JP4100627B2 - Image processing apparatus, image processing method, program, and information recording medium - Google Patents

Description

  The present invention relates to the field of image processing, and more particularly to an apparatus and method for performing processing for frequency characteristic control on code data compressed with an image.

  Image sharpening and smoothing are basic techniques for improving and restoring image quality. Image quality enhancement is enhancement of image features such as edges, borders, contrast, and includes noise reduction, edge sharpening and sharpening, filtering, and other well-known processes. A popular sharpening method is called unsharp masking, which filters images using a special high-pass filter. One drawback of unsharp masking is that even the noise in the image is sharpened and often looks unnatural.

  To solve the problem of sharpening such noise pixels, as described in Patent Document 1, wavelet transform is applied to an image to decompose it into a plurality of decomposition levels, and at least two of these decomposition levels are included. A technique for scaling coefficients at one decomposition level using scale-dependent parameters separately for each decomposition level is effective.

  Now, image data is very often compressed prior to storage or transmission. JPEG is currently widely used as an encoding method for compressing still images, but JPEG 2000, which was internationally standardized in 2001, is also expected as a standard encoding method succeeding JPEG.

  FIG. 28 is a block diagram showing a JPEG compression / decompression algorithm, and FIG. 30 is a block diagram showing a JPEG2000 compression / decompression algorithm. One of the differences between the two encoding methods is the difference in the applied frequency conversion.

  In JPEG, at the time of compression, discrete cosine transform (DCT) is applied to image data in units of blocks of 8 pixels × 8 pixels, and each block is converted into 64 coefficients as shown in FIG. Such DCT coefficients correspond to one frequency band, each representing a specific horizontal / vertical frequency component.

In JPEG2000, upon compression, an image is divided into rectangular areas that do not overlap known as 1 or more tiles, 2-dimensional wavelet transform in tile a plurality of frequency bands of the multi-resolution by applying the (discrete wavelet transform) (Sub (Band). For example, the tile image as shown in FIG. 31A is decomposed into four frequency bands as shown in (b) by applying a one-dimensional discrete wavelet transform in order in the vertical and horizontal directions, Applying the same two-dimensional wavelet transform to the 1LL band to decompose into a frequency band as shown in (c), and further applying the two-dimensional wavelet transform to the 2LL band into a frequency band as shown in (d). Repeat the above operation as many times as necessary. In the figure, numerals prefixed names of frequency band shows a decomposition level is the number (decomposition count) of the applied two-dimensional wavelet transform before the coefficient is obtained right L or H in the vertical direction L or H on the left side means a low frequency component or a high frequency component by horizontal wavelet transformation. It is also possible to apply wavelet transform only in one dimension.

  The present invention uses an encoding method, such as JPEG and JPEG2000, that decomposes image data into components of a plurality of frequency bands by discrete cosine transform and wavelet transform, and quantizes each frequency band component and then performs variable length coding. The code data of the compressed image is to be processed.

  For example, there are Patent Documents 2 and 3 as prior documents related to the present invention.

  Patent Document 2 discloses a compression processing unit that performs DCT conversion on input image data, quantizes the input image data, performs variable length coding, and transfers the compressed data as compressed data, and variable length decoding of the transferred compressed data. A first quantization matrix used for quantization in the compression processing means and an inverse in the decompression processing means. An image compression encoding and image compression decoding apparatus is described in which the second quantization matrix used for quantization is at least relatively different.

  Patent Document 3 discloses a color image transmission apparatus including means for converting a color image into source image data and a compression engine that compresses the source image data into compressed image data. The compression engine converts the source image data into source image data. Converting the converted image data into quantized image data according to conversion means for converting into converted image data, means for storing a first quantization table comprising a plurality of elements, and elements included in the first quantization table And means for storing a second multi-element quantization table related to the first quantization table but not the same, and using the entropy table to convert the quantized image data into encoded image data Entropy encoding means for converting, and the encoded image data, the second quantization table, and the entropy table are captured. And encapsulation means for forming encapsulated data file by Le of, and means for transmitting the encapsulated data files, further comprises a color image transmission apparatus is described.

JP 2001-212327 A JP-A-9-224246 JP-A-8-289324

  An object of the present invention is to provide a novel image processing apparatus and image processing method capable of flexibly controlling frequency characteristics (frequency characteristics of image data obtained by decompressing code data) without decompressing code data of an image. Is to provide.

The invention of claim 1
First means for reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second means for reading frequency characteristic information for changing the frequency characteristic;
Third means for setting a correction value of the quantization parameter for each frequency band according to the frequency characteristic information read by said second means,
A correction set by the third means for the quantization parameter for each frequency band, which is recorded in the code data read by the first means and applied to the quantization when the code data is generated And a fourth means for correcting the value.

According to a second aspect of the present invention, in the image processing apparatus according to the first aspect, the third means sets a correction value for each region of the image.

According to a third aspect of the present invention, the value of the quantization parameter applied to the quantization at the time of generation of the code data , wherein the third means is recorded in the code data read by the first means. 3. The image processing apparatus according to claim 1, wherein the correction value of the quantization parameter is set according to the frequency characteristic information read by the second means with reference to the above .

The invention of claim 4 further includes fifth means for analyzing the frequency characteristics of the image from the code data read by the first means, and the third means is read by the second means. 3. The image processing apparatus according to claim 1, wherein a correction value of the quantization parameter is set according to the frequency characteristic information and the frequency characteristic analyzed by the fifth means .

According to the invention of claim 5, the second means reads visual characteristics together with frequency characteristic information for changing the frequency characteristics, and the third means reads the frequency characteristic information read by the second means and 3. The image processing apparatus according to claim 1, wherein the correction value of the quantization parameter is set according to the visual characteristic.

