the code generation unit 1305 A code string of a predetermined format is formed from the digitized data and output to the signal line 207. Also in the third embodiment, a code string having the configuration shown in FIG. 9 can be generated as in the first embodiment. In the image processing apparatus according to the third embodiment, the prediction selection signal m is a fixed value. Therefore, if the prediction selection signal m at the decoding side apparatus is also fixed, it is not necessary to store the prediction selection signal m in the header.

  As described above, in the image processing apparatus according to the third embodiment, a vector corresponding to the number of encoded pixels Xa, Xb, Xc, and Xd around the pixel X to be encoded is used. The presence / absence of an information codeword and the code assignment in vector information encoding are changed. This further improves the reversible compression performance of image data that is likely to have pixels of the same color around the pixel of interest, such as characters, line drawings, and CG images, and the same color around the pixel of interest, such as a natural image. It is possible to perform efficient encoding without generating extra addition even in image data in which there is a low possibility of the presence of pixels.

<Modification of Third Embodiment>
The third embodiment may be realized by software. An example of this will be described below. Since the apparatus configuration when implemented by software is the same as that of FIG. 14 with reference to the modification of the first embodiment, the description thereof will be omitted, and the processing procedure of the encoding application executed on the apparatus will be described.

  The encoding processing procedure in the third embodiment is shown in FIG. As shown in the figure, the flow is substantially the same as the flowchart of FIG. 10 shown as the flow of the encoding process in the first embodiment described above, and thus the same reference numerals are given to the same processes. The difference is that the input of the prediction selection signal m shown in step S1000 in FIG. 10 is not necessary, and a process of obtaining the color number Nc by the color number determination unit 1301 in step S1015 is added. Hereinafter, modifications of the third embodiment will be described. Note that the various types of buffers are secured in the RAM 1402 by the CPU 1401 prior to starting the processing.

  First, the code string forming unit 104 generates and outputs a header including additional information of the encoding target image (step S1001). Next, a counter y that holds the vertical position of the pixel of interest is set to 0 (step S1002), and a counter x that holds the horizontal position of the pixel of interest is set to 0 (step S1003). Similarly, the counter C holding the component number of interest is set to 0 (step S1004).

  The component value P (x, y, C) of the target pixel is input from the signal line 206 and stored in the buffer 201. The component value P (x, y, C) is predictively encoded by the component value predicting unit 202, the subtractor 203, and the prediction error encoding unit 1304, and at the same time, the neighborhood match determining unit 1302 Comparison is performed (step S1005). 1 is added to the component number C of interest (step S1006). If C <3 as compared with 3, which is the number of color components of the RGB image, the process returns to step S1005, and the next component is similarly processed. If not, the process proceeds to step S1008.

  When all the components are input for the pixel of interest, the color number determination unit 1301 obtains the color number Nc (step S1015). Using this, the neighborhood match determination unit 1302 generates the vector information 106 and the match pixel presence information 105. However, when Nc is 4, the vector information is not encoded (step S1008).

  Based on the code word of the neighborhood matching information output from the matching pixel position encoding unit 1303 and the prediction encoded data of each component output from the prediction error encoding unit 1304, the matching pixel exists for the code generation unit 1305. A code is generated for the pixel of interest in accordance with the information 105, and the result is output via the code string forming unit 104 (step S1009).

  Next, 1 is added to the counter x that holds the horizontal position of the pixel of interest (step S1010). If x <W compared to the horizontal pixel count W of the image, the process returns to step S1004, and the next step Pixel encoding processing is performed. If x = W, the process advances to step S1012 to add 1 to the counter y that holds the vertical position of the pixel of interest. Therefore, if y <H compared with y indicating the vertical position of the pixel of interest and the number H of vertical pixels in the image, the process returns to step S1003, and the same process is performed for the pixels on the next line. . If y = H, the encoding for the entire image is complete, and the encoding process is terminated.

  As described above, this modification can also provide the same operational effects as those of the third embodiment.

[Fourth Embodiment]
Next, an example of decoding the encoded image data generated in the third embodiment and its modification will be described as a fourth embodiment. Basically, the process may be performed in the reverse order of the encoding process of the third embodiment.

  FIG. 17 is a block diagram illustrating a functional configuration of an image processing apparatus that decodes image data according to the fourth embodiment. As shown in FIG. 17, the image processing apparatus according to the present embodiment includes a matching pixel position decoding unit 1701, a prediction error decoding unit 1702, a selector 1703, a color number determination unit 1301, a code buffer 1101, a header analysis unit 1102, and a switch 1106. , A buffer 201, a component value prediction unit 202, and an adder 1111. In the figure, reference numerals 1107, 1108, 1109 and 1110 are signal lines. Blocks that perform the same operations as the processing blocks of the image processing apparatuses according to the first to third embodiments are denoted by the same reference numerals, and description thereof is omitted.

Hereinafter, with reference to FIG. 17, the difference between the processing performed by the image processing apparatus according to the fourth embodiment and the image processing apparatus described in the second embodiment will be described.
As in the second embodiment, the encoded data to be decoded is input to the image processing apparatus via the signal line 1107, stored as appropriate in the code buffer 1101, and additional information is acquired by the header analysis unit 1102. Is called.

  Also in the fourth embodiment, the pixel data finally obtained by decoding is output from the switch 1106, but the image data for two lines is stored in the buffer 201 (decoding process is started). Is cleared to zero).

  The number-of-colors determination unit 1301 obtains and outputs the number of colors Nc in the neighboring pixels a, b, c, and d of the pixel of interest by the processing described in the third embodiment.

  When the number of colors Nc output from the number-of-colors determination unit 1301 is 1 to 3, the matching pixel position decoding unit 1701 first selects one of the encoded data of each pixel from FIGS. Whether to use the table is determined, and the code word is decoded into vector information. The decoded vector information is output to the selector 1703 via the signal line 1109.

  Also, the coincidence pixel position decoding unit 1701 compares the decoded vector information and the number of colors Nc, and when both are equal, outputs “1” as a signal 1108 for controlling the prediction error decoding unit 1702 and the switch 1106, and is different. In this case, “0” is output. On the other hand, when the number of colors Nc is 4, the control signal “1” is output from the signal line 1108 and “4” is output to the signal line 1109 as vector information.

