JP5267252B2 - Image encoding device, image encoding method, image encoding recording medium, and image encoding program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image encoder capable of optimally encoding a foreground image in an MRC model at a high speed and improving image quality. <P>SOLUTION: In the image encoder, when original images recorded on an HDD 2 are read onto a RAM 3 by an instruction from a CPU 4, the CPU 4 reads the original images on the RAM 3, executes image encoding processing and quantization processing to them, and performs encoding to the MRC model. The CPU 4 is provided with an image generation means for generating a first image which is a pattern part image, a second image which is an image indicating the color of a non-pattern part and a third image which is an image indicating the shape of the non-pattern part from color input images, and an image encoding means for individually encoding the first image to the third image generated there. A frequency conversion encoding means for frequency-converting and encoding the first image and the second image provided in the image encoding means makes the degree of the quantization of high frequency components in the second image higher than the degree of the quantization of high frequency components in the first image. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to an image encoding device, an image encoding method, an image encoding recording medium, and an image encoding program for optimally encoding a foreground image in an MRC (Mixtorster Content) model at high speed.

  Conventionally, as a technique related to the MRC model, for example, in order to prevent deterioration of characters and line drawings at a high compression rate, the image data is first image data, second image data, first image data, and first image data. An image processing apparatus (see Patent Document 1) that separates the data into three pieces of selection data for selecting any one of the two pieces of image data, inputs the compressed data that has been encoded, and restores the image; Examples thereof include an encoding processing device, an encoding processing method, a program, and an information recording medium (see Patent Document 2) that generate JPM encoded data for dividing and improving accessibility to partial images.

  Patent Document 1 described above discloses the basic configuration of an MRC model, and Patent Document 2 discloses a technique for improving the accessibility to partial images by dividing a mask in MRC. In general, foreground and background images are compressed at a high compression ratio (for example, 1/40) in order to increase the compression ratio. Under such a high compression ratio, the skill of quantization greatly affects the image quality. Therefore, the quantization applied when compressing the foreground image and the background image needs to be optimized.

  Here, as a method for creating the MRC code, a method of reducing the foreground image and expressing the foreground image with several colors (one color in an extreme case) is typical. If such color reduction is used, the foreground image becomes an artificial image, and a high compression rate can be obtained by using an encoding method for the artificial image.

  However, the method of performing color reduction requires processing time for color reduction and causes image quality degradation due to color reduction. In order to prevent image quality deterioration due to this color reduction, it is necessary to narrow the area expressed by one color. For this purpose, it is necessary to divide the foreground image into many partial images, and in any case, processing time is required. (In the first place, since the MRC model requires a lot of processing time for generating the foreground image and the background image, there is a situation where it is desired to reduce the time required for encoding as much as possible). In addition, it is difficult to completely separate the foreground image and the background image. When a portion that should originally be a background image is separated as a foreground image, if it is reduced in color, it is recognized as an image defect. There is also a problem.

  Therefore, if an MRC code is to be created at a high speed, a technique capable of encoding at a high speed and with a high compression rate is required without excessively reducing and dividing the foreground image. When the foreground color reduction is omitted, it is necessary to compress the foreground image in which noise is superimposed by scanning, and thus a high compression rate cannot be obtained with the encoding method for artificial images. Therefore, in order to obtain a high compression ratio by omitting color reduction, a method using frequency conversion (encoding method for natural images) is applied to both the foreground image and the background image.

  However, as long as the color reduction is not performed, it is desired to optimize the quantization of the foreground image. However, since the reason why the color reduction is not performed is the high-speed processing, it is not possible to take much processing time for the optimization. For example, there is known a method of analyzing feature amounts and contents of an image and selecting an encoding method or a quantization method according to the analysis. However, such analysis generally has a large amount of processing and is difficult to employ.

  In short, the techniques disclosed in Patent Document 1 and Patent Document 2 related to the above-described MRC model have a problem that image quality cannot be improved by optimally encoding a foreground image at high speed.

  The present invention has been made to solve such problems, and its technical problem (objective) is to provide an image encoding apparatus and image that can improve the image quality by encoding the foreground image in the MRC model at high speed and optimally. To provide an encoding method, an image encoding recording medium, and an image encoding program.

  The present invention solves the above technical problem, and the invention according to claim 1 is a first image that is a picture portion image from a color input image, and a second image that is a color representing a color of a non-picture portion. And an image generation means for generating a third image that is an image representing the shape of the non-picture part, and the first image to the third image generated by the image generation means are individually encoded. An image encoding device comprising: an image encoding device comprising: a frequency conversion encoding unit configured to frequency-convert and encode the first image and the second image. And the frequency transform coding means makes the degree of quantization of the high frequency component in the second image stronger than the degree of quantization of the high frequency component in the first image.

The invention according to claim 2 is a first image that is a pattern portion image from a color input image, a second image that is an image that represents the color of the non-pattern portion, and an image that represents the shape of the non-pattern portion. An image encoding apparatus comprising: an image generation unit that generates a third image; and an image encoding unit that individually encodes the first image to the third image generated by the image generation unit. The image encoding means includes luminance / color difference conversion encoding means for encoding the first image and the second image by luminance / color difference conversion, and the luminance / color difference conversion encoding means. is that the degree of quantization of the color difference component on the degree of luminance quantization in the second image is smaller than the degree of quantization of the color difference component on the degree of luminance quantization in said first image Features.

  The invention according to claim 3 is a first image that is a pattern portion image from a color input image, a second image that is an image representing the color of the non-pattern portion, and an image that represents the shape of the non-pattern portion. An image encoding apparatus comprising: an image generation unit that generates a third image; and an image encoding unit that individually encodes the first image to the third image generated by the image generation unit. The image encoding means includes luminance / color difference conversion encoding means for encoding the first image and the second image by luminance / color difference conversion, and the luminance / color difference conversion encoding means. Is characterized in that the high-frequency component of luminance in the second image is most quantized and the high-frequency component of color difference in the first image is most quantized.

  According to a fourth aspect of the present invention, in the image encoding device according to any one of the first to third aspects, the image encoding unit performs the quantization process by performing frequency coefficient quantization and frequency coefficient This is performed by any one of a combination of discarding the lower bits, the quantization, and the discard. According to the fourth aspect of the present invention, specific quantization can be realized when the function of the image in MRC is reflected in both the frequency component and the luminance / color difference component.

  The invention according to claim 5 is a first image that is a picture portion image from a color input image, a second image that is an image that represents the color of the non-picture portion, and an image that represents the shape of the non-picture portion. An image encoding method comprising: an image generation step for generating a third image; and an image encoding step for individually encoding the first image to the third image generated in the image generation step. In the image encoding step, the first image and the second image are frequency-converted and encoded, and the degree of quantization of the high-frequency component in the second image is determined as the high-frequency in the first image. It is characterized by having a frequency transform coding step for making it stronger than the degree of quantization of the component.