According to a sixth aspect of the present invention, the second means reads image area information together with frequency characteristic information for changing frequency characteristics, and the third means reads image area information read by the second means. The image processing apparatus according to claim 2, wherein the correction value of the quantization parameter is set according to the frequency characteristic information read by the second means for each image area.

In the invention of claim 7, the second means reads the character area information and the character component frequency band information together with the frequency characteristic information for changing the frequency characteristic, and the third means is the second means. Based on the read character area information, the character area is distinguished from other areas, and the character area is specified by the specific frequency band specified by the character component frequency band information read by the second means and others. 3. The image processing apparatus according to claim 2, wherein the correction value of the quantization parameter is set in accordance with the frequency information read by the second means by distinguishing the frequency band from the second frequency band.

The invention of claim 8, wherein the quantization parameter into loaded code data by the first means is recorded in the form of a quantized representative value, said third means a correction value of the quantization representative value set, characterized in that said fourth means modifies said quantized representative value that is recorded in the loaded code data by the first means, the correction value set by the third means An image processing apparatus according to any one of claims 1 to 7.

According to a ninth aspect of the invention, a quantization parameter is recorded in the form of a quantization index in the code data read by the first means , and the third means sets a correction value of the quantization index. , claims the fourth means, the quantization index which is recorded in the code data read by the serial first means, characterized in that to correct the correction value set by the third means An image processing apparatus according to any one of Items 1 to 7.

In the invention of claim 10, when the quantization parameter is recorded in the form of a quantized representative value in the code data read by the first means, the fourth means is the third means. The quantization parameter correction value set in step (2) is converted into a quantization representative value correction value, and the quantized representative value recorded in the code data read by the first means is converted into the converted correction value. an image processing apparatus according to the invention of any one of claims 1 to 7, characterized in that to fix the.

In the invention of claim 11, when the quantization parameter is recorded in the form of a quantization index in the code data read by the first means, the fourth means is provided by the third means. The set correction value of the quantization parameter is converted into the correction value of the quantization index, and the quantization index recorded in the code data read by the first means is corrected to the converted correction value . An image processing apparatus according to any one of claims 1 to 7, wherein the image processing apparatus according to any one of claims 1 to 7 is provided.

The invention of claim 12 includes a first step of reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second step of reading frequency characteristic information for changing the frequency characteristic;
A third step of setting a correction value of a quantization parameter for each frequency band in accordance with the frequency characteristic information read in the second step;
The correction parameter set in the third step , which is recorded in the code data read in the first step and applied to the quantization at the time of generation of the code data, is set in the third step. It is an image processing method characterized by having the 4th process corrected to a value.

The invention of claim 13 includes a first step of reading code data generated by variable length coding after image data is decomposed into a plurality of frequency band components and quantized, and
A second step of reading frequency characteristic information for changing the frequency characteristic;
A third step of setting a correction value of the quantized representative value according to the frequency characteristic information read by the second step;
The quantization representative value recorded in the code data read in the first step and applied to the quantization at the time of generation of the code data is corrected to the correction value set in the third step. And an image processing method characterized by comprising: a fourth step.

The invention of claim 14 includes a first step of reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second step of reading frequency characteristic information for changing the frequency characteristic;
A third step of setting a correction value of the quantization index according to the frequency characteristic information read by the second step;
The quantization index recorded in the code data read in the first step and applied to the quantization at the time of generation of the code data is corrected to the correction value set in the third step. And an image processing method characterized by comprising a fourth step.

  A fifteenth aspect of the invention is a program that causes a computer to function as each unit of the image processing apparatus according to any one of the first to eleventh aspects of the invention.

  The invention of claim 16 is a program for causing a computer to execute each step of the image processing method of the invention of claim 12, 13 or 14.

  The invention of claim 17 is a computer-readable information recording medium on which the program of the invention of claim 15 or 16 is recorded.

  According to the first to fourteenth aspects, the quantization parameter for each frequency band recorded in the code data is corrected according to the frequency characteristic information, or the quantization parameter in the code data is a quantized representative value. Alternatively, if the quantization index is recorded, the quantization representative value or the quantization index is modified according to the frequency characteristic information, and the quantization parameter for each frequency band is indirectly modified according to the frequency characteristic information. Thus, it is possible to change the frequency characteristics as if edge enhancement filter processing, smoothing filter processing, and other filters are applied without decoding the code data, and flexibly by changing the frequency characteristic information. Various frequency characteristics can be controlled. When two-dimensional frequency conversion is used in encoding of code data, it is possible to control frequency characteristics in various directions in the same manner as when applying an edge enhancement filter or smoothing filter for each direction. Become. According to the invention of claim 2, it is possible to control the frequency characteristics in units of regions so that the frequency characteristics are different between the central portion and the peripheral portion of the image. According to the third aspect of the present invention, it is possible to control relative frequency characteristics based on the quantization parameter, the quantization representative value, or the quantization index recorded in the code data. According to the invention of claim 4, adaptive frequency characteristic control according to the original frequency characteristic is possible. Further, according to the invention of claim 5, the frequency characteristic control reflecting the visual characteristic can be performed as in the case of applying the filter incorporating the visual characteristic. According to the invention of claim 6, it is possible to perform different frequency characteristic control for each image area of the image. According to the seventh aspect of the invention, it is possible to control the frequency characteristics in which the character area is distinguished from the other area so that the image quality of the character area of the image is higher than that of the other area. In addition, according to the inventions of claims 15 to 17, it is possible to easily achieve the inventions of claims 1 to 14 by using a computer.

  Each step of the means constituting the image processing apparatus according to the present invention or the processing procedure of the image processing method according to the present invention can be realized by hardware, firmware, software, or a combination thereof. One typical embodiment of the present invention is a form realized by one or more programs in a general-purpose computer such as a personal computer or a microcomputer incorporated in various devices. One or more such programs and various information recording (storage) media that can be read by a computer such as a magnetic disk, an optical disk, a magneto-optical disk, and a semiconductor storage element on which the program is recorded are also included in the present invention.