When the control signal 1108 from the matching pixel position decoding unit 1701 is “1”, the prediction error decoding unit 1702 performs decoding that is paired with the encoding process of the prediction error encoding unit 1304 described in the third embodiment. The prediction error of each component of the pixel of interest is decoded by the procedure. Similar to the prediction error encoding unit 1304, the prediction error decoding unit 1702 stores a counter N that stores the number of encoded pixels, and the sum of absolute values of encoded prediction errors for each component. Counter A [C] (C is a component number, 0-2). Counter N is set to 1 and counters A [0] -A [2] are set to 2 at the start of decoding. In decoding the component value of interest of the pixel of interest, the same value as the k parameter used at the time of encoding is derived from N and A [0] to A [2] by the same method as at the time of encoding, and this is used. Decodes the non-negative integer value V. The decoding of the Golomb code is performed in the reverse procedure of the encoding. First, the number of consecutive “0” s from the decoding start point is checked and held at the integer value m. K bits are extracted immediately after “1” that terminates the continuation of “0”, m is shifted left by k bits, and an OR operation with the extracted k bits is performed to decode the non-negative integer value V. The prediction error e is decoded from the non-negative integer value V by the following calculation.
When V is an odd number, e = − (V + 1) / 2
When V is an even number, e = V / 2
Similar to the second embodiment, the component value predicting unit 202 generates a predicted value p, and an adder 1111 adds the predicted value p and the prediction error e, thereby restoring the target color component of the target pixel. And output. After decoding each component, update the counter N [C] by adding the absolute value | e | of the decoded prediction error, and update the counter N by adding 1 when all the components have been decoded for the pixel of interest. To do. In order to limit A [C] and N within a predetermined range, a process of reducing A [C] and N to 1/2 at a timing when N reaches a certain value (for example, 32) is applied. good. However, in order to obtain the same k parameter, this processing must be performed in common on the encoding side and the decoding side.

  The selector 1703 selects and outputs one of the component values of the surrounding pixels Xa, Xb, Xc, and Xd of the pixel of interest X read from the buffer 201 in accordance with the vector information 1109 from the matching pixel position decoding unit 1701. That is, if the vector information 1109 is “0”, each component value of Xa is selected, and if it is “1”, the second value different from Xa in the order of Xa, Xb, Xc, Xd. Each component value of a pixel having a pixel value (color) is selected. Similarly, when “2” is selected, each component value of the first pixel value (Xa) and the third pixel value different from the second pixel value is selected in the order of Xa, Xb, Xc, and Xd. And output. When the vector information 1109 is “3” or “4”, the output from the selector 1703 is invalid. In this case, since the switch 1106 is connected to the terminal a, the output from the selector 1703 may be anything.

  When the signal line 1108 is “1” from the matching pixel position decoding unit 1701, the switch 1106 selects the terminal a and outputs pixel data obtained by predictive decoding. When the signal 1108 is “0”, the terminal b is selected, and pixel data obtained based on the vector information is output as a decoding result. As described above, the pixel data output from the switch 1106 is stored in the buffer 1201 and used as surrounding pixels when decoding encoded data to be subsequently input.

  As described above, the original image data can be completely restored from the encoded data.

<Modification of Fourth Embodiment>
The third embodiment may be realized by software. An example of this will be described below.

  Note that the processes corresponding to the various processing units shown in FIG. 17 are realized by functions, subroutines, etc. on the software, but in this modification as well, the names of the respective processing units in FIG. 17 are used as they are for convenience. It shall be. Since the apparatus configuration is the same as in FIG. 14, the description thereof is omitted, and the processing procedure of the decryption application executed on the apparatus will be described with reference to the flowchart of FIG.

  As shown in FIG. 23, the processing procedure is almost the same as the flowchart of FIG. 12 described as the flow of the decoding processing of the second embodiment. Therefore, the same processes as those in the flowchart of FIG. Further, as a particularly different point, a process for detecting the number of colors Nc in step S1219 is added. In step S1205, vector information is decoded according to the number of colors Nc, and in step S1206, decoding is performed in step S1220. It is determined whether or not the vector information obtained in this way and the number of colors Nc are equal. Note that the program according to the figure is loaded in the RAM 1402, and the CPU 1401 executes it.

  First, encoded data to be decoded is stored in the code buffer 1101 from the signal line 1107, and the header analysis unit 1102 analyzes the header of the encoded data to extract additional information necessary for decoding (step S1201). A counter y that holds the vertical position of the pixel of interest is set to 0 (step S1202), and a counter x that holds the horizontal position is set to 0 (step S1203). Further, the counter C holding the component number of interest is set to 0 (step S1204).

  In decoding each pixel, first, the color number determination unit 1301 obtains the color number Nc of surrounding pixels, and the matching pixel position decoding unit 1701 decodes the vector information 1109 corresponding to the value of Nc. At this time, matched pixel presence information 1108 is also generated and output as a control signal to the prediction error decoding unit 1702 and the switch 1106 (step S1205).

  Next, it is determined whether or not the vector information obtained by decoding is equal to the number of colors Nc (step S1220). If they are equal (YES), the process proceeds to step S1207, and if they are not equal (NO), the process proceeds to step S1208. Move.

  When the process moves to step S1207 (at this time, the signal line 1108 is “1”), the terminal a of the switch 1106 is selected, and the component value P of interest is selected by the component value prediction unit 202, the prediction error decoding unit 1702, and the adder 1111. (X, y, C) is decoded. On the other hand, if NO is determined in step S1220 (the signal line 1108 is “0” at this time), the process proceeds to step S1208, the terminal b of the switch 1106 is selected, and the vector information decoded by the matching pixel position decoding unit 1701 is added. Based on this, the selector 1703 selects any one component value of peripheral pixels read from the buffer 1201 and outputs the result as the component value P (x, y, C) of the pixel of interest. In this way, the component value P (x, y, C) output in step S1207 or step S1208 is output from the signal line 1110 to the outside of the apparatus and stored in the buffer 201.

  In step S1209, 1 is added to the counter C that holds the component number of interest, and if C <3 as compared with 3, which is the number of color components of the RGB image, the process returns to step S1206 to decode the next component. To do. If it is determined that C = 3, the process proceeds to step S1211, and 1 is added to the counter x that holds the horizontal position of the pixel of interest. If it is determined in step S1212 that x <W, the process returns to step S1204, and the next pixel is decoded. If it is determined that x = W, the process proceeds to step S1213, and 1 is added to the counter y that holds the vertical position of the pixel of interest. Next, if y <H compared with the number H of vertical pixels of the image, the process returns to step S1203, and the same process is performed for each pixel of the next line. Also. If it is determined that y = H, it means that the decoding process for the entire image has been completed, and thus this decoding process ends.

