JP2012175140A - Image processing apparatus and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing method capable of compressing image data including noise with a high compression ratio without losing the features of the compressed image.SOLUTION: The image processing apparatus is provided with: remainder component separation means which divides the pixel value of each pixel of an inputted image by 2 or an integer larger than 2, which is inputted by a user or preset, to separate the inputted image into quotient component data composed of the pixel value of the calculated quotient and remainder component data composed of the pixel value of the calculated remainder; and data compression means which compresses the respective components separated by the remainder component separation means and outputs compressed data of each component.

Description

  The present invention relates to an image processing apparatus and an image processing method for compressing image data.

Data compression is mainly used to reduce the storage area and transfer time when data is stored in a storage device or when data is transferred via a network.
Data compression includes lossless compression and lossy compression. The lossless compression can completely restore the data before compression, whereas the lossy compression cannot completely restore the data before compression.

Lossless compression is used for general file compression, but irreversible compression is also used for image and audio compression.
An example of lossless compression is Huffman coding (Non-Patent Document 1). This is to reduce the overall data amount by replacing a symbol that frequently appears in data with a short code and conversely with a symbol that has a low appearance frequency.
In addition, in lexical coding represented by LZ77, when a phrase that already appears in the data appears behind the data, the phrase that appears frequently is replaced by a reference to the appearance position of the previous phrase. Many algorithms have been proposed that achieve high compression ratios by combining with the Huffman coding and the like. In the example of image compression, lossless compression is used in the png format or the gif format (Patent Document 1).

An example of irreversible image compression is the jpeg format. This achieves a high compression rate by separating an image into frequency components using discrete cosine transform and removing high frequency components that cannot be recognized by humans.
The image data is used not only for browsing purposes but also for extracting image features by image analysis processing. In many cases, an enormous number of image data is used for analysis processing, and image compression is an important technique for reducing the data storage area and transfer time.

Introduction to document data compression algorithm-Huffman code, arithmetic code, LZ code, etc. realized in C (IF Essence Series), Tomohiko Uematsu, CQ Publishing (1994/10)

JP-A-8-237138

However, in lossless compression, compression is performed using data features such as frequent occurrence of specific symbols and phrases, so the compression rate when compressing image data that contains components with strong randomness such as noise Has the problem of becoming lower.
In the case of irreversible compression, there is a problem that features important for image analysis processing may be removed. For example, two pixels having different brightness may have the same brightness after compression, or the brightness magnitude relationship may be reversed. Since lossy compression is mainly performed by removing components and features that cannot be recognized by humans, there is a problem that the removed features and components are likely to affect the mechanical analysis processing.

  An object of the present invention is to provide an image processing apparatus and an image processing method capable of compressing image data including noise at a high compression rate without losing image characteristics after compression.

In order to solve the above-described problem, an image processing apparatus according to the present invention uses, as a divisor, an input by a user or a preset integer of 2 or more as a divisor for an input image, and the pixel value of each pixel of the image is a divisor. And a component separating means for calculating a quotient and a remainder when dividing by quotient and separating the quotient component data consisting of the quotient pixel value and the remainder component data consisting of the remainder pixel value, and each component separated by the component separating means And a data compression means for outputting compressed data of each component.
The divisor having the highest compression ratio is further searched for each input image.
Further, the compressed data of each component is restored, the restored quotient component is multiplied by the divisor for each pixel, the restored remainder component is added to the multiplied value to restore the image before compression, and the restored image is It further comprises compressed data restoration means for outputting.
Further, the data compression means discards the remainder component and outputs compressed data of only the quotient component.

The image processing method according to the present invention uses a quotient obtained by dividing the pixel value of each pixel of the image by the divisor, using the input by the user or a preset integer of 2 or more as the divisor for the input image. Calculating a remainder, separating the quotient component data consisting of the quotient pixel value and the remainder component data consisting of the remainder pixel value, compressing each of the separated components, and outputting the compressed data of each component And a step of performing.
Further, the method further includes a step of searching for the divisor having the highest compression rate for each input image.
In addition, the compressed data of each component is restored, the restored quotient component is multiplied by the divisor for each pixel, and the restored remainder component is added to the multiplied value to restore the pre-compression image. The method further includes the step of outputting an image.
The data compressing means includes a step of discarding the remainder component and outputting compressed data of only the quotient component.

According to the present invention, in the case of an image with many noise components such that a pattern with high randomness appears in the remainder component, the quotient component and the remainder component are separated and compressed respectively, so that the compression ratio of the entire image is high. Become. When reversible compression is used for compression of each component, the entire process is also reversible, so that the influence on the image analysis process can be reduced.
When the remainder component is discarded, lossy compression is performed. However, if an appropriate algorithm is selected for compression of the quotient component, the magnitude relationship between the pixel values is not reversed, so that the influence on the image analysis processing can be reduced.

It is a system configuration figure showing an embodiment of the present invention. It is a figure showing the data format in embodiment of FIG. It is a flowchart figure which shows the whole process of embodiment of FIG. It is a flowchart figure which shows the separation compression process of embodiment of FIG. It is a flowchart figure which shows the remainder component separation process of embodiment of FIG. It is a figure which shows the image restoration process of embodiment of FIG.