  Hereinafter, embodiments of the present invention will be specifically described.

  In this embodiment, code data in which a quantization parameter for each frequency band is recorded is to be processed. In order to change the frequency characteristic of the code data, the quantization parameter in the code data according to the frequency characteristic information is used. To correct. Here, linear quantization is assumed, and the quantization parameter is a numerical value used as a quantization denominator for each frequency band (however, it may be a numerical value for uniquely determining the numerical value).

  An example of the quantization parameter for each frequency band is shown in FIG. FIG. 2 shows an example of the quantization parameter for each frequency band by the one-dimensional wavelet transform as shown in FIG.

  Depending on the format of the code data, the quantization parameter may be recorded in a specific header part or may be recorded in a form independent of the header part, either of which may be used. . The same applies to code data in which quantization representative values or quantization indexes are recorded instead of quantization parameters (see Examples 2 and 3 below).

  FIG. 15 is a block diagram for explaining the configuration of the image processing apparatus according to the present embodiment. The image processing apparatus 90 includes a code reading unit 91a that reads code data from an external storage device and the like, a quantization parameter reading unit 91b that reads a quantization parameter recorded therein from the code data read thereby, A frequency characteristic information reading unit 91e for reading frequency characteristic information for changing the frequency characteristic from the storage device, a quantization parameter setting unit 91d for setting a correction value of the quantization parameter according to the read frequency characteristic information, and By rewriting the quantization parameter read by the quantization parameter reading unit 91b of the original code data with the correction value by the quantization parameter setting unit 91d, only the quantization parameter of the original encoded data is corrected. It comprises a quantization parameter correction unit 91c that generates correct code data. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  FIG. 16 is a flowchart showing the flow of processing operations in the image processing apparatus 90. This embodiment is also an embodiment of the image processing method of the present invention, and FIG. 16 is also a flowchart showing the processing procedure.

  The processing operation will be described below. First, frequency characteristic information for changing frequency characteristics is read (frequency characteristic information reading unit 91e, step 11). Based on the read frequency characteristic information, a correction value of the quantization parameter is set (quantization parameter setting unit 91d, step 12).

  Next, code data is read (code reading unit 91a, step 14), and a quantization parameter of one frequency band recorded in the code data is read (quantization parameter reading unit 91b, step 16). If the quantization parameter of this frequency band is a correction target (step 17, YES), the quantization parameter is corrected to a correction value (quantization parameter correction unit 91c, step 18). The processing loop of steps 15 to 18 is repeated for each frequency band, and when the processing for the last frequency band is finished (step 15, YES), the quantization parameter correction processing is completed, and the quantization parameter is corrected according to the frequency characteristic information. New code data is generated. Note that the processing after step 14 is executed only when it is determined that the frequency characteristic needs to be changed (step 13, YES).

  Here, a specific example of how to obtain the frequency characteristic information and the correction value of the quantization parameter according to the frequency characteristic information will be described.

  The frequency characteristic information read by the frequency characteristic information reading unit 91e is, for example, a numerical value indicating the degree of enhancement of the coefficient for each frequency band based on human visual characteristics as shown in FIG. More specifically, for example, it is a numerical value indicating the ratio of the coefficient before and after the change for each frequency band as shown in FIG.

The corrected value of the quantization parameter is
Q '= Q / R (S) Equation (1)
Is calculated by Here, Q ′ is a corrected quantization parameter, Q is a corrected quantization parameter, and R (S) is a coefficient conversion ratio in the frequency band S.
FIG. 4 shows values obtained by correcting the quantization parameter shown in FIG. 1 according to the correction equation (1) based on the frequency characteristic information shown in FIG.

  Here, when the quantization parameter q of a specific frequency band is changed to q ', the effect on the restored image by this change will be described with reference to FIG.

  In the encoding process, the frequency conversion coefficient of the image data is quantized. If the quantization parameter (quantization parameter before correction) is q and the coefficient value before quantization is W, the coefficient value after quantization is (Where, linear quantization is assumed, and the quantization parameter is the quantization denominator, that is, the number of quantization steps).

  Here, if q is corrected to q ′ according to the frequency characteristic information, the coefficient is inversely quantized with q ′ as the quantization parameter at the time of decoding, and therefore the coefficient value after inverse quantization is q * W / q. Become.

That is, when the coefficient value W is desired to be enhanced or reduced by E times, the quantization parameter value may be corrected from E = q ′ / q so that q ′ = E * q. Therefore, by performing the same quantization parameter control for each frequency band according to the frequency characteristic information, the frequency characteristic of the code data, in other words, the frequency characteristic of the decoded image can be changed. For example, the same effect as the edge enhancement filter process can be obtained by enhancing the high frequency component, and the same effect as the smoothing filter process can be obtained by reducing the high frequency component.

  So far, the description has been made assuming that one-dimensional frequency conversion (band division) is used. Next, a case where two-dimensional frequency conversion (band division) is used will be described.

FIG. 5 shows an example of the quantization parameter for each frequency band divided by two-dimensional (horizontal and vertical direction) frequency conversion. In the frequency band in each direction, level 1 is the lowest frequency band, and level 7 is the highest frequency band. Here, the two-dimensional frequency band is represented as (Sx, Sy). However, Sx is a horizontal frequency band level , and Sy is a vertical frequency band level . In the example of FIG. 5, the value of the quantization parameter for the frequency band (1, 1) is 1, the value of the quantization parameter for the frequency band (2, 1) is 2, and so on. For example, in the case of discrete cosine transform used in JPEG, the frequency band (1, 1) corresponds to a DC component (or DC component and a low-frequency AC component), and a higher frequency band corresponds to another AC. It corresponds to the component (see FIG. 29).