  With the above operation, the original image data can be completely restored from the encoded data.

<Fifth Embodiment>
In the image processing apparatus according to the third embodiment, whether or not to encode vector information according to the number of colors of the encoded surrounding pixel group for the pixel of interest, and a table for encoding are displayed. Switched. As a result, an increase in the amount of code due to encoding of vector information can be suppressed even for an image that is unlikely to have surrounding pixels having the same pixel value (color) as the target pixel value, such as a natural image. did it. Further, for computer graphics images and the like that are likely to have neighboring pixels having the same pixel value (color) as the target pixel value, further compression efficiency can be expected. In order to realize a further compression rate, it is also effective to combine with run length coding. Therefore, as the fifth embodiment, an example in which run-length encoding is combined with the third embodiment will be described.

  FIG. 18 is a block diagram showing a functional configuration of an image processing apparatus that performs image coding according to the fifth embodiment. As illustrated in FIG. 18, the image processing apparatus according to the fifth embodiment includes a run length encoding unit 1801, a switch 1802, a color number determination unit 1301, a neighborhood match information encoding unit 1306, and a component value predictive encoding unit. 1307, a code generation unit 1305, a buffer 201, and a code string formation unit 104. Reference numerals 105, 206, and 207 denote signal lines. Blocks that perform the same operations as the processing blocks of the image processing apparatus according to the first embodiment are denoted by the same reference numerals and description thereof is omitted.

  Hereinafter, with reference to FIG. 18, an image encoding process in the image processing apparatus according to the fifth embodiment will be described.

  The image processing apparatus of the fifth embodiment has a configuration in which a run-length encoding unit 1801 and a switch 1802 are added to the image processing apparatus of the third embodiment shown in FIG. In FIG. 13, the neighborhood match information encoding unit 1306 is a combination of broken line portions including the neighborhood match determination unit 1302 and the match pixel position encoding unit 1303 in FIG. 13, and the component value prediction encoding unit 1307 is the same. The broken line portion including the component value prediction unit 202, the subtractor 203, and the prediction error encoding unit 1304 in the figure is grouped.

  In the fifth embodiment, the image data to be encoded is CMYK color image data, and is composed of pixel data expressing a density value in the range of 0 to 255 with 8 bits of each component (color component). Shall be. 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 C, M, Y, and K. The image is assumed to be composed of horizontal W pixels and vertical H pixels. Note that C (cyan), M (magenta), Y (yellow), and K (black) are used as the color space in the embodiments so far, the encoding target image is RGB color image data, and other examples are shown. Is for.

  Next, the operation of each unit in the image processing apparatus according to the fifth embodiment will be described. As in the third embodiment described above, the prediction method used in the component value prediction unit 202 (inside the component value prediction encoding unit 1307) is assumed to be fixed, and the prediction selection signal m = 4 is always input. Shall.

  The encoding target image data is sequentially input from the signal line 206. The input order of the pixel data is a raster scan order, and component data is input in the order of C, M, Y, and K for each pixel. For the density values of each component of the pixel, C, M, Y, and K are defined as component numbers 0, 1, 2, and 3, respectively, and the upper left corner of the image is the coordinate (0, 0). The value of the component number C of the pixel at the direction pixel position y is represented as P (x, y, C). For example, the values of the C, M, Y, and K components of the pixel at the horizontal pixel position 104 and the vertical pixel position 335 are P (104, 335, 0), P (104, 335, 1), and P (104, respectively. , 335,2) and P (104,335,3). Further, when pixel values are expressed as a group of components, they are expressed as a vector X having C, M, Y, and K component values as elements. For example, a pixel having C, M, Y, and K component values of c, m, y, and k is described as X = (c, m, y, k).

  In this image processing apparatus, a code corresponding to the run length encoding is basically output from the code string forming unit 104 for each pixel of image data sequentially input from the signal line 206. Then, a code is output for a run (number of consecutive) of pixels.

  The buffer 201 has an area for storing image data for two lines, and sequentially stores the image data input from the signal line 206. As described above, in the image data, the component values are input in the order of raster in each pixel and in the order of C, M, Y, and K in each pixel, but from the buffer 201, a, b, The same component values of the surrounding three pixels indicated by c and d are read out and supplied to the color number determination unit 1301, the neighborhood match information encoding unit 1306, and the component value predictive encoding unit 1307. When the component value of the pixel of interest is P (x, y, C), a = P (x-1, y, C), b = P (x, y-1, C), c = P (x −1, y−1, C), d = P (x + 1, y + 1, C). When a, b, c, and d are outside the image at the first line of the image or at the beginning / end of each line, a common value is assumed on the encoding side and the decoding side. In this embodiment, it is 0.

For each component of the pixel X of interest, the color number determination unit 1301 reads the same component values of the surrounding four pixels indicated by a, b, c, and d in FIG. 3 from the buffer 201 to obtain pixels Xa, Xb, Xc, and Xd. . The position of the pixel of interest is (x, y), and Xa, Xb, Xc, Xd are expressed using this as follows.
Xa = (P (x-1, y, 0), P (x-1, y, 1), P (x-1, y, 2), P (x-1, y, 3))
Xb = (P (x, y-1,0), P (x, y-1,1), P (x, y-1,2), P (x, y-1,3))
Xc = (P (x-1, y-1,0), P (x-1, y-1,1), P (x-1, y-1,2), P (x-1, y- 1,3))
Xd = (P (x + 1, y-1,0), P (x + 1, y-1,1), P (x + 1, y-1,2), P (x + 1, y- 1,3))

  The number-of-colors determination unit 1301 detects how many kinds of colors exist in the set of Xa, Xb, Xc, and Xd, and outputs the number of colors Nc. Specifically, for 6 combinations that extract 2 pixels from 4 pixels, the number of the extracted 2 pixels is the same. If 0, 4 colors, 1 if 3 colors, 2 or 3 If there are two colors and six, it is determined that there is one color, and the number of colors Nc is output. This is the same as in the third embodiment.