The invention according to claim 6 is a first image that is a pattern portion image from a color input image, a second image that is an image that represents the color of the non-pattern portion, and an image that represents the shape of the non-pattern portion. An image encoding method comprising: an image generation step for generating a third image; and an image encoding step for individually encoding the first image to the third image generated in the image generation step. Te, wherein the image encoding step, the first image and the second image coded by converting luminance and color difference, the degree of quantization of color difference components for the degree of quantization of luminance in the second image Is provided with a luminance / color difference transform coding step for making the color difference component smaller than the degree of quantization of the color difference component with respect to the degree of luminance quantization in the first image.

  The invention according to claim 7 is a first image that is a pattern portion image from a color input image, a second image that is an image representing the color of the non-pattern portion, and an image that represents the shape of the non-pattern portion. An image encoding method comprising: an image generation step for generating a third image; and an image encoding step for individually encoding the first image to the third image generated in the image generation step. In the image encoding step, the first image and the second image are encoded by luminance / color difference conversion, and the high frequency component of the luminance in the second image is most quantized and the first image is encoded. A luminance-color-difference conversion coding step for quantizing the high-frequency component of the color difference in the image of the image most.

  The invention according to claim 8 is the image encoding method according to any one of claims 5 to 7, wherein the image encoding step includes quantization of frequency coefficients, and lower bits of frequency coefficients. This is performed by any one of a combination of discarding, quantization, and discarding.

  The invention according to claim 9 is a first image that is a picture portion image from a color input image, a second image that is an image that represents the color of the non-picture portion, and an image that represents the shape of the non-picture portion. Processing information of an image generation step for generating a third image and processing information of an image encoding step for individually encoding the first image to the third image generated in the image generation step are recorded. In the image encoding recording medium, the processing information of the image encoding step encodes the first image and the second image by frequency conversion, and the high-frequency component quantum in the second image is encoded. Processing information of a frequency conversion encoding step for making the degree of conversion stronger than the degree of quantization of the high frequency component in the first image, and luminance / color difference conversion of the first image and the second image Encode the second First luminance / color difference in which the degree of quantization of the color difference component relative to the degree of luminance quantization in the image is smaller than the degree of quantization of the color difference component relative to the degree of luminance quantization in the first image The processing information of the transform encoding step, the first image and the second image are encoded by luminance / color difference conversion, and the high frequency component of the luminance in the second image is most quantized, and the first At least one of the processing information of the second luminance / color difference transform coding step that most quantizes the high-frequency component of the color difference in the image.

  According to a tenth aspect of the present invention, in the image encoding recording medium according to the ninth aspect, the processing information of the image encoding step includes quantization of a frequency coefficient, discarding lower bits of the frequency coefficient, Instruction information to be performed by any one of the combination of the quantization and the discard is included.

The invention according to claim 11 is a first image that is a picture portion image from a color input image, a second image that is a color representing the color of the non-picture portion, and an image that represents the shape of the non-picture portion. Processing information of an image generation step for generating a third image, and processing information of an image encoding step for individually encoding the first image to the third image generated in the image generation step. And an image encoding program to be executed by the information processing apparatus, wherein the processing information of the image encoding step is obtained by performing frequency conversion on the first image and the second image and encoding the first image and the second image. Processing information of a frequency transform encoding step for making the degree of quantization of the high-frequency component in the second image stronger than the degree of quantization of the high-frequency component in the first image, the first image, and the second image Picture The encoded by converting luminance and color difference, the degree of quantization of the color difference component on the degree of luminance quantization in the second image, the quantization of the color difference component on the degree of quantization of luminance in the first image and processing information of the first luminance and color difference conversion encoding step to be smaller than the extent of, the first image and the second image coded by converting luminance and color difference, the luminance in the second image It is characterized in that it has at least one of the processing information of the second luminance / color difference conversion encoding step that most quantizes the high frequency component and most quantizes the high frequency component of the color difference in the first image.

  According to a twelfth aspect of the present invention, in the image coding program according to the eleventh aspect, the processing information of the image coding step includes quantization of a frequency coefficient, discarding lower bits of the frequency coefficient, It is characterized by including instruction information performed by any one of the combination of quantization and the discarding.

  According to the technical gist of the invention described in claim 1 or 5, it is possible to perform encoding by reflecting the function of the image in MRC in the quantization of frequency components. According to the technical gist of the invention described in claim 2 or 6, it is possible to perform encoding by reflecting the function of the image in MRC in the quantization of luminance / chrominance components. According to the technical gist of the invention described in claim 3 or 7, it is possible to perform encoding by reflecting the function of the image in MRC in both the frequency component and the luminance / color difference component. According to the technical gist of the invention described in claim 4 or 8, specific quantization is realized in any case where the function of the image in MRC is reflected in the frequency component, the luminance / chrominance component, or both. can do. According to the technical gist of the invention described in claim 9 or 10, a recording medium that covers the above effects can be provided. According to the technical summary of the invention described in claim 11 or 12, it is possible to provide a program that covers the above effects.

It is the block diagram which showed the basic composition of the image coding apparatus which concerns on Example 1 of this invention. It is a flowchart for demonstrating the encoding process of JPEG2000. An example of an original image, a coordinate system, and mirroring after a DC level shift in a process of performing wavelet transform in two dimensions on a luminance component of a 16 × 16 image is illustrated. FIG. 4 illustrates an example of the arrangement of coefficients after filtering in the vertical direction in the process of FIG. 3. FIG. 4 shows an example of an arrangement of coefficients after filtering in the horizontal direction in the process of FIG. 3. FIG. 4 shows an example of the rearranged coefficient arrangement in the process of FIG. 3. FIG. 4 illustrates an example of an array of coefficients that has been subjected to the wavelet transform described in the process of FIG. 3 and rearranged twice. It illustrates the relationship between the decomposition level and the resolution level. The relationship between an image (tile), a subband, a precinct, and a code block is illustrated. 1 shows a schematic configuration of an MRC model by an image coding apparatus according to Embodiment 1 of the present invention. It is the flowchart which showed the operation | movement process of the image coding of the MRC model by the image coding apparatus of Example 1 of this invention. 12 illustrates a first weight matrix as a Sobel operator of a Sobel filter used in the image encoding operation process shown in FIG. 11. 12 is a diagram illustrating a second weight matrix as a Sobel operator of a Sobel filter used in the image encoding operation process shown in FIG. 11. 11 illustrates an example of the number of lower bit planes that are not encoded according to the signal processing for the YCbCr signal when encoding the foreground 1 of the picture portion image in the image encoding operation process shown in FIG. (B) relates to the Cb signal, and (c) relates to the Cr signal. FIG. 11 illustrates an example of the number of lower bit planes that are not encoded according to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the first method in the image encoding operation processing shown in FIG. (A) relates to the Y signal, (b) relates to the Cb signal, and (c) relates to the Cr signal. FIG. 11 illustrates an example of the number of lower bit planes that are not encoded according to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the second method in the image encoding operation processing shown in FIG. (A) relates to the Y signal, (b) relates to the Cb signal, and (c) relates to the Cr signal. FIG. 11 illustrates the number of lower bit planes that are not encoded according to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the third technique in the image encoding operation process shown in FIG. (A) relates to the Y signal, (b) relates to the Cb signal, and (c) relates to the Cr signal. It is the flowchart which showed the operation | movement process of the image encoding of the MRC model by the image coding apparatus which concerns on Example 2 of this invention. This is an example of the number of quantization steps that are not considered for visual characteristics related to signal processing for YCbCr signals when encoding color images that are generally applied (typical used with 9 × 7 filter of JPEG2000). , (A) relates to the Y signal, (b) relates to the Cb signal, and (c) relates to the Cr signal. The visual sensitivity (typical thing used with a 9 * 7 filter of JPEG2000) concerning signal processing to a YCbCr signal when coding a color image that is generally applied is illustrated, and (a) shows a Y signal. (B) relates to the Cb signal, and (c) relates to the Cr signal. Quantization step number considering visual characteristics related to signal processing for YCbCr signal when coding foreground 1 of picture portion image in operation processing of image coding shown in FIG. 18 (typical used with 9 × 7 filter of JPEG2000 (A) relates to the Y signal, (b) relates to the Cb signal, and (c) relates to the Cr signal. 18 illustrates the number of quantization steps in consideration of visual characteristics related to signal processing for a YCbCr signal when encoding a foreground 2 of a color image of a character (line drawing) in the image encoding operation processing shown in FIG. a) relates to the Y signal, (b) relates to the Cb signal, and (c) relates to the Cr signal. It is the flowchart which showed the operation | movement process of the image encoding of the MRC model by the image coding apparatus which concerns on Example 3 of this invention.