The frequency characteristic information in this case designates the conversion ratio of the coefficient for each two-dimensional frequency band based on the human visual characteristic as shown in FIG. 9 , and the quantization parameter setting unit 91d Using the conversion ratio, the correction value of the quantization parameter for each frequency band is calculated by Equation (1).

  FIG. 6 shows an example of a quantization parameter value for a frequency band when using a two-dimensional wavelet transform such as JPEG2000, and FIG. 7 shows an example of a coefficient conversion ratio (that is, frequency characteristic information) for each frequency band. FIG. 8 shows values obtained by correcting the quantization parameter values of FIG. 6 using the conversion ratio of FIG. Here, the relationship between the frequency band by the two-dimensional wavelet transform and the frequency band in each direction is as shown in FIG. Note that the mark * in FIG. 7 means that the corrected quantization parameter value is 0 (that is, the coefficient of the frequency band becomes 0 by inverse quantization).

  In this way, when two-dimensional frequency conversion is used, a filter process that increases or decreases a specific horizontal / vertical direction frequency component, for example, frequency characteristics similar to the edge enhancement filter process or the smoothing filter process for each direction. The frequency characteristics can be controlled flexibly by changing the frequency characteristic information.

  In this embodiment, code data in which a quantized representative value is recorded is processed. In order to change the frequency characteristic of the code data, the quantized representative value in the code data is modified according to the frequency characteristic information. To do.

  Here, the quantization representative value is a numerical value that determines a quantization parameter for a specific frequency band, and is a numerical value that is a basis for determining the quantization parameter for each other frequency band by calculation. For example, in JPEG2000, the quantization parameter is determined by calculation from a pair of exponent and mantissa (εb, μb), but a method of specifying the pair of exponent and mantissa for each decomposition level (explicit quantization method) And specify only the exponent and mantissa pair for the uppermost decomposition level LL subband and use it to calculate the mantissa and exponent pair for the remaining subbands (implicit or implicit) Called a standard quantization method). A pair of exponent and mantissa specified in the latter implicit quantization method is an example of the quantization representative value.

  That is, in this embodiment, the quantization representative value is corrected according to the frequency characteristic information, so that the quantization parameter for each frequency band is indirectly corrected according to the frequency characteristic information.

FIG. 17 is a block diagram for explaining the configuration of the image processing apparatus according to the present embodiment. The image processing apparatus 100 includes a code reading unit 101a that reads code data from an external storage device and the like, and a quantized representative value reading unit 101b that reads a quantized representative value recorded therein from the read code data. , A frequency characteristic information reading unit 101e that reads frequency characteristic information for changing frequency characteristics from an external storage device or the like, and a quantized representative value setting unit that sets a correction value of the quantized representative value according to the read frequency characteristic information 101d, and, the original code data, by rewriting the quantization representative value read quantized representative value reading unit 101b to fix value due to quantization representative value setting unit 101d, the quantization of the original encoded data representative The quantization representative value correcting unit 101c generates new code data in which only the value is corrected. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  FIG. 18 is a flowchart showing the flow of processing operations in the image processing apparatus 100. This embodiment is also an embodiment of the image processing method of the present invention, and FIG. 18 is also a flowchart showing the processing procedure.

  The processing operation will be described below. First, frequency characteristic information for changing frequency characteristics is read (frequency characteristic information reading unit 101e, step 21). Based on the read frequency characteristic information, a correction value of the quantized representative value is set (quantized representative value setting unit 101d, step 22).

  Next, code data is read (code reading unit 101a, step 24), and a quantized representative value recorded in the code data is read (quantized representative value reading unit 101b, step 25). Then, by correcting the quantized representative value to the corrected value, code data in which the quantized representative value is corrected (frequency characteristic is changed) is generated (quantized representative value correcting unit 101c, step 26). Note that steps 24 to 26 are executed only when it is determined that the frequency characteristic needs to be changed (step 23, Yes).

  In this embodiment, code data in which a quantization index is recorded is to be processed, and the quantization index in the code data is modified according to the frequency characteristic information in order to change the frequency characteristic of the code data. Here, the quantization index is a numerical value associated with a set of quantization parameters for each frequency band. When decoding code data, a set of quantization parameters for each frequency band for inverse quantization can be obtained from the value of the quantization index in the code data. That is, in the present embodiment, the quantization parameter set for each frequency band is indirectly modified according to the frequency characteristic information by correcting the quantization index according to the frequency characteristic information.

  The pair of exponent and mantissa specified for each decomposition level in the explicit quantization method of JPEG 2000 described in connection with the second embodiment is an example of the quantization index. From one set of the specified exponent and mantissa, the quantization parameter for each frequency band at the same decomposition level is determined by calculation.

  FIG. 19 is a block diagram for explaining the configuration of the image processing apparatus according to the present embodiment. The image processing apparatus 80 includes a code reading unit 81a that reads code data from an external storage device and the like, a quantization index reading unit 81b that reads a quantization index recorded therein from the code data read thereby, A frequency characteristic information reading unit 81e for reading frequency characteristic information for changing the frequency characteristic from the storage device, a quantization index setting unit 81d for setting a correction value of the quantization index according to the frequency characteristic information read thereby, and By rewriting the quantization index read by the quantization index reading unit 81b of the original code data with the correction value by the quantization index setting unit 81d, a new quantization index of only the original encoded data is corrected. Quantization index correction unit 8 for generating code data 1c. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  FIG. 20 is a flowchart showing the flow of processing operations in the image processing apparatus 80. This embodiment is also an embodiment of the image processing method of the present invention, and FIG. 20 is also a flowchart showing the processing procedure.

  The processing operation will be described below. First, frequency characteristic information for changing frequency characteristics is read (frequency characteristic information reading unit 81e, step 31). Based on the read frequency characteristic information, a modification value of the quantization index is set (quantization index setting unit 81d, step 32).