  When the number of colors Nc of the surrounding four pixels output from the color number determination unit 1301 is 1 to 3, the neighborhood match information encoding unit 1306 is similar to the third embodiment, and the target pixel X and the surrounding four pixels The vector information for specifying the coincidence / non-coincidence with (Xa, Xb, Xc, Xd) and the neighboring pixels in the case of coincidence is encoded, and the code word is output. The signal line 105 outputs peripheral pixel state information indicating whether or not there is a pair having the same color in the four surrounding pixels.

  On the other hand, as in the third embodiment, the C, M, Y, and K components of the pixel of interest are encoded by the component value predictive encoding unit 1307. It should be noted that a regular mode encoding method in the international standard method JPEG-LS may be applied to the predictive encoding of each component in the component value predictive encoding unit 1307.

  The code generation unit 1305 performs predictive encoding of the vector information codeword output from the neighborhood match information encoding unit 1306 and each component output from the component value prediction encoding unit 1307 according to the neighboring pixel state information of the signal line 105. Encoded data for the pixel of interest is generated from the data. The operation of the code generation unit 1305 is as described in the third embodiment.

  The run-length encoding unit 1801 stores therein a counter RL that measures the number of consecutive identical pixel values, and starts and continues measurement when the color number Nc output from the color number determination unit 1301 becomes 1. However, when the pixel X of interest is different from the previous pixel value Xa or reaches the last pixel of one line, the run length held in the counter RL is encoded and output. . For example, various methods can be applied to the run-length encoding. Here, the same method as the run-length encoding by the run mode in the international standard method JPEG-LS is used. Is omitted. Here, for the sake of simplicity, the description has been made on the assumption that the continuous number of pixels is started based on the number of colors Nc output from the number-of-colors determination unit 1301. Any condition that can predict the occurrence of (continuation of pixel values) may be used. For example, a method may be used in which run measurement is started when the pixel immediately before Xa is Xe and Xa = Xe.

  The switch 1802 is connected to the terminal b side when the number of colors Nc output from the color number determination unit 1301 is 1, and is fixed to the b side while the run length encoding unit 1801 is counting the run length. To do. After the run length is terminated and a code is output from the run length encoding unit 1801, the terminal is switched to the terminal a.

  As described above, the image processing apparatus according to the fifth embodiment includes the neighborhood match information encoding unit 1306, the component value prediction encoding unit 1307, and the run length encoding unit 1801, The encoding is performed by selectively operating these encoding methods according to the number of colors of the encoded pixels Xa, Xb, Xc, and Xd around the pixel X. Since these selections are performed by encoded pixels, it is not necessary to transmit information for switching the encoding method. In addition to the same effects as those of the third embodiment, by introducing run-length encoding, lossless compression performance such as characters, line drawings, and CG images can be further improved.

<Modification of Fifth Embodiment>
The fifth embodiment may be realized by software. An example of this will be described below. Since the apparatus configuration when implemented by software is the same as that of FIG. 14 with reference to the modification of the first embodiment, the description thereof will be omitted, and the processing procedure of the encoding application executed on the apparatus will be described.

  An encoding process procedure according to the fifth embodiment is shown in a flowchart of FIG. FIG. 19 is a flowchart showing the flow of encoding processing by the image processing apparatus of this embodiment. The program according to the figure is loaded into the RAM 1402 and is executed by the CPU 1401. When realized by software, the processing corresponding to each processing unit shown in FIG. 18 is realized by a function, a subroutine, or the like on the software, but in this modification as well, for convenience, the name of each processing unit in FIG. Use as is.

  First, the code string forming unit 104 generates and outputs a header including additional information of the encoding target image (step S1901). Next, a counter y that holds the vertical position of the pixel of interest is set to 0, and a counter RL that is held inside the run-length encoding unit 1801 is set to 0 (step S1902). Further, a counter x that holds the horizontal position of the pixel of interest is set to 0 (step S1903).

  Next, for the four pixels around the pixel of interest X located at the coordinates (x, y) coordinates, the color number determination unit 1301 determines the color number Nc (step S1904). If the number of colors Nc is 1 or the counter RL is a value other than 0 (YES), the process proceeds to step S1914. Otherwise (NO), the process proceeds to step S1906 (step S1905). . If the determination in step S1905 is YES, switch 1802 selects terminal b, and if NO, terminal a is selected.

  In step S1906, the neighborhood coincidence information encoding unit 1306 generates and encodes vector information for identifying the coincidence / non-coincidence with the surrounding 4 pixels, and the matching surrounding pixel position when they coincide, and also encodes the surrounding 4 pixels. Neighboring pixel state information indicating whether or not a pair of the same color exists is generated and output as a signal line 105.

  Next, it is determined whether or not the control signal 105 is “1” (step S1907). If it is “1” (YES), the component value predictive coding unit 1307 performs the component value predictive coding process. It performs with respect to each component of C, M, Y, and K, and produces | generates a code | symbol (step S1908). Subsequently, the code generation unit 1305 generates a code for the target pixel in accordance with the control signal from the signal line 105, and outputs the code to the code string formation unit 104 via the switch 1802. A code string is formed (step S1909).

  Thereafter, 1 is added to the counter x that holds the horizontal position of the pixel of interest (step S1910). If x <W as compared with the number of horizontal pixels W of the image (YES), the process proceeds to step S1904. Then, the next pixel is encoded. If x = W, the process proceeds to step S1912 (step S1911).

  On the other hand, when the processing proceeds from step S1905 to step S1914, the pixel value X of interest is compared with the immediately preceding pixel value Xa. If X ≠ Xa (NO), the process proceeds to step S1920. If X = Xa (YES), the process proceeds to step S1915.

  When the processing proceeds to step S1915, the counter RL held in the run-length encoding unit 1801 is incremented by 1 (step S1915), and then 1 is added to the counter x that holds the horizontal position of the target pixel (step S1915). S1916). Next, if x <W as compared with the number of pixels W in the horizontal direction of the image, the process returns to step S1904, and the next pixel is encoded. If it is determined that x = W, since the run has reached the right end of the image, the run length is determined at this point, and the run length held in the counter RL is encoded by the run length encoding unit 1801. Output the sign. The code output from the run-length encoding unit 1801 is sent to the code string forming unit 104 via the switch 1802 to form a code string of a predetermined format (step S1918). When the run-length encoding is completed, the counter RL is returned to 0 (step S1919). At this time, the switch 1802 is changed to the terminal a side. Thereafter, the process proceeds to step S1912, and the target of the encoding process is moved to the next line.