  Hereinafter, an image encoding device, an image encoding method, an image encoding recording medium, and an image encoding program of the present invention will be described in detail with reference to the drawings.

  First, the technical focus of the present invention will be described. The present applicant pays attention to the structure and function of the MRC model when encoding a foreground image without color reduction at high speed, and performs encoding using frequency conversion and quantization on the foreground image. In the case of adopting it, when considering the function of the layer itself “color of foreground image”, it was found that the difference in function without analyzing the contents of the image can be reflected in the quantization. Until now, the quantization which reflects the function concerned with a foreground image has not been performed.

  As explained in the technical problem, the information that the foreground image has is the color and not the shape, but the color component is mainly the low frequency component of the image, whereas the shape component is the high frequency component (edge component) of the image. ). That is, in the foreground image, it is considered that high frequency, more specifically, high frequency of luminance should be quantized. On the other hand, since the background image is a pattern portion excluding characters and line drawings, almost normal quantization for natural images can be applied.

  Therefore, in the image encoding method of the present invention, from the color input image, the first image that is the image portion image, the second image that is the image that represents the color of the non-pattern portion, and the image that represents the shape of the non-pattern portion An image generation step for generating a third image, and an image encoding step for individually encoding the first image to the third image generated in the image generation step. In the encoding step, the first image and the second image are frequency-converted and encoded, and the degree of quantization of the high-frequency component in the second image is stronger than the degree of quantization of the high-frequency component in the first image. It is assumed that a frequency transform encoding step is performed. According to the first image encoding method, encoding can be performed by reflecting the function of the image in MRC in the quantization of frequency components.

  By the way, in normal encoding, luminance / color difference conversion is performed, and quantization is performed for each luminance and each color difference. In general natural images, the degree of luminance quantization << the degree of color difference quantization is generally reflected, reflecting human visual characteristics that are more sensitive to luminance than color.

  However, in the foreground image of MRC, even if it is a natural image, the degree of luminance quantization <the degree of color difference quantization or the degree of luminance quantization≈the degree of color difference quantization is a requirement. It is desirable to change the quantization ratio of luminance and color difference from that of the background image.

The second image encoding method in consideration of such points includes an image generation step and an image encoding step similar to those in the case of the first image encoding method, and the image encoding step includes: the image and the second image coded by converting luminance and color difference, the degree of quantization of color difference components for the degree of quantization of luminance in the second image (the degree of quantization of the degree / intensity of the quantized color difference) Is provided with a luminance / color difference transform coding step for making the smaller than the degree of quantization of the color difference component with respect to the degree of luminance quantization in the first image. According to the second image encoding method, it is possible to perform encoding by reflecting the function of the image in MRC in the quantization of luminance / chrominance components.

  Further, according to the above-described points, the highest frequency component to be quantized in the foreground image is the high frequency component of color difference and the highest frequency component in the background image is to be quantized.

  The third image encoding method considering such points includes an image generation step and an image encoding step similar to those in the case of the first image encoding method, and the image encoding step includes: Luminance / color difference conversion that encodes the image and the second image by performing luminance / color difference conversion, and most quantizes the high frequency component of luminance in the second image, and most quantizes the high frequency component of color difference in the first image. It is assumed to have an encoding step. According to the third image encoding method, encoding can be performed by reflecting the function of the image in MRC in both the frequency component and the luminance / color difference component.

  By the way, the quantization in each of the image coding methods described above differs in implementation form depending on the coding method and the selection of the coding parameter. First, when performing linear quantization or nonlinear quantization of frequency coefficients (for example, coefficient When dividing by the value S, and second, a method in which the low-order bits of the frequency coefficient are not encoded, and the low-order bits of the frequency coefficient are encoded but eventually discarded, and the third As described above, there are three cases where both the first and second can be applied simultaneously.

  Omitting or discarding encoding for n bit planes in the second quantization is equivalent to dividing the coefficient by 2 to the nth power in the first quantization, and the third quantization is: This is equivalent to dividing the coefficient by S × (2 to the power of n).

  The image encoding method considering such points is the image encoding method in each of the image encoding methods described above, and the image encoding step includes any one of the quantization, the quantization of the frequency coefficient, the discard of the lower bits of the frequency coefficient, and the combination of the quantization and the discard. This is what you do. By applying this method, specific quantization can be realized when the function of the image in MRC is reflected in both the frequency component and the luminance / color difference component.

  FIG. 1 is a block diagram showing a basic configuration of an image coding apparatus according to Embodiment 1 of the present invention. This image encoding apparatus can apply each of the above-described image encoding methods and the quantization method thereto, and an HDD 2 is connected to a RAM 3 and a CPU 4 in a personal computer (PC) via a data bus 1. The information processing system consists of standard specifications.

  In this image encoding device, an original image (original color image) is encoded in the following flow. First, an original image recorded on the HDD 2 is read onto the RAM 3 by a command from the CPU 4. Secondly, the CPU 4 reads the original image on the RAM 3, performs the image encoding process of the present invention and the quantization process thereof, and encodes the MRC model. Third, the CPU 4 writes the encoded data in another area on the RAM 3. Fourth, the encoded data (image) is read from the RAM 3 according to an instruction from the CPU 4 and recorded on the HDD 2.