  Next, the code data is read (code reading unit 81a, step 34), and the quantization index recorded in the code data is read (quantization index reading unit 81b, step 35). Then, by correcting the quantization index to the correction value, code data in which the quantization index is corrected (frequency characteristics are changed) is generated (quantization index correction unit 81c, step 36).

  Steps 34 to 36 are executed only when it is determined that the frequency characteristics need to be changed (step 33, Yes).

  The recording format for the quantization parameter in the code data is not uniform. For example, as described in the first to third embodiments, the quantization parameter for each frequency band (or the numerical value for determining it) is recorded, the quantization parameter for a specific frequency band (or There is a format in which a quantized representative value is recorded or a format in which a quantization index is recorded. In this embodiment, the frequency characteristics are controlled by correcting the quantization parameters according to the frequency characteristic information for code data having different recording formats related to the quantization parameters.

  The configuration of the image processing apparatus according to the present embodiment is basically the same as that of the first embodiment, and will be described with reference to FIG. The quantization parameter correction unit 91c in the image processing apparatus of the present embodiment has the same basic function as that of the first embodiment, but the conversion of the recording format related to the quantization parameter is performed as necessary. This is different from the case of the first embodiment.

  FIG. 21 is a flowchart for explaining the processing operation in the image processing apparatus of the present embodiment. This embodiment is also an embodiment of the image processing method of the present invention, and FIG. 21 is a flowchart showing the processing procedure.

  First, frequency characteristic information for changing the frequency characteristic is read (frequency characteristic reading unit 91e, step 41), and a correction value of the quantization parameter is set based on the frequency characteristic information (quantization parameter setting unit 91d, step 42). ). Next, code data is read (code reading unit 91a, step 44), and a part (quantization parameter, quantized representative value or quantization index) related to the quantization parameter is read (quantization parameter reading unit 91b, step 45). . Then, the quantization parameter correcting unit 91c determines whether or not conversion of the recording format of the quantization parameter is necessary (step 46).

Here, the recording format of the quantization parameter for each frequency band is set as a reference recording format, and the quantization parameter correction value for each frequency band is also set in the quantization parameter setting unit 91d . . When the recording format in the code data is the reference format, it is determined that conversion of the recording format is unnecessary, and the quantization parameter in the code data is set to the value set by the quantization parameter setting unit 91d in the quantization parameter correction unit 91c. It is corrected, and new code data after correction is generated (step 47). When the recording format is not the standard format, that is, when the quantization representative value or the quantization index is recorded, the quantization parameter set by the quantization parameter setting unit 91d is set in the quantization parameter correction unit 91c. The corrected value is converted into a corresponding quantized representative value or quantization index, and the quantized representative value or quantized index in the code data is rewritten by the quantized representative value or quantized index after conversion, and the quantized value is converted. New code data modified with respect to the parameters is generated (step 48).

  In addition, when it is judged that the change of a frequency characteristic is not required (step43, NO), the process of steps 44-48 is not performed.

  When the quantization parameter for each frequency band is recorded in the original code data, the recording format is changed and rewritten to the quantization representative value or the quantization index in order to reduce the data amount. Such a mode is also included in the present invention.

  The present embodiment is intended for code data whose quantization parameter can be controlled in image area units, and is intended to control the frequency characteristics of the code data for each area.

  The configuration of the image processing apparatus according to the present embodiment is basically the same as that of the first embodiment, and will be described with reference to FIG.

  In the image processing apparatus according to the present embodiment, the code data is read by the code reading unit 91a, and the quantization parameter of each of the plurality of areas recorded in the code data is read by the quantization parameter reading unit 91b. Further, the frequency characteristic information reading unit 91e reads frequency characteristic information for changing the frequency characteristic related to at least one region. In the quantization parameter setting unit 91d, the correction value of the quantization parameter for each frequency band for each region is set according to the frequency characteristic information. Then, in the quantization parameter correction unit 91c, the quantization parameter for each frequency band of each region recorded in the code data is rewritten with the correction value set by the quantization parameter setting unit 91d. New code data in which is corrected is generated.

  By correcting the quantization parameter for each region as described above, it is possible to control the frequency characteristics in units of regions, for example, making the frequency characteristics different between the central portion and the peripheral portion of the image.

  Similarly, for code data in which a quantized representative value or area index is recorded in units of areas, the frequency characteristics of the code data are also corrected by correcting the quantized representative value or quantization index in units of areas. It is obvious that can be changed in units of regions. Such embodiments are also encompassed by the present invention.

  Since the configuration of the image processing apparatus according to the present embodiment is basically the same as that of the first embodiment, FIG. In the image processing apparatus according to the present embodiment, the value of the quantization parameter recorded in the code data is also given to the quantization parameter setting unit 91d. In the quantization parameter setting unit 91d, the correction value of the quantization parameter is not determined based only on the frequency characteristic information for changing the frequency characteristic, but the value of the quantization parameter recorded in the code data is used as a reference. As described above, the quantization parameter correction value is set by increasing or decreasing the quantization parameter value according to the frequency characteristic information (for example, increasing or decreasing by 10%).

  FIG. 22 is a flowchart illustrating the flow of processing operations in the image processing apparatus according to the present embodiment. This embodiment is also an embodiment of the image processing method of the present invention, and FIG. 22 is also a flowchart showing the processing procedure.

  First, frequency characteristic information for changing frequency characteristics is read by the frequency characteristic information reading unit 91e (step 51). Next, code data is read by the code reading unit 91a, and the quantization parameter is read by the quantization parameter reading unit 91b (step 52). The read quantization parameter is passed to both the quantization parameter modification unit 91c and the quantization parameter setting unit 91d.

  In the quantization parameter setting unit 91d, as described above, the value of the passed quantization parameter is used as a reference, the value is increased or decreased according to the frequency characteristic information, and the increased or decreased value is used as a modification value of the quantization parameter. Set (step 53).