  Further, when the process proceeds from step S1914 to step S1920, it means that the run is terminated due to the appearance of a pixel value X different from the immediately preceding pixel value Xa, and therefore the run length encoding unit 1801 holds the counter RL. The run length is encoded and the code is output. The code output from the run-length encoding unit 1801 is sent to the code string forming unit 104 via the switch 1802 to form a code string of a predetermined format. When the run-length encoding is completed, the counter RL is returned to 0 (step S1921). At this time, the switch 1802 is changed to the terminal a side. Subsequently, the process proceeds to step S1906, and the encoding process for the pixel of interest X in which the run is terminated is continued.

  When the encoding of one line of image data is thus completed, the process advances to step S1912 to add 1 to the counter y that holds the vertical position of the pixel of interest, and then compare it with the number of vertical pixels H of the image. If y <H (YES), the process returns to step S1903, and the same process is performed for pixels on the next line. If it is determined that y = H, this means that all the encoding of the image data to be encoded has been completed, and the encoding process is terminated (step S1913).

  As described above, the present modification (example realized by software) can achieve the same effects as those of the fifth embodiment.

<Sixth Embodiment>
An apparatus for decoding the encoded data generated in the fifth embodiment (and its modifications) will be described as a sixth embodiment.

  FIG. 20 is a block diagram of an image processing apparatus that decodes encoded image data according to the sixth embodiment. As illustrated in FIG. 20, the image processing apparatus according to the sixth embodiment includes a color number determination unit 1301, a code buffer 1101, a header analysis unit 1102, a switch 1106, a buffer 201, and a component value prediction encoded data decoding unit 1704. , A neighborhood match information decoding unit 1705, a run length decoding unit 2001, and a switch 2002. In the figure, reference numerals 1107, 1108, and 1110 denote signal lines. Blocks that operate in the same manner as the processing blocks of the conventional image processing apparatus and the image processing apparatuses of the first to fifth embodiments are denoted by the same reference numerals, and description thereof is omitted.

  The image decoding process performed by the image processing apparatus according to this embodiment will be described below with reference to FIG. The image processing apparatus of the sixth embodiment has a configuration in which a run-length decoding unit 2001 and a switch 2002 are added to the image processing apparatus of the fourth embodiment shown in FIG. The neighborhood coincidence information decoding unit 1705 is a combination of broken line portions including the coincidence pixel position decoding unit 1701 and the selector 1703 in FIG. 17, and the component value predictive encoded data decoding unit 1704 is a component value shown in FIG. A broken line portion including the prediction unit 202, the prediction error decoding unit 1702, and the adder 1111 is gathered together.

  Similar to the fourth embodiment, encoded data to be decoded is input to the image processing apparatus via a signal line 1107 and stored in the code buffer 1101 as appropriate. The header analysis unit 1102 analyzes the header part at the head of the encoded data and acquires information necessary for the decoding process.

  The color number determination unit 1301 obtains and outputs the color number Nc by the processing described in the third embodiment.

  When the surrounding color number Nc output from the color number determination unit 1301 is a prime color number Nc from 1 to 3, the neighborhood match information decoding unit 1705 performs the same processing as in the fourth embodiment, and the pixel of interest X and the surrounding 4 pixels. Match / mismatch with (Xa, Xb, Xc, Xd), and if they match, vector information that identifies peripheral pixels having the same color as the pixel of interest is decoded. If the vector information obtained by decoding is different from the number of colors Nc, one of Xa, Xb, Xc, and Xd is output according to the vector information. The neighborhood coincidence information decoding unit 1705 outputs neighborhood pixel state information indicating whether a pair having the same color exists in the four pixels around the pixel of interest, that is, whether the vector information and Nc are equal. Output to line 1108.

  The component value predictive encoded data decoding unit 1704 reads the encoded data from the code buffer 1101 and decodes the C, M, Y, and K components of the pixel as in the fourth embodiment.

  On the other hand, the run-length decoding unit 2001 reads the encoded data from the code buffer 1101 when the color number Nc output from the color number determination unit 1301 is 1, and calculates the continuous number RL of the same pixel value as the previous pixel Xa. Decode and output RL consecutive Xa as decoded pixel values. The decoding process of the run-length decoding unit 2001 is paired with the encoding process of the run-length encoding unit 1801 in the fifth embodiment.

  The switch 2002 is connected to the terminal b side when the color number Nc output from the color number determination unit 1301 is 1. At this time, while output of RL pieces of Xa decoded by the run length decoding unit 1801 is performed, the output is maintained on the b side, and when output of RL pieces of Xa is completed, the terminal a is connected.

<Modification of Sixth Embodiment>
FIG. 21 is a flowchart showing a processing procedure when the processing of the sixth embodiment is realized by software. Since the apparatus configuration when implemented by software is the same as that of FIG. 14 referring to the modification of the first embodiment, description thereof will be omitted, and the processing procedure of the decryption application executed on the apparatus will be described.

  The program according to the figure is loaded into the RAM 1402 and is executed by the CPU 1401. When realized by software, the processing corresponding to each processing unit shown in FIG. 20 is realized by a function, a subroutine, or the like on the software, but in this modification as well, for convenience, the name of each processing unit in FIG. Use as is.

  When encoded data to be decoded is input to the code buffer 1101 from the signal line 1107, the header analysis unit 1102 analyzes the header of the encoded data and extracts additional information necessary for decoding (step S2101). Next, a counter y that holds the vertical position of the pixel of interest is set to 0 (step S2102), and a counter x that holds the horizontal position of the pixel of interest is set to 0 (step S2103). Focusing on the pixel located at the (x, y) coordinate, the number of colors Nc is obtained with reference to the surrounding pixels Xa to Xd decoded by the number-of-colors determination unit 1301 (step S2104). If the number of colors Nc is 1 (YES), the process proceeds to step S2114, and otherwise (NO), the process proceeds to step S2106 (step S2105). If the determination in step S2105 is YES, switch 2002 is connected to terminal b, and if NO, it is connected to terminal a.