  The CPU 4 in this case is a first image that is a picture portion image, a second image that is a color representing the color of the non-pattern portion, and a third image that is a shape representing the shape of the non-picture portion from the color input image. It functions as an image generating means for generating an image and an image encoding means for individually encoding the first image to the third image generated by the image generating means. The image encoding means has frequency conversion encoding means for frequency-converting and encoding the first image and the second image, and the frequency conversion encoding means is for quantizing high-frequency components in the second image. It has a functional configuration that makes the degree stronger than the degree of quantization of the high-frequency component in the first image.

The image encoding means includes luminance / color difference conversion encoding means for encoding the first image and the second image by luminance / color difference conversion, and the luminance / color difference conversion encoding means includes the second the degree of quantization of the color difference component on the degree of quantization of luminance in the image (the degree of quantization of the degree / intensity of the quantization of the color difference), the quantization of the color difference component on the degree of quantization of luminance in the first image It may have a functional configuration that is smaller than the degree of.

  Further, the image encoding means includes luminance / color difference conversion encoding means for encoding the first image and the second image by luminance / color difference conversion, and the luminance / color difference conversion encoding means includes the second It may have a functional configuration that most quantizes the high-frequency component of luminance in the image and most quantizes the high-frequency component of color difference in the first image.

  In any functional configuration, the image encoding means preferably performs the quantization process by any one of frequency coefficient quantization, discard of lower bits of the frequency coefficient, and a combination of quantization and discard.

  By the way, it is not the structure which performs image coding processing by the function structure of CPU4, but the image coding which recorded the processing information of the image generation step which applied each image coding method mentioned above, and the processing information of the image coding step It is also possible to read a recording medium (such as a hard disk) by the HDD 2 and perform an equivalent processing operation.

In this case, the processing information of the image encoding step in the image encoding recording medium is encoded by frequency-converting the first image and the second image, and the degree of quantization of the high-frequency component in the second image is The processing information of the frequency conversion encoding step for making the frequency component higher than the degree of quantization of the high frequency component in the first image, the first image and the second image are encoded by luminance / color difference conversion, and the second image is encoded. the degree of quantization of the color difference component on the degree of quantization of luminance (the degree of quantization of the degree / intensity of the quantization of the color difference), the degree of quantization of color difference components for the degree of quantization of luminance in the first image Information of the first luminance / color difference conversion coding step to be smaller than the first and second images and luminance / chrominance-converted and encoded, and the high-frequency component of luminance in the second image is most quantized. As As long as it has at least one of the second luminance and color difference conversion encoding step of processing information that best quantized high-frequency component of the color difference in the first image. However, for this image encoding recording medium, the processing information of the image encoding step is information indicating that quantization is performed by any one of quantization of frequency coefficient, discard of lower bits of frequency coefficient, and combination of quantization and discard As long as it contains.

  Similarly, the CPU 4 is caused to execute an image encoding program having the processing information of the image generation step to which each of the image encoding methods described above is applied and the processing information of the image encoding step to perform an equivalent processing operation. It is also possible.

  In such a case, the processing information of the image encoding step in the image encoding program includes processing information of the frequency conversion encoding step similar to that described above, processing information of the first luminance / color difference conversion encoding step, It is sufficient if it has at least one of the processing information of the luminance / color difference conversion encoding step. However, also for this image encoding program, the processing information of the image encoding step includes instruction information for performing quantization by any one of quantization of frequency coefficient, discard of lower bits of frequency coefficient, and combination of quantization and discard. It may be included.

  Prior to specific description of the first embodiment, an outline of the JPEG2000 encoding method used in the first embodiment will be described below.

  JPEG2000 is a JPEG successor image encoding method that became an international standard in 2001, and the encoding process is generally performed as shown in FIG.

  First, the image is divided into rectangular tiles (the number of divisions ≧ 1), and each tile is subjected to a DC level shift when compressing a color image made up of, for example, three RGB components, and then the luminance / color difference components. Perform component conversion (color conversion).

  After DC level shift and color conversion, wavelet conversion is performed, quantization is performed for each subband as necessary, and bitplane coding is performed from the upper bitplane to the (necessary) lower bitplane for each subband. (Entropy coding in bit plane units). Furthermore, after unnecessary codes are discarded and the necessary codes are collected to generate a packet, the packets are arranged to form a code.

  When decompressing, inverse wavelet transform is applied to the wavelet coefficients of each component obtained through inverse quantization after bit plane decoding (entropy decoding in bit plane units), and then inverse color transform is performed. It is shown that the pixel values are returned to RGB.

  Here, JPEG2000 DC level shift conversion formula and inverse conversion formula are obtained by subtracting 1 from the bit depth of each component i (i = 0, 1, 2 for RGB image) of Ssiz (i). 2 ^ Ssiz (i) means 2 to the power of Ssiz (i), and I (x, y) is the original signal value (pixel value) at the coordinates (x, y). I (x, y) ← I (x, y) −2 ^ Ssiz (i). For the inverse transformation, I (x, y) ← I (x, y) + 2 ^ Ssiz (i ).

  When this DC level shift is a positive number such as an RGB signal value, the forward conversion subtracts half of the dynamic range of the signal from each signal value, and the inverse conversion uses the signal dynamics for each signal value. A level shift that adds half the range is performed. However, this level shift is not applied to signed integers such as Cb and Cr signals of YCbCr signals.

  In JPEG2000, reversible conversion (RCT) and irreversible conversion (ICT) are defined as component conversion (color conversion).

  The forward transformation for reversible transformation (RCT) is Y0 (x, y) = floor ((I0 (x, y) + 2 * (I1 (x, y) + I2 (x, y)) / 4), Y1 (x Y) = I2 (x, y) -I1 (x, y), Y2 (x, y) = I0 (x, y) -I1 (x, y), and the inverse transformation is represented by I1 (x Y) = Y0 (x, y) -floor ((Y2 (x, y) + Y1 (x, y)) / 4), I0 (x, y) = Y2 (x, y) + I1 (x, y) , I2 (x, y) = Y1 (x, y) + I1 (x, y).

  In each equation, I represents an original signal, Y represents a signal after conversion, and 0 to 2 following I and Y are suffixes. In the case of an RGB signal, I0 = R, I1 = G, I2 = B in the I signal, and Y0 = Y, Y1 = Cb, and Y2 = Cr in the Y signal. Floor (X) is a function that replaces the real number X with an integer that does not exceed X and is closest to X.

  The forward transformation for the irreversible transformation (ICT) is Y0 (x, y) = 0.299 * I0 (x, y) + 0.587 * I1 (x, y) + 0.144 * I2 (x, y), Y1 (x, y) = − 0.1675 * I0 (x, y) −0.33126 * I1 (x, y) + 0.5 * I2 (x, y), Y2 (x, y) = 0.5 * I0 (x, y) −0.41869 * I1 (x, y) −0.0811 * I2 (x, y), and the inverse transformation is I0 (x, y) = Y0 (x, y ) + 1.402 * Y2 (x, y), I1 (x, y) = Y0 (x, y) −0.34413 * Y1 (x, y) −0.71414 * Y2 (x, y), I2 ( x, y) = Y0 (x, y) + 1.772 * Y1 (x, y).