  Then, in the quantization parameter correction unit 91c, if the quantization parameter for each frequency band is a change target (step 56, YES), it is changed to the set correction value (step 57). When the quantization parameter changing process is completed for all frequency bands (step 55, YES), the process is terminated. If it is determined that the frequency characteristics need not be changed (step 54, NO), steps 55 to 57 are not executed.

  As described above, in this embodiment, the correction value is set in consideration of the value of the quantization parameter already recorded in the code data. For example, by modifying the quantization parameter value of a certain frequency band to be reduced by 10%, the coefficient of the frequency band is increased by 10% from the original value by inverse quantization. That is, according to the present embodiment, the value of the original quantization parameter is changed by relatively changing the value of the quantization parameter according to the frequency characteristic information with reference to the value of the original quantization parameter of the code data. The frequency characteristic can be changed relative to the frequency characteristic used when decoding.

  When code data in which a quantized representative value or quantization index is recorded is targeted, the value is increased or decreased according to the frequency characteristic information with reference to the original value of the quantized representative value or quantized index. The corrected value can be set. In addition, when code data in a format in which the quantization parameter, the quantization representative value, or the quantization index can be controlled for each region of the image, it is obvious that the same correction can be performed for each region. is there. Such an embodiment is also included in the present invention.

  FIG. 23 is a block diagram for explaining the configuration of the image processing apparatus according to the present embodiment. The image processing apparatus 130 includes a code reading unit 131a that reads code data from an external storage device and the like, a quantization parameter reading unit 131b that reads a quantization parameter recorded therein from the code data read thereby, Frequency characteristic information reading unit 131e for reading frequency characteristic information for changing the frequency characteristic from the storage device, the frequency characteristic analysis unit 131f for analyzing the frequency characteristic of the image from the code data, and the frequency characteristic of the image obtained thereby And a quantization parameter setting unit 131d for setting a modification value of the quantization parameter according to the frequency characteristic information for changing the frequency read by the frequency characteristic information reading unit 131e, and a quantization parameter reading unit for the original code data The quantization parameter read in 131b is converted into the quantization parameter. By rewriting the correction value set by the setting unit 131d, the quantization parameter correction unit 131c generates new code data in which only the quantization parameter of the original encoded data is corrected. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  The analysis method in the frequency characteristic analysis unit 131f is not particularly limited. For example, the code data is not decoded and the code amount for each frequency band is analyzed in the code state, the code data is decoded, and the decoded image data is used. An analysis method can be used.

  In this embodiment, the quantization parameter setting unit 131d can determine the correction value of the quantization parameter in accordance with the frequency characteristics obtained by analyzing the code data. According to such a correction value determination method, the frequency characteristic after the quantization parameter correction can be adaptively controlled according to the frequency characteristic of the image before the correction. That is, adaptive filter processing according to the frequency characteristic before correction is possible.

  FIG. 24 is a flowchart illustrating the flow of processing operations in the image processing apparatus according to the present embodiment. This embodiment is also an embodiment of the image processing method of the present invention, and FIG. 24 is also a flowchart showing the processing procedure.

  First, code data is read by the code reading unit 131a, and the quantization parameter is read by the quantization parameter reading unit 131b (step 61). The frequency characteristic of the original image data of the code data is analyzed by the frequency characteristic analysis unit 131f, and the result is passed to the quantization parameter setting unit 131d (step 62). Next, frequency characteristic information for changing the frequency characteristic is read by the frequency characteristic information reading unit 131e (step 63). In the quantization parameter setting unit 131d, the correction value of the quantization parameter is set by the method described above based on the frequency characteristic of the original image and the frequency characteristic information for changing the frequency characteristic (step 64). .

  Then, if the quantization parameter for each frequency band is to be changed (step 67, YES), the quantization parameter correction unit 131c changes it to the set correction value (step 68). When the quantization parameter changing process is completed for all frequency bands (step 66, YES), the process is terminated. If it is determined that the frequency characteristics need not be changed (step 65, NO), the processing of steps 66 to 68 is not executed.

  When code data in which a quantized representative value or quantization index is recorded is targeted, the quantized representative value or quantization index can be similarly modified. In addition, when code data in a format in which the quantization parameter, the quantization representative value, or the quantization index can be controlled for each region of the image, it is obvious that the same correction can be performed for each region. is there. Such an embodiment is also included in the present invention.

  Alternatively, the frequency characteristic analysis unit 131f may be omitted, and the frequency characteristic of the original image of the code data may be captured from the outside. Such a configuration is also included in the present invention.

  FIG. 25 is a block diagram for explaining the configuration of the image processing apparatus according to the present embodiment. The image processing apparatus 140 according to the present embodiment includes a code reading unit 141a that reads code data from an external storage device and the like, and a quantization parameter reading that reads a quantization parameter recorded therein from the code data read thereby. 141b, a frequency characteristic information reading unit 141e for reading frequency characteristic information for changing frequency characteristics and an applied visual characteristic from an external storage device, etc., and a quantization parameter of the quantization parameter according to the read frequency characteristic information and the visual characteristic. By rewriting the quantization parameter read by the quantization parameter reading unit 141b in the quantization parameter setting unit 141d for setting the correction value and the original code data to the correction value by the quantization parameter setting unit 141d, New code with only the quantization parameter of the encoded data modified It comprises a quantization parameter correction unit 141c that generates signal data. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  As is clear from the above description, it is a feature of the present embodiment that the visual characteristics captured from the outside are reflected when the correction value of the quantization parameter is determined. In addition, the structure which memorize | stores the visual characteristic to apply inside an apparatus is also possible, and such a structure is also contained in this invention.

  In the present embodiment, similar to the fifth embodiment, code data whose quantization parameters can be controlled in units of image regions are targeted.