  In step S <b> 2106, the neighborhood match information decoding unit 1705 decodes the vector information that identifies the match / mismatch with the surrounding four pixels and specifies the neighborhood pixel if they match. If the vector information obtained by decoding and the number of colors Nc are different values, any one of Xa to Xd is selected and output according to the vector information. Further, information indicating whether the vector information and the number of colors Nc are equal is output to the signal line 1108 as neighboring pixel state information. It is determined whether or not the signal line 1108 is “1” (step S2107). If it is “1” (YES), the component value predictive encoded data decoding unit 1704 performs C, M, Y, For each of the K components, a component value is decoded from the predicted encoded data (step S2108). When the signal line 1108 is “0”, the switch 1106 is connected to the terminal a. When the signal line 1108 is “1”, the switch 1106 is connected to the terminal b, and the neighborhood match information decoding unit 1705 or the component value predictive encoding is connected. The pixel value output from the data decoding unit 1704 is output as the decoded pixel value at the position (x, y) from the signal line 1110 via the switches 1106 and 2002 (step S2109).

  Next, 1 is added to the counter x that holds the horizontal position of the pixel of interest (step S2110), and if x <W compared with the number of horizontal pixels W of the image (YES), the process proceeds to step S2104. Then, the next pixel is decoded. If x = W, the process proceeds to step S2112.

  On the other hand, when the processing moves to step S2114, the run length RL is decoded inside the run length decoding unit 2001. The decrypted RL is compared with 0 (step S2115). If RL = 0 (YES), the process proceeds to step S2106. In this case, the connection of the switch 2002 is changed from the terminal b to the terminal a. If RL ≠ 0 (NO), Xa is output as the decoded pixel value at the position (x, y) from the signal line 1110 via the switch 2002 (step S2116). Then, 1 is subtracted from the counter RL (step S2117), and 1 is added to the counter x that holds the horizontal position of the pixel of interest (step S2118). In step S2119, the counter x is compared with the horizontal pixel count W of the image. If x <W (YES), the process returns to step S2115, and if x = W, the process proceeds to step S2112.

  When the decoding process for one line is completed as described above, in step S2112, 1 is added to the counter y that holds the vertical position of the pixel of interest, and in step S2113, the number of vertical pixels in the image is determined. Compare with H. If y <H (YES), the process returns to step S2103, and the same process is performed for the pixels on the next line. If y = H, it means that the decoding process for the entire image has been completed, and the decoding process is terminated.

  With the above operation, the same operational effects as in the sixth embodiment can be achieved.

  The present invention is not limited to the embodiments described above. The application example will be described below.

  In the third to sixth embodiments, according to the number of colors Nc of the surrounding pixels, the vector information for identifying the coincidence / non-coincidence with Nc types of pixel values and the position of the neighboring pixels when they coincide is encoded. Therefore, Nc + 1 value symbols are encoded as vector information. For example, if the number of colors Nc is 3, 0 if it matches the first pixel value X1, 1 if it matches the second pixel value X2, and 1 if it matches the third pixel value X3. In FIG. 15C, a quaternary symbol is encoded by using the code table of FIG. 15C, such as 3 when it does not match any of 2 and X1 to X3. However, it is not necessary to make all Nc types of pixel values subject to matching / mismatching. For example, when Nc = 3, it is determined whether the first pixel value X1 and the second pixel value X2 match or do not match. When X = X1, 0, when X = X2, 1, X ≠ When X1 and X ≠ X2, a symbol having three values of 0, 1, and 3 is coded such as 3 (in consideration of the process of comparing the vector information and Nc, “2” has been excluded). Anyway.

  Further, the method of obtaining the first pixel value X1, the second pixel value X2, the third pixel value X3, and the like is not limited to the above-described embodiment, and for example, the order of Xb, Xa, Xc, Xd is different. The first, second, and third pixel values may be acquired in order, or may be changed depending on the matching status among Xa, Xb, Xc, and Xd. For example, the first, second, and third pixel values are acquired in the order of Xa, Xb, Xc, and Xd. If Xa = Xb, the second pixel value X2 is set to Xd. Absent.

  In addition, as a method of encoding vector information, a method using a code word determined in advance assuming a probability distribution has been shown. However, a code word different from the above example may be used, or a different algorithm such as an arithmetic code may be used. An encoding technique may be applied.

  In addition, as a prediction method of a component value to be focused on, several prediction methods may be prepared and switched adaptively, or a component that focuses on an average value of prediction errors generated in each encoded component value It is also possible to use non-linear prediction such as feedback to the prediction.

  Although an example using Huffman coding or Golomb coding as the entropy coding of the component value prediction error is shown here, other entropy coding may be used.

  In addition, an example in which Xa, Xb, Xc, and Xd are referred to as the surrounding pixel values of the pixel of interest X has been described. However, more pixels may be referred to, or the number of reference pixels may be reduced such as only Xa and Xb. It doesn't matter.

  In addition, although description has been made by taking, as an example, encoding of an RGB image and a CMYK image as a color space of a color image, any image data composed of a plurality of components such as a Lab or a YCrCb image can be applied. The present invention is not limited. Further, the example in which the image data is input in the raster scan order has been described. However, the present invention is applied to a method in which an image is decomposed into bands each having a predetermined number of lines and the vertical direction is preferentially scanned in units of bands. It doesn't matter. In such a case, the reference pixel may be changed from the encoded pixel according to the scanning method.

  In the above-described embodiment, the encoding device and the decoding device have been described as different examples. However, it goes without saying that these two functions may be provided in one device and common portions may be integrated.

  Further, the present invention can be applied to a system constituted by a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.), but an apparatus (for example, a copying machine, a facsimile machine, etc.) comprising a single device. You may apply to.

  As is clear from the description of the above embodiment, the present invention can also be applied to software executed on a general-purpose information processing apparatus such as a personal computer. In addition, software (including applications) is usually stored in a computer-readable storage medium such as a CD-ROM, and can be executed by setting the storage medium in a computer and copying or installing it in the system. Of course, such computer-readable storage media are clearly within the scope of the present invention.