  In each equation, I represents the original signal, and Y represents the signal after conversion. In the case of RGB signals, I0 = R, I1 = G, I2 = B in the I signal, and I0 = Y, I1 = Cb, and I2 = Cr in the Y signal.

  The component after conversion (referred to as a tile component) is divided into four subbands abbreviated as LL, HL, LH, and HH by wavelet conversion. Incidentally, when wavelet transform (decomposition) is recursively repeated for the LL subband, one LL subband and a plurality of HL, LH, and HH subbands are finally generated.

  Here, the 5 × 3 wavelet transform and 9 × 7 wavelet transform (either one is selected and used) used in JPEG2000 will be described.

  The 5 × 3 wavelet transform is a transformation in which an output (low pass coefficient) of one low pass filter is obtained using 5 pixels and an output (high pass coefficient) of one high pass filter is obtained using 3 pixels. Similarly, the 9 × 7 wavelet transform is a conversion in which an output (low-pass coefficient) of one low-pass filter is obtained using 9 pixels and an output (high-pass coefficient) of one high-pass filter is obtained using 7 pixels. . The main difference is the difference in the filter range, and the same is true in that a low-pass filter is applied to the center of the even position and a high-pass filter is applied to the center of the odd position.

  The forward transformation in the 5 × 3 wavelet transformation formula is performed by C (2i + 1) = P (2i + 1) −floor ((P (2i) + P (2i + 2)) / 2) in the first step and C (2i in the second step. ) = P (2i) + floor ((C (2i-1) + C (2i + 1) +2) / 4), and the inverse transformation is P (2i) = C (2i) -floor (( C (2i-1) + C (2i + 1) +2) / 4), P (2i + 1) = C (2i + 1) + floor ((P (2i) + P (2i + 2)) / 2) in the second step.

  The forward transformation in the 9 × 7 wavelet transformation formula is performed by C (2n + 1) = P (2n + 1) + α * (P (2n) + P (2n + 2)) in the first step and C (2n) = P (in the second step. 2n) + β * (C (2n-1) + C (2n + 1)), third step C (2n + 1) = P (2n + 1) + γ * (C (2n) + C (2n + 2)), fourth step C (2n ) = C (2n) + δ * (C (2n−1) + C (2n + 1)), C (2n + 1) = K * C (2n + 1) in the fifth step, C (2n) = (1 / K in the sixth step ) * C (2n), and the inverse transformation is performed by P (2n) = K * C (2n) in the first step, P (2n + 1) = (1 / K) * C (2n + 1) in the second step , P (2n) = X (2n) −δ * (P (2n−1) + P (2n + 1)) in the third step, Step P (2n + 1) = P (2n + 1) −γ * (P (2n) + P (2n + 2)), fifth step P (2n) = P (2n) −β * (P (2n−1) + P ( 2n + 2)), P (2n) = P (2n + 1) −α * (P (2n) + P (2n + 2)) in the sixth step.

  However, in this case, α = −1.586134342059924, β = −0.0529801188572961, γ = 0.8291910755530934, δ = 0.443456852043971, and K = 1.230174141491001.

  Next, the procedure of wavelet transform and the definition of the decomposition level, resolution level, and subband will be described.

  FIG. 3 shows an original image and a coordinate system after DC level shift in the process of performing wavelet transform (5 × 3 transform) in two dimensions (vertical direction and horizontal direction) on a luminance component of a 16 × 16 image. This is an example of mirroring. FIG. 4 shows an example of the arrangement of coefficients after filtering in the vertical direction in the same process. FIG. 5 exemplifies the arrangement of the coefficients after filtering in the horizontal direction in the same process. FIG. 6 shows an example of the rearranged coefficient array in the same process.

  Referring to FIG. 3, xy coordinates are taken, and for a certain x, the pixel value of a pixel whose y coordinate is y is represented as P (y) (0 ≦ y ≦ 15). In JPEG2000, first, a high-pass filter is applied to a pixel whose y coordinate is odd (y = 2i + 1) in the vertical direction (Y coordinate direction) to obtain a coefficient C (2i + 1), and then the y coordinate is even (y = 2i). ) To obtain a coefficient C (2i) by applying a low-pass filter centering on the pixel of (), and such processing is performed for all x. Here, the high-pass filter and the low-pass filter are expressed by the relational expressions of the first step and the second step described above in order. When a filter is applied to the edge of an image, there are cases where there are not enough adjacent pixels with respect to the central pixel. In such a case, the pixel value is appropriately compensated by a technique called mirroring. Mirroring assumes a virtual pixel outside the edge of the image, and the pixel value inside the edge is copied symmetrically with respect to the pixel outside the edge, with the pixel at the edge of the image as the axis of symmetry. This is a well-known technique. Mirroring is performed at all four edges of the image (tile).

  For simplicity of explanation, if the coefficient obtained by the high-pass filter is denoted by H and the coefficient obtained by the low-pass filter is denoted by L, the image shown in FIG. 3 is converted to the L coefficient as shown in FIG. , Converted to an array of H coefficients.

  Subsequently, a high-pass filter is first applied to the coefficient array as shown in FIG. 4 centering on the coefficient whose x coordinate is odd (y = 2i + 1) in the horizontal direction, and then the x coordinate is even (x = 2i). ), The low pass filter is applied to the center, and these processes are performed for all y. In this case, P (2i) and the like in the first step and the second step are read as representing coefficient values.

  Furthermore, this time, a high-pass filter is first applied to the coefficient array shown in FIG. 4 centering on a coefficient whose x coordinate is an odd number (y = 2i + 1) in the horizontal direction, and then an x coordinate is an even number (x = 2i). A low pass filter is applied centering on the coefficients, and processing is performed for all y. In this case, P (2i) and the like in the first step and the second step are read as representing coefficient values.

  Similarly, in order to simplify the description, the coefficient obtained by applying the low-pass filter centered on the L coefficient described above is LL, the coefficient obtained by applying the high-pass filter centered on the L coefficient is HL, and the coefficient obtained by applying the high-pass filter is centered on the low-pass filter. If the coefficient obtained by applying the filter is expressed as LH, and the coefficient obtained by applying the high-pass filter centered on the H coefficient is expressed as HH, the coefficient array shown in FIG. 4 becomes a coefficient array as shown in FIG. Converted. Here, the coefficient group to which the same symbol is attached is called a subband, and is composed of four subbands in the form shown in FIG.

  As described above, each wavelet transform (one decomposition (decomposition)) is completed for each of the vertical and horizontal directions, and only the above-mentioned LL coefficients are collected and each subband as shown in FIG. If only the LL subband is taken out, an “image” having a resolution that is half the original image is obtained. Incidentally, such classification for each subband is called deinterleaving, and arranging in the processing state as described with reference to FIG. 1 is called interleaving.