  FIG. 26 is a block diagram illustrating the configuration of the image processing apparatus according to the present embodiment. The image processing apparatus 150 according to the present embodiment includes a code reading unit 151a that reads code data from an external storage device and the like, and a quantization parameter reading that reads a quantization parameter recorded therein from the code data read thereby. 151b, a frequency characteristic information reading unit 151e for reading frequency characteristic information and image area information for changing frequency characteristics from an external storage device, etc., and an image area based on the image area information according to the read frequency characteristic information Separately, the quantization parameter setting unit 151d for setting a correction value of the quantization parameter and the quantization parameter read by the quantization parameter reading unit 151b of the original code data are rewritten to the correction value by the quantization parameter setting unit 151d. As a result, only the quantization parameter of the original encoded data is corrected. It comprises a quantization parameter correction unit 151c that generates code data. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  As is apparent from the above description, different quantization parameters can be corrected for each image area based on image area information captured from the outside. For example, if the quantization parameter value of a specific frequency band is corrected to half the value of another image area with respect to a certain image area, the components of the specific frequency band in that image area will be The image area is emphasized twice as much. In this way, filter processing for each image area is possible.

  In addition, when the code data in which the quantization representative value or the quantization index is recorded are targeted, it is obvious that the quantization representative value or the quantization index can be similarly corrected for each image area. Such an embodiment is also included in the present invention.

  In the present embodiment, similar to the fifth embodiment, code data whose quantization parameters can be controlled in units of image regions are targeted. In addition, the present embodiment is effective when processing code data of an image including a character area and other areas as illustrated in FIG.

  FIG. 27 is a block diagram of the image processing apparatus according to the present embodiment. The image processing apparatus 160 according to this embodiment includes a code reading unit 161a that reads code data from an external storage device and the like, and a quantization parameter reading that reads a quantization parameter recorded therein from the code data read by the code reading unit 161a. Unit 161b, frequency characteristic information for changing frequency characteristics from an external storage device, etc., frequency characteristic information reading unit 161e for reading character region information and character component frequency band information, and frequency characteristic information read thereby The quantization parameter setting unit 161d that sets a modification value of the quantization parameter for each region based on the above, and the quantization parameter read by the quantization parameter reading unit 161b of the original code data are set to the quantization parameter setting unit 161d. By rewriting to the corrected value by, the quantization parameter of the original encoded data It consists of a quantization parameter correction unit 161c that generates new code data in which only the meter is corrected. The generated new code data is stored in, for example, an external storage device or transferred to the external device through a transmission path.

  Here, the character area information indicates the position of the character area. The character component frequency band information indicates a frequency band in which the character component is remarkably reflected. FIG. 11 shows an example of the code amount per unit area for each frequency band in the character region (an example measured for each frequency band of the one-dimensional wavelet transform). In the example of FIG. 11, since it can be seen that the character component is remarkably reflected in the frequency band 4H, for example, the frequency band 4H is designated by the character component frequency band information.

  In the quantization parameter setting unit 161d, the correction value of the quantization parameter for each frequency band is set according to the frequency characteristic information read by the frequency characteristic information reading unit 161e. Regarding the region specified by the character region information, For the quantization parameter of the frequency band designated by the character component frequency band information, for example, a large correction value is set so that the frequency band is emphasized.

  If the quantization parameter for the specific frequency band of the character area is corrected, the frequency characteristic of the image obtained by decoding the code data with the corrected quantization parameter is changed according to the frequency characteristic information as a whole, and the image quality of the character area is changed. Can be improved over other regions.

  It is obvious that the quantization representative value or quantization index focusing on the same character area can be corrected for code data in which the quantization representative value or quantization index is recorded for each area. Such an embodiment is also included in the present invention.

  As a modification of the present invention, the same means as the quantization parameter setting unit of each of the above embodiments is added to the device for decoding the code data, and the quantization parameter recorded in the code data is changed to each of the above. It is also possible to change the frequency characteristics of the restored image by making corrections in the same manner as in the embodiment and using the corrected quantization parameters for inverse quantization.

It is a figure which shows the example of the quantization parameter for every frequency band. It is a figure which shows the example of the quantization parameter for every frequency band. It is a figure which shows the example of the conversion ratio of the quantization parameter for every frequency band. It is a figure which shows the example of the correction value of the quantization parameter for every frequency band. It is a figure which shows the example of the quantization parameter for every frequency band. It is a figure which shows the example of the quantization parameter for every frequency band. It is a figure which shows the example of the conversion ratio of the quantization parameter for every frequency band. It is a figure which shows the example of the correction value of the quantization parameter for every frequency band. It is a figure which shows a visual characteristic. It is a figure which shows the example of the image area in an image. It is a figure which shows the example of the code amount per unit area for every frequency band in a character area. It is a figure which shows the example of a band division | segmentation by 1-dimensional wavelet transform. It is a figure which shows a response | compatibility with the frequency band by two-dimensional wavelet transform, and FIGS. It is a figure for demonstrating correction of a quantization parameter and its effect. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. 4 is a flowchart illustrating an example of a processing operation flow of the image processing apparatus according to the present invention. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. 4 is a flowchart illustrating an example of a processing operation flow of the image processing apparatus according to the present invention. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. 4 is a flowchart illustrating an example of a processing operation flow of the image processing apparatus according to the present invention. 4 is a flowchart illustrating an example of a processing operation flow of the image processing apparatus according to the present invention. 4 is a flowchart illustrating an example of a processing operation flow of the image processing apparatus according to the present invention. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. 3 is a flowchart showing a processing operation flow of the image processing apparatus according to the present invention. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. It is a block diagram which shows the structural example of the image processing apparatus which concerns on this invention. It is a block diagram which shows the basic algorithm of JPEG. It is a figure which shows the frequency band division | segmentation by discrete cosine transform in JPEG. It is a block diagram which shows the basic algorithm of JPEG2000. It is a figure which shows the frequency band division | segmentation by the two-dimensional discrete wavelet transform in JPEG2000.