It is a block diagram which shows the function structure of the image processing apparatus which concerns on 1st Embodiment. It is a block diagram of the conventional image processing apparatus. It is a figure which shows the position of the surrounding pixels a, b, and c with respect to the encoding object pixel x. The prediction formula corresponding to the prediction selection signal m is shown. It is a figure which shows a response | compatibility with the prediction error e and group number SSSS. It is a figure which shows a response | compatibility with group number SSSS and a code word. It is a figure which shows the correspondence of neighborhood match information and a code word. It is a figure which shows the structure of pixel coding data. It is a figure which shows the structure of the output code sequence of this image processing apparatus. It is a flowchart which shows the flow of the encoding process of the encoding target image data by the image processing apparatus of 1st Embodiment. It is a block diagram which shows the function structure of the image processing apparatus which concerns on 2nd Embodiment. It is a flowchart which shows the flow of the decoding process by the image processing apparatus of 2nd Embodiment. It is a block diagram which shows the function structure of the image processing apparatus which concerns on 3rd Embodiment. It is an apparatus block diagram at the time of encoding and decoding by software in each embodiment. It is a figure which shows the code table used for the encoding of neighborhood matching information in the image processing apparatus which concerns on 3rd, 5th embodiment. It is a figure which shows the example of a Golomb code. It is a block diagram which shows the function structure of the image processing apparatus which concerns on 4th Embodiment. It is a block diagram which shows the function structure of the image processing apparatus which concerns on 5th Embodiment. It is a flowchart which shows the flow of the encoding process of the encoding target image data by the image processing apparatus of 5th Embodiment. It is a block diagram which shows the function structure of the image processing apparatus which concerns on 6th Embodiment. It is a flowchart which shows the flow of the decoding process by the image processing apparatus of 6th Embodiment. It is a flowchart which shows the flow of the encoding process of the encoding target image data by the image processing apparatus of 3rd Embodiment. It is a flowchart which shows the flow of the decoding process by the image processing apparatus of 4th Embodiment.

Claims (12)