  For the second wavelet transform, the LL subband is regarded as the original image, and the same transformation as described above may be performed. In this case, when rearrangement is performed, the form shown in FIG. 7 is obtained. The coefficient prefixes 1 and 2 here indicate how many times the wavelet transform has been obtained with respect to horizontal and vertical, and are called decomposition levels.

  FIG. 8 illustrates the relationship between the decomposition level and the resolution level. Here, it is shown that the resolution level is almost opposite to the composition level.

  In the above description, when only one-dimensional wavelet transformation is performed, processing in only one direction may be performed, and the number of times wavelet transformation has been performed in any one direction becomes the decomposition level.

  When a 9 × 7 filter is selected as the wavelet filter, linear quantization of the coefficients is performed for each subband after the above-described division. When a 5 × 3 filter is selected, for each subband, The specification does not perform linear quantization.

  As the next processing, each subband is divided into rectangles called precincts. The precinct is a collection of three subbands HL, LH, and HH obtained by dividing a subband into rectangles. In a broad sense, the precinct represents a position in an image. Three precincts are one group, but when the LL subband is divided, one group is one. The precinct can be the same size as the subband, and the code block is obtained by further dividing the precinct into rectangles.

  FIG. 9 illustrates the relationship between an image (tile), subband, precinct, and code block. Here, as an order of physical size, it is shown that image ≧ tile> subband ≧ precinct ≧ code block.

  After the above-described division, entropy coding of coefficients (bit plane coding in bit plane units) is performed for each code block in precinct units and in bit plane order. In the first embodiment, as described with reference to FIG. 2, the bit plane encoding is performed from the most significant bit plane to the lower bit plane defined in advance in the table. When considered as a necessary code under such conditions, there is no unnecessary code. For these codes “from the most significant bit plane to the lower bit plane preliminarily defined in the table”, a code with a header is called a packet. The packet header includes information about codes included in the packet, and each packet can be handled independently. In other words, a packet is a unit of code.

  For all precincts (= all code blocks = all subbands), packets are arranged to form a code.

  In the following, a technical overview of the image coding apparatus according to the first embodiment will be described based on the processing contents described above.

  FIG. 10 shows a schematic configuration of an MRC model by the image coding apparatus according to the first embodiment. FIG. 11 is a flowchart showing an operation process of image encoding of the MRC model by the image encoding apparatus.

  Referring to FIG. 10, the MRC model here has a white background as a base page, a black first mask 1 (MMR specification), and a pattern portion of the printer at a predetermined location based on a light moss green color. A first foreground 1 (JPEG2000 specification) having a colorless transparent base tone that preserves the printed portion (MMR specification), and a second foreground 2 (JPEG2000 specification) in which magenta and yellow-green are separated into two colors. It is an image configuration in which images are stacked one after another. On the composite image (assumed after the image encoding process), the shape outline and color of the pattern of the printer are clear, and the print portion is divided into a red-purple portion and a yellow-green portion.

  In the image encoding operation process shown in FIG. 11, when an original image is first input, it is determined whether or not it is a pixel (hereinafter referred to as a character pixel) constituting a character (line drawing) in units of pixels with respect to the original image. Then, the second mask 2 is created (step S1).

  This is performed by a known image area discrimination technique. In this example, a Sobel filter known as an edge detection operator is applied to each pixel of the original image. The Sobel filter multiplies the first weight matrix shown in FIG. 12 by multiplying the top, bottom, left, and right nine pixels centered on the target pixel to calculate the sum HS, and similarly, the second HS shown in FIG. To calculate the sum VS. Then, the square root (HS ^ 2 + VS ^ 2) is set as the output value of the filter. When the output value of this filter is 30 or more, for example, it is determined that the target pixel is a character pixel, the value of the target pixel position is set to 1, and the values of the other pixels are set to 0, whereby the second mask 2 is generated. In the case of a binary image, black = 1 and white = 0 are common.

  Next, the color of a pixel that does not belong to the character (line drawing) region (hereinafter referred to as a non-character pixel) is replaced with the color of the character pixel that is closest to the pixel to create the second foreground 2 (step S2). ) Since this process is performed for all non-character pixels in the image, as described with reference to FIG. 11, the image is replaced with the character color. When the character color of the original image is one color, the image itself is one color.

  Next, the color of the pixel in the character (line drawing) area in the original image is replaced with the color of the non-character pixel located closest to the pixel to create the first foreground 1 (step S3). The foreground 1 here may be called a background.

  Subsequently, a first mask 1 is created as an image having a pixel value 1 having the same size as the original image (step S4), and a background image is created as a white image having the same size as the original image (step S5).

  Then, the first foreground 1 is decomposed twice using a 5 × 3 filter, encoded according to the JPEG2000 specification under the conditions shown in FIG. 14 (step S6), and the second foreground 2 is similarly converted to 5 × 3. After decomposing twice using the filter and encoding according to the JPEG2000 specification under the conditions shown in FIG. 15, FIG. 16 or FIG. 17 (step S7), the first mask 1 and the second mask 2 (both 2 The value image) is encoded by a well-known MMR specification (step S8).

  Finally, the background is encoded according to the JPM specification (entropy encoding is not performed, the background color is specified as the code: step S9), all the foreground, mask, and background codes are combined, and a desired header is attached. JPM code is acquired (step S10).

  FIG. 14 described above exemplifies the number of lower bit planes that are not encoded according to the signal processing for the YCbCr signal when encoding the foreground 1 that is a picture portion image. FIG. 14A relates to the Y signal. FIG. 4B relates to the Cb signal, and FIG. 4C relates to the Cr signal.

  Here, as shown in FIG. 14, in the foreground 1 that is a picture portion image, the number of lower bitplanes that are not encoded is increased with respect to the color difference rather than the luminance, and the degree of luminance quantization is determined by the color difference quantization. It is stronger than the degree. In RCT, the number of bits of color difference is one bit greater than the number of bits of luminance, but as is clear from FIG. 14, the difference in the number of lower bit planes that are not encoded is more than that.

  FIG. 15 illustrates the number of low-order bit planes that are not encoded according to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the first method in the image encoding operation process described above. FIG. 9A relates to the Y signal, FIG. 10B relates to the Cb signal, and FIG. 10C relates to the Cr signal.

The first method here corresponds to the first image encoding method described above, and in the foreground 2 which is an image representing a color, as shown in the thick frame region in FIG. In the decomposition level 1, the number of low-order bit planes that are not encoded is larger than that of the foreground 1 with respect to luminance and color difference, and the degree of quantization of high-frequency components is stronger than that of the foreground 1 .

  FIG. 16 exemplifies the number of lower bit planes that are not encoded according to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the second method in the image encoding operation process described above. FIG. 9A relates to the Y signal, FIG. 10B relates to the Cb signal, and FIG. 10C relates to the Cr signal.