Explanation of symbols

81a Code reading unit 81b Quantization index reading unit 81c Quantization index correction unit 81d Quantization index setting unit 81e Frequency characteristic information reading unit 91a Code reading unit 91b Quantization parameter reading unit 91c Quantization parameter correction unit 91d Quantization parameter setting unit 91e Frequency characteristic information reading unit 101a Code reading unit 101b Quantized representative value reading unit 101c Quantized representative value correcting unit 101d Quantized representative value setting unit 101e Frequency characteristic information reading unit 131a Code reading unit 131b Quantization parameter reading unit 131c Quantization Parameter correction unit 131d Quantization parameter setting unit 131e Frequency characteristic information reading unit 131f Frequency characteristic analysis unit 141a Code reading unit 141b Quantization parameter reading unit 141c Quantization parameter correction unit 141d Quantization Parameter setting unit 141e Frequency characteristic information etc. reading unit 151a Code reading unit 151b Quantization parameter reading unit 151c Quantization parameter modification unit 151d Quantization parameter setting unit 151e Frequency characteristic information etc. reading unit 161a Code reading unit 161b Quantization parameter reading unit 161c Quantization parameter correction unit 161d Quantization parameter setting unit 161e Frequency characteristic information reading unit

Claims (17)

First means for reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second means for reading frequency characteristic information for changing the frequency characteristic;
Third means for setting a correction value of the quantization parameter for each frequency band according to the frequency characteristic information read by said second means,
A correction set by the third means for the quantization parameter for each frequency band, which is recorded in the code data read by the first means and applied to the quantization when the code data is generated And a fourth means for correcting the value. The image processing apparatus according to claim 1, wherein the third unit sets a correction value for each region of the image. The third means is based on the value of the quantization parameter applied to the quantization at the time of generation of the code data recorded in the code data read by the first means . The image processing apparatus according to claim 1, wherein the correction value of the quantization parameter is set according to the frequency characteristic information read by the means of 2. A fifth means for analyzing the frequency characteristics of the image from the code data read by the first means ;
The third means sets the correction value of the quantization parameter in accordance with the frequency characteristic information read by the second means and the frequency characteristic analyzed by the fifth means. 2. The image processing apparatus according to 2. The second means reads visual characteristics together with frequency characteristic information for changing the frequency characteristics,
It said third means, the image processing apparatus according to claim 1 or 2, characterized in that to set the correction value of the quantization parameters in accordance with the frequency characteristic information and visual characteristics read by the second means. The second means reads image area information together with frequency characteristic information for changing frequency characteristics,
Said third means is characterized by setting for each image area based on the read image area information by the second means, in accordance with the frequency characteristic information read by said second means a correction value of the quantization parameter The image processing apparatus according to claim 2. The second means reads the character area information and the character component frequency band information together with the frequency characteristic information for changing the frequency characteristic,
Said third means, the second distinction between the character area and other areas on the basis of the character read area information by means and character components read by said second means for a character area The specific frequency band specified by the frequency band information is distinguished from other frequency bands, and the correction value of the quantization parameter is set according to the frequency information read by the second means. An image processing apparatus according to 1. A quantization parameter is recorded in the form of a quantization representative value in the code data read by the first means ,
The third means sets a correction value of the quantized representative value;
Said fourth means, wherein, characterized in that modifying the quantization representative value recorded in the loaded code data by said first means, the correction value set by the third means Item 8. The image processing apparatus according to any one of Items 1 to 7. A quantization parameter is recorded in the form of a quantization index in the code data read by the first means ,
The third means sets a correction value of the quantization index;
Said fourth means, claims, characterized in that to modify the quantization index which is recorded in the code data read by the first means, the correction value set by the third means The image processing apparatus according to any one of 1 to 7. When the quantization parameter is recorded in the form of the quantization representative value in the code data read by the first means, the fourth means is the quantization parameter set by the third means. Is converted into a corrected representative value of the quantized representative value, and the quantized representative value recorded in the code data read by the first means is corrected to the converted corrected value. The image processing apparatus according to any one of claims 1 to 7. When the quantization parameter is recorded in the form of a quantization index in the code data read by the first means, the fourth means includes the quantization parameter set by the third means . claims the correction value into a correction value of the quantization indices, the quantization index recorded in the code in the data read by said first means, characterized in that to correct the correction value the converted The image processing apparatus according to any one of 1 to 7. A first step of reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second step of reading frequency characteristic information for changing the frequency characteristic;
A third step of setting a correction value of a quantization parameter for each frequency band in accordance with the frequency characteristic information read in the second step;
The correction parameter set in the third step , which is recorded in the code data read in the first step and applied to the quantization at the time of generation of the code data, is set in the third step. An image processing method comprising a fourth step of correcting the value. A first step of reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second step of reading frequency characteristic information for changing the frequency characteristic;
A third step of setting a correction value of the quantized representative value according to the frequency characteristic information read by the second step;
The quantization representative value recorded in the code data read in the first step and applied to the quantization at the time of generation of the code data is corrected to the correction value set in the third step. image processing method characterized by comprising a fourth step of. A first step of reading code data generated by performing variable length coding after decomposing and quantizing image data into a plurality of frequency band components;
A second step of reading frequency characteristic information for changing the frequency characteristic;
A third step of setting a correction value of the quantization index according to the frequency characteristic information read by the second step;
The quantization index recorded in the code data read in the first step and applied to the quantization at the time of generation of the code data is corrected to the correction value set in the third step. An image processing method comprising: a fourth step.   A program that causes a computer to function as each unit of the image processing apparatus according to claim 1.   A program for causing a computer to execute each step of the image processing method according to claim 12, 13 or 14.   A computer-readable information recording medium on which the program according to claim 15 or 16 is recorded.

Images