An image encoding device that inputs image data in which one pixel is composed of a plurality of color component data and performs lossless encoding,
For each of a plurality of color component data constituting the input pixel-of-interest data, a prediction code that obtains a prediction value by referring to the color component data of an encoded pixel in the vicinity of the pixel of interest and generates predictive encoded data And
A color number counting means for counting the number of colors represented by a plurality of pixels in the vicinity of the pixel of interest and at the encoded pixel position, and generating color number information;
Wherein the target pixel data, wherein there in the vicinity of the target pixel is compared with the plurality of pixel data each other in the encoded pixel position, whether neighboring pixels of the same color as the target pixel is present, and the presence a vector information generating means for generating a vector information specifying the relative positions of the corresponding neighboring pixels for the target pixel in the case of,
Vector information encoding means for encoding the vector information;
Based on the color number information obtained from the color number counting means and the vector information obtained by the vector information generation means, output from the encoded data obtained by the predictive coding means and the vector information coding means. Encoded data generation means for generating encoded data for use,
The encoded data generation means includes
Number of colors represented by the color number information is equal to or less than a predetermined threshold value, and, if the vector information indicates the presence of a neighboring pixel having the same color as the target pixel, the generated vector information encoding means encodes data and then output as encoded data of the target pixel,
In the counted number of colors than the predetermined threshold value by the number of colors counting means, and, if the vector information indicates absence of neighboring pixels having the same color as the target pixel, generated by the vector information encoding means predictive coded data and coded data said generated by predictive coding means, and output as coded data of the target pixel,
If the number of colors that are counted in the number of colors counting means exceeds a predetermined threshold value, image, characterized in that the predictive coding data generated by said prediction encoding means, and outputs it as coded data of the target pixel Encoding device. The vector information generating means generates vector information corresponding to the number of colors indicated by the color number information obtained by the color number counting means,
The vector information encoding means generates encoded data of the vector information using a code configured according to a probability distribution assumed for each number of colors indicated by the color number information. The image encoding device according to claim 1. When a sequence of pixels of the same color is expected from encoded pixels, the number of subsequent input pixels having the same color as a predetermined encoded pixel is counted, and the number of consecutive pixels of the same color counted Run-length encoding means for outputting encoded data;
At the timing when the continuity encoded data of the same color pixel is generated by the run-length encoding means, the encoded data from the run-length encoding means is selected and output, and the same from the encoded pixels. The image code according to claim 1, further comprising: a switching unit that selects and outputs the encoded data from the encoded data generation unit when continuation of color pixels is not expected. Device. A method of controlling an image encoding device that inputs image data in which one pixel is composed of a plurality of color component data and performs lossless encoding,
For each of a plurality of color component data constituting the input pixel-of-interest data, a prediction code that obtains a prediction value by referring to the color component data of an encoded pixel in the vicinity of the pixel of interest and generates predictive encoded data Conversion process,
A color number counting step of counting the number of colors represented by a plurality of pixels in the vicinity of the pixel of interest and at the encoded pixel position, and generating color number information;
Wherein the target pixel data, wherein there in the vicinity of the target pixel is compared with the plurality of pixel data each other in the encoded pixel position, whether neighboring pixels of the same color as the target pixel is present, and the presence a vector information generation step of generating vector information specifying the relative positions of the corresponding neighboring pixels for the target pixel in the case of,
A vector information encoding step for encoding the vector information;
Based on the color number information obtained in the color number counting step and the vector information obtained in the vector information generation step, output of the encoded data obtained in the predictive coding step and the vector information coding step and a coded data generating step of generating encoded data of use,
The encoded data generation process includes:
Number of colors represented by the color number information is equal to or less than a predetermined threshold value, and, when said vector information indicates the presence of a neighboring pixel having the same color as the target pixel, the generated vector information encoding step the coded data and then output as encoded data of the target pixel,
In the counted number of colors than the predetermined threshold value by the number of colors counting step, and, if the vector information indicates absence of neighboring pixels having the same color as the target pixel, generated by the vector information encoding step predictive encoded data generated by the encoded data and the predictive coding step, and outputs the encoded data of the target pixel,
If the number of colors that are counted in the number of colors counting step exceeds a predetermined threshold value, image, characterized in that the predictive coding data generated by the predictive coding process, and outputs it as coded data of the target pixel A method for controlling an encoding device. By causing read Komase executed by a computer, the computer, one pixel receives image data that is composed of a plurality of color component data, a computer program that causes functions as an image coding apparatus for lossless encoding,
For each of a plurality of color component data constituting the input pixel-of-interest data, a prediction code that obtains a prediction value by referring to the color component data of an encoded pixel in the vicinity of the pixel of interest and generates predictive encoded data And
Color number counting means for counting the number of colors represented by a plurality of pixels at the encoded pixel position in the vicinity of the target pixel and generating color number information;
Wherein the target pixel data, wherein there in the vicinity of the target pixel is compared with the plurality of pixel data each other in the encoded pixel position, whether neighboring pixels of the same color as the target pixel is present, and the presence a vector information generating means for generating a vector information specifying the relative positions of the corresponding neighboring pixels for the target pixel in the case of,
Vector information encoding means for encoding the vector information;
Based on the color number information obtained from the color number counting means and the vector information obtained by the vector information generation means, output from the encoded data obtained by the predictive coding means and the vector information coding means. Function as encoded data generation means for generating encoded data for
The encoded data generation means includes
Number of colors represented by the color number information is equal to or less than a predetermined threshold value, and, if the vector information indicates the presence of a neighboring pixel having the same color as the target pixel, the generated vector information encoding means encodes data and then output as encoded data of the target pixel,
In the counted number of colors than the predetermined threshold value by the number of colors counting means, and, if the vector information indicates absence of neighboring pixels having the same color as the target pixel, generated by the vector information encoding means predictive coded data and coded data said generated by predictive coding means, and output as coded data of the target pixel,
If the number of colors that are counted in the number of colors counting means exceeds a predetermined threshold value, characterized in that makes function the predictive encoded data generated by said prediction encoding means, as output as coded data of the target pixel A computer program.   A computer-readable storage medium storing the computer program according to claim 5. An image decoding apparatus that decodes encoded image data in which one pixel is composed of a plurality of color component data and is encoded in a predetermined pixel order,
Storage means for storing a predetermined number of decoding results and initializing with predetermined color data prior to starting decoding;
Predictive decoding means for obtaining prediction values with reference to color component data of decoded pixels in the vicinity of input pixel-of-interest data, and decoding predictive encoded data;
A color number counting means for counting the number of colors represented by a plurality of pixels in the vicinity of the pixel of interest and in the decoded pixel position, and generating color number information;
If the number of colors by the color count information obtained by the said color counting means indicates that more than a predetermined number, the coded data of the target pixel as a predictive coded data, decoding obtained by said prediction decoding unit the result, a first decoded data output means for outputting as the decoded data of the target pixel,
When the color count by the color count information obtained by the color number counting means is equal to or less than the predetermined number, the coded data of the target pixel, it is determined that the vector information, vector information decoding means for decoding the vector information When,
Vector information obtained by the vector information decoding means, has the same color as the target pixel, indicating decoded pixel position near, select the appropriate position of the pixel data in said storage means, said target pixel Second decoded data output means for outputting the decoded data as:
Vector information obtained by the vector information decoding means, indicating that the pixel position in the vicinity with the same color as the target pixel does not exist, the prediction decoding encoded data input subsequent to the vector information is decrypted by means, image decoding apparatus, characterized in that it comprises a third decoded data output means for outputting as the decoded data of the target pixel.   The said vector information decoding means decodes vector information using the code | symbol comprised according to the probability distribution assumed for every color number shown by color number information, The description of Claim 7 characterized by the above-mentioned. Image decoding apparatus. Run-length decoding means for decoding the number of consecutive pixels having the same color as a predetermined encoded pixel when a sequence of pixels having the same color is expected from the decoded pixels;
When the pixel data of the same color is generated by the run-length decoding unit, the decoded data from the run-length decoding unit is output, and when the continuation of pixels having the same color is not expected from the decoded pixels, The image decoding apparatus according to claim 7, further comprising: a switching unit that selects and outputs decoded data from the first to third decoded data generation units. A storage means for storing a predetermined number of lines of decoding results and initializing with data of a predetermined color prior to the start of decoding, and a code in which one pixel is composed of a plurality of color component data and is encoded in a predetermined pixel order A method of controlling an image decoding device for decoding digitized image data,
A predictive decoding step of obtaining predictive values with reference to color component data of decoded pixels in the vicinity of input pixel-of-interest data and decoding predictive encoded data;
A color number counting step of counting the number of colors represented by a plurality of pixels in the vicinity of the pixel of interest and in the decoded pixel position, and generating color number information;
If the number of colors by the color count information obtained by the said color number counting step indicates that more than a predetermined number, the coded data of the target pixel as a predictive coded data, decoding the obtained by predictive decoding process the result, a first decoded data output step of outputting as the decoded data of the target pixel,
When the color count by the color count information obtained by the color number counting step is equal to or less than a predetermined number, the coded data of the target pixel, it is determined that the vector information, and vector information decoding step of decoding the vector information ,
Vector information obtained by the vector information decoding step, with the same color as the target pixel, indicating decoded pixel position near, select the appropriate position of the pixel data in said storage means, said target pixel A second decoded data output step for outputting the decoded data as
Vector information obtained by the vector information decoding step, indicating that the pixel position in the vicinity with the same color as the target pixel does not exist, the prediction decoding encoded data input subsequent to the vector information is decoded in step, the control method of the image decoding device characterized by comprising a third decoded data output step of outputting as the decoded data of the target pixel. By causing read Komase executed by a computer, the computer, one pixel is composed of a plurality of color component data, function causes a computer program as the image decoding apparatus for decoding a coded image data at a predetermined pixel order Because
Means for storing a decoding result for a predetermined number of lines and securing a buffer memory that is initialized with data of a predetermined color prior to the start of decoding;
Predictive decoding means for obtaining prediction values with reference to color component data of decoded pixels in the vicinity of input pixel-of-interest data, and decoding predictive encoded data;
A color number counting means for counting the number of colors represented by a plurality of pixels in the vicinity of the pixel of interest and in the decoded pixel position, and generating color number information;
If the number of colors by the color count information obtained by the said color counting means indicates that more than a predetermined number, the coded data of the target pixel as a predictive coded data, decoding obtained by said prediction decoding unit the result, a first decoded data output means for outputting as the decoded data of the target pixel,
When the color count by the color count information obtained by the color number counting means is equal to or less than the predetermined number, the coded data of the target pixel, it is determined that the vector information, vector information decoding means for decoding the vector information When,
Vector information obtained by the vector information decoding means, has the same color as the target pixel, indicating decoded pixel position near, select the appropriate position of the pixel data in the buffer memory, the pixel of interest Second decoded data output means for outputting the decoded data as:
Vector information obtained by the vector information decoding means, indicating that the pixel position in the vicinity with the same color as the target pixel does not exist, the prediction decoding encoded data input subsequent to the vector information is decrypted by means, third computer program characterized by causing functions as decoded data output means for outputting as the decoded data of the target pixel.   A computer readable storage medium storing the computer program according to claim 11.

Images