The second method here corresponds to the second image encoding method described above. As is clear from the comparison with FIG. 14 in FIG. 16, in the foreground 1, high-frequency quantization of luminance is performed. In the foreground 2, although the degree of high-frequency quantization of the color difference is high, the degree of high-frequency quantization of the luminance is equal to the degree of high-frequency quantization of the color difference. The ratio is changed between foregrounds 1 and 2. That is, the degree of quantization of the color difference component on the degree of quantization of luminance in the second image (the degree of quantization of the degree / intensity of the quantization of the color difference), is smaller than that of the first image . Note that the same number of lower bit planes that are not encoded with respect to the foreground 1 are the same as those in FIG.

  FIG. 17 illustrates the number of low-order bit planes that are not encoded according to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the third method in the image encoding operation process described above. FIG. 9A relates to the Y signal, FIG. 10B relates to the Cb signal, and FIG. 10C relates to the Cr signal.

  The third method here corresponds to the above-described third image encoding method, and as is clear from the comparison with FIG. 14 in FIG. 17, the high frequency of the color difference is quantized most in the foreground 1. However, in FIG. 17, the high frequency of luminance is most quantized. In this case as well, the same number of lower bit planes that are not encoded with respect to the foreground 1 are the same as those in FIG.

  FIG. 18 is a flowchart showing an operation process of image coding of the MRC model by the image coding apparatus according to Embodiment 2 of the present invention. Compared to the processing described in FIG. 11, the image encoding operation processing here is performed by decomposing twice foreground 1 and foreground 2 using a 9 × 7 filter, linearly quantizing the coefficients, The difference is that the bit plane is encoded.

  More specifically, in the image encoding operation process shown in FIG. 18, after the input of the original image, the character pixels constituting the character (line drawing) in units of pixels, as in the case of the operation process of FIG. The process of creating the second mask 2 (step S11), the color of the non-character pixel that does not belong to the character (line drawing) region is the color of the character pixel that is closest to the pixel. The process of replacing and creating the second foreground 2 (step S12), replacing the color of the pixel in the character (line drawing) area in the original image with the color of the non-character pixel located closest to the pixel, Process for creating foreground 1 (step S13), process for creating first mask 1 as an image with the same pixel size 1 as the original image (step S14), and creating a background image as a white image of the same size as the original image The process of (Step S15) is performed.

  Then, the first foreground 1 is decomposed twice using a 9 × 7 filter, quantized under the conditions shown in FIG. 21, and then all bit planes are encoded (step S16). Foreground 2 is decomposed twice using a 9 × 7 filter, quantized under the conditions shown in FIG. 22, and then all bit planes are encoded (step S17).

  Thereafter, as in the case of the operation process of FIG. 11, the process of encoding the first mask 1 and the second mask 2 (both binary images) with the well-known MMR specification (step S18), and the background. JPM coding (entropy coding is not performed, background color is specified as a code: step S19), all foreground, mask, and background codes are combined, and a desired header is added to the JPM code Is acquired (step S20).

  Hereinafter, the linear quantization in the second embodiment will be described. FIG. 19 exemplifies the number of quantization steps (a typical one used with a 9 × 7 filter of JPEG2000) that does not consider visual characteristics related to signal processing for a YCbCr signal when encoding a general color image. FIG. 6A relates to the Y signal, FIG. 5B relates to the Cb signal, and FIG. 5C relates to the Cr signal. The number of quantization steps is obtained as a value obtained by multiplying the inverse of the square root of the subband gain of the 9 × 7 filter by a constant.

  FIG. 20 exemplifies the visual sensitivity (typical used with the 9 × 7 filter of JPEG2000) related to the signal processing for the YCbCr signal when encoding a general color image. Is related to the Y signal, (b) is related to the Cb signal, and (c) is related to the Cr signal. The visual sensitivity indicates that the larger the value, the larger the visual influence of the error generated in the subband.

  FIG. 21 shows the number of quantization steps in consideration of the visual characteristics related to the signal processing for the YCbCr signal when encoding the foreground 1 of the picture part image in the image encoding operation processing (typical used together with the JPEG2000 9 × 7 filter). (A) is related to the Y signal, (b) is related to the Cb signal, and (c) is related to the Cr signal. Since the effect of visual sensitivity cannot be taken with the number of quantization steps not considering the visual characteristics described above, in JPEG 2000, the number of quantization steps obtained from the subband gain is divided by the visual sensitivity, as shown in FIG. It is desirable to calculate the number of quantization steps in consideration of visual characteristics.

  FIG. 22 exemplifies the number of quantization steps in consideration of the visual characteristics related to the signal processing for the YCbCr signal when the foreground 2 of the color image of the character (line drawing) is encoded by the image encoding operation processing. FIG. 4A relates to the Y signal, FIG. 5B relates to the Cb signal, and FIG. 4C relates to the Cr signal.

  That is, in the second embodiment, for the foreground 2, the number of quantization steps in FIG. 22 is applied. For the foreground 1, the number of quantization steps in FIG. 21 is applied.

  As is apparent from FIG. 22, in the foreground 2, the number of quantization steps in the foreground 2 is larger than that in the foreground 1 with respect to luminance and color difference at the decomposition level 1 that is a high-frequency component. The quantization step value is smaller than that in the foreground 1. In FIG. 21, the color difference high frequency is most quantized, but in FIG. 22, the luminance high frequency is most quantized.

  FIG. 23 is a flowchart showing an operation process of image coding of the MRC model by the image coding apparatus according to Embodiment 3 of the present invention. Compared with the processing described in FIG. 11, the image encoding operation processing here is performed by decomposing twice foreground 1 and foreground 2 using a 9 × 7 filter and encoding a predetermined lower bit plane. The difference is that the quantization is performed while omitting.

  More specifically, in the image encoding operation process shown in FIG. 23, after the input of the original image, as in the case of the operation process of FIG. The process of creating the second mask 2 (step S21), the color of the non-character pixel that does not belong to the character (line drawing) region is the color of the character pixel that is closest to the pixel. The process of replacing and creating the second foreground 2 (step S22), replacing the color of the pixel in the character (line drawing) area in the original image with the color of the non-character pixel located closest to the pixel, Processing for creating the foreground 1 (step S23), processing for creating the first mask 1 as an image having the same pixel size 1 as the original image (step S24), creating a background image as a white image having the same size as the original image The process of (Step S25) is performed.

  Then, the first foreground 1 is decomposed twice using a 9 × 7 filter, quantized under the conditions shown in FIG. 21, and then encoded under the conditions shown in FIG. 14 (step S26). The second foreground 2 is decomposed twice using a 9 × 7 filter, quantized under the conditions shown in FIG. 22, and then encoded under the conditions shown in FIG. 15 (step S27).

  Thereafter, as in the case of the operation process of FIG. 11, the process of encoding the first mask 1 and the second mask 2 (both binary images) with the well-known MMR specification (step S28), the background. JPM coding (entropy coding is not performed, background color is specified as a code: step S29), all foreground, mask, and background codes are combined, and a desired header is added to the JPM code Is acquired (step S30).

  Here, with regard to the layout and size information of each layout object with respect to the background in JPM, the layout object ID, the stacking order, the number of vertical and horizontal pixels, and the layout offset with respect to the background are described in the Layout Object Header Box. In addition, for information on the arrangement and size of the foreground and background constituting each layout object, an arrangement offset with respect to the background is described in the Object Header Box. Further, the number of pixels of each object is described in the code of each object.

  In each of the embodiments described above, JPM is used as the file format. However, any file format may be used as long as it is an MRC type, and an MRC type PDF is also in its category. In each embodiment, the case where the wavelet transform is used for the frequency transform is illustrated. However, the present invention is not limited to this, and other frequency transforms represented by the discrete cosine transform can be applied. Furthermore, the above-mentioned “high-frequency component” includes only the highest frequency band (coefficient), at least the highest frequency band (coefficient), and the lowest frequency band (coefficient). It shows two ways of not including.

  The present invention can be applied to general apparatuses for encoding multi-layer images in addition to general-purpose image processing apparatuses and image forming apparatuses.

1 Data bus 2 HDD
3 RAM
4 CPU

Japanese Patent No. 3275807 JP 2007-5844 A

Claims (12)

  An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image An image encoding device comprising: generation means; and image encoding means for individually encoding the first image to the third image generated by the image generation means, wherein the image encoding means Comprises frequency transform coding means for frequency transforming and coding the first image and the second image, the frequency transform coding means for quantizing high frequency components in the second image. An image coding apparatus characterized in that the degree is made stronger than the degree of quantization of high-frequency components in the first image. An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image An image encoding device comprising: generation means; and image encoding means for individually encoding the first image to the third image generated by the image generation means, wherein the image encoding means Has a luminance / color difference conversion encoding means for encoding the first image and the second image by luminance / color difference conversion, and the luminance / color difference conversion encoding means in the second image the degree of quantization of the color difference component on the degree of quantization of luminance, the image coding apparatus characterized by smaller than the degree of quantization of the color difference component on the degree of quantization of luminance in the first image.   An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image An image encoding device comprising: generation means; and image encoding means for individually encoding the first image to the third image generated by the image generation means, wherein the image encoding means Has a luminance / color difference conversion encoding means for encoding the first image and the second image by luminance / color difference conversion, and the luminance / color difference conversion encoding means in the second image An image coding apparatus characterized in that the high frequency component of luminance is most quantized and the high frequency component of color difference in the first image is most quantized.   4. The image encoding device according to claim 1, wherein the image encoding unit performs the quantization process by quantizing a frequency coefficient, discarding lower bits of the frequency coefficient, the quantization, and the like. An image coding apparatus characterized by performing any one of the discard combinations.   An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image An image encoding method comprising: a generation step; and an image encoding step for individually encoding the first image to the third image generated in the image generation step, wherein the image encoding step includes: The first image and the second image are frequency-transformed and encoded, and the degree of quantization of the high-frequency component in the second image is greater than the degree of quantization of the high-frequency component in the first image. An image encoding method comprising a frequency transform encoding step for strengthening. An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image An image encoding method comprising: a generation step; and an image encoding step for individually encoding the first image to the third image generated in the image generation step, wherein the image encoding step includes: the degree of quantization of the first image and the second image coded by converting luminance and color difference, the color difference components for the degree of quantization of luminance in the second image, in said first image image encoding method characterized by having a luminance and color difference conversion encoding step to be smaller than the degree of quantization of the color difference component on the degree of quantization of luminance.   An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image An image encoding method comprising: a generation step; and an image encoding step for individually encoding the first image to the third image generated in the image generation step, wherein the image encoding step includes: The first image and the second image are encoded by luminance / color difference conversion, and the high frequency component of luminance in the second image is most quantized, and the high frequency component of color difference in the first image is An image encoding method comprising a luminance / color difference transform encoding step that is most quantized.   8. The image encoding method according to claim 5, wherein the image encoding step includes quantizing the quantization with a frequency coefficient, discarding lower bits of the frequency coefficient, the quantization, and the discarding. An image encoding method, which is performed by any one of the combinations. An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image A recording medium for image encoding in which processing information of a generation step and processing information of an image encoding step for individually encoding the first image to the third image generated in the image generation step are recorded. Then, the processing information of the image encoding step is encoded by frequency-converting the first image and the second image, and the degree of quantization of the high-frequency component in the second image is determined by the first information. Processing information of a frequency conversion encoding step for making the frequency component higher than the degree of quantization of the high-frequency component in the image, and the first image and the second image are encoded by luminance / color difference conversion, and the second image is encoded. The amount of luminance at The degree of quantization of the color difference component on the extent of reduction, the process information of the first luminance and color difference conversion encoding step to be smaller than the degree of quantization of the color difference component on the degree of quantization of luminance in the first image The first image and the second image are encoded by luminance / color difference conversion, and the high frequency component of luminance in the second image is most quantized, and the high frequency component of color difference in the first image is also quantized. A recording medium for encoding an image, characterized by having at least one of processing information of a second luminance / color-difference conversion encoding step that most quantifies the image.   10. The recording medium for image encoding according to claim 9, wherein the processing information of the image encoding step includes quantization of a frequency coefficient, discarding lower bits of the frequency coefficient, a combination of the quantization and the discarding. An image encoding recording medium comprising: instruction information to be performed by any of the above. An image for generating a first image that is a picture part image, a second image that is an image showing the color of the non-picture part, and a third image that is an image showing the shape of the non-picture part from the color input image Processing information of the generation step, and processing information of the image encoding step for individually encoding the first image to the third image generated in the image generation step, and causing the information processing apparatus to execute the processing information. An image encoding program for processing, wherein the processing information of the image encoding step is obtained by frequency-converting and encoding the first image and the second image, and quantizing high-frequency components in the second image. Processing information of a frequency conversion encoding step for making the degree of conversion stronger than the degree of quantization of the high frequency component in the first image, and luminance / color difference conversion of the first image and the second image Mark However, the first to be smaller than the degree of quantization of the degree of quantization of the color difference component on the degree of quantization of luminance in the second image, the color difference component on the degree of luminance quantization in the first image The luminance / color difference conversion encoding step processing information and the first image and the second image are encoded by luminance / color difference conversion, and the high-frequency component of luminance in the second image is most quantized. An image encoding program comprising: at least one of processing information of a second luminance / color difference conversion encoding step that most quantizes a high-frequency component of color difference in the first image.   12. The image encoding program according to claim 11, wherein the processing information of the image encoding step includes any one of a combination of quantization of a frequency coefficient, discard of lower bits of the frequency coefficient, the quantization, and the discard. An image encoding program characterized by including instruction information to be performed.