FIG. 1 is a system configuration diagram showing an embodiment of the present invention.
In FIG. 1, a processing apparatus 101 executes a program (102) including a separation / compression processing unit (103) and an image restoration processing unit (104). In addition, a storage area for storing work areas and input / output data necessary for program execution and a network environment for transmitting / receiving input / output data are provided as necessary.
When the image data 105 is input via the storage device or the network, the program 102 processes the image data 105 by the separation / compression processing unit 103, and stores the component separation data 106 in the storage device or outputs it via the network. .
When the component separation data 107 is input via the storage device or the network, it is processed by the image restoration processing unit 104, and the component separation data 108 is stored in the storage device or output via the network.

FIG. 2 is a diagram illustrating a data format used in the present embodiment.
The data format includes image data handled in the present embodiment and component separation data when processing is performed on the image data.
First, image data will be described.
The image data includes an image width (201), height (202), data amount per pixel (203), other header information (204), and pixel value array (205).
The data amount per pixel (203) is the number of bits necessary to represent the color information per pixel. By multiplying this by the width (201) and the height (202), the size of the pixel value array (205) can be calculated. In this embodiment, a monochrome image is a target. However, a color image can be easily expanded by, for example, processing separately for each component of red, green, and blue.
Other header information (204) stores various information related to the image, such as the creation date and time of image data, the direction when the pixel value array (205) is expanded as an image, and the color format of pixels such as RGB and CMY. It is.
The image data 201 to 204 may be collectively referred to as image header information.

Next, component separation data will be described.
The component separation data is data obtained by dividing the quotient and the remainder when the pixel value is divided by a specified divisor into separate arrays, and the image width (206), height (207), and data amount per pixel (208 ), Other header information (209), divisor (210), quotient component compression method ID (211), quotient component array size (212), quotient component array (213), residue component compression method ID (214), residue component array It consists of a size (215) and a remainder component array (216).
206 to 209 are the same as the header information 201 to 204 of the image data.
When image data is set as component separation data, the header information 206 to 209 is copied as it is from the header information 201 to 204 of the image.
The divisor (210) is an integer value used when the pixel value of the image is separated into a quotient component and a remainder component. When component separation data is expressed as C, the divisor may be expressed as CM.
When “1” is set as the divisor, all pixel value data is stored only in the quotient component, and the remainder component has no meaning.
The quotient component compression method ID (211) is an ID for specifying the compression method used when compressing the quotient component array (213).
A compression method that has already been proposed as lossless compression or lossy compression may be used. Also, 'raw' can be used as a method ID that leaves the pixel value as it is without compression.
It is assumed that a data restoration method corresponding to the available compression method is also prepared.
The quotient component array size (212) stores the array size of the quotient component array (213).
The quotient component array (213) stores a data string obtained by compressing the pixel value array by the compression method designated by the pixel value of the image or the quotient component compression method ID (211). The quotient component array of the component separation data C may be expressed as CP.
The remainder component compression method ID (214), the remainder component array size (215), and the remainder component array (216) are obtained by replacing the information 211 to 213 for the quotient component with the remainder component, respectively.
The remainder component array of the component separation data C may be expressed as CQ.
In addition, when executing the compression process and the decompression process on the quotient component array (213) and the remainder component array (216) and replacing the array data, the corresponding array size (212, 215) is also corrected at the same time. Subsequent data shall be automatically moved to the next position in the array.

FIG. 3 is a flowchart showing the overall processing of this embodiment.
First, if image data to be compressed is input, it is checked whether the image data is image data or component separation data (step 301). If it is image data, compression processing is performed in the procedure from step 302 onward, and if it is component separation data, image restoration processing is performed (step 317). Details of the image restoration processing will be described with reference to FIG.
Hereinafter, compression processing when input data is an image will be described.
First, separation data C to be output as a result is prepared (step 302).
Next, image I data is set in C (step 303). Here, the header information of the image I is copied as it is, and the pixel value array is stored as a C quotient component.

Next, it is checked whether to search for a divisor that minimizes the compression size (step 304). When searching, proceed to the following.
First, “1” is assigned to the divisor CM of C. A divisor of 1 means that the data is not separated into a quotient component and a remainder component.
Next, the quotient component CP is compressed (step 306). The compression method used at this time may be a known one. The compression method to be used may be determined in advance, or an ID for identifying the compression method may be input for each image and the compression method may be automatically selected.
After the array is compressed, the C quotient component compression method ID is set, and the CP array size is updated simultaneously.

Next, the separation data T is prepared (step 307).
Next, 2 is substituted into the divisor TM (308). The initial value can be arbitrarily set as long as it is an integer of 2 or more.
Next, it is checked whether TM is less than a preset upper limit value (step 309). Although the upper limit of the divisor can be arbitrarily set, in this embodiment, the upper limit is half of the maximum value of the pixel value calculated from the data amount per pixel (203) specified by the image data.
If TM is less than the upper limit, first, an image I is set in T (step 310). The setting method is the same as in step 303.
Next, a separation compression process is executed for T (step 311). Details of the separation compression processing will be described with reference to FIG.

Next, the compressed sizes of the separated data C and T are compared (step 312). If T is smaller, T is substituted for C (step 313).
Next, “1” is added to TM (step 314), and the process returns to step 309. If TM is greater than or equal to the upper limit in step 309, the process ends.
Thus, the divisor CM that minimizes the compression size is obtained.
If the divisor search is not performed in step 304, first, the divisor CM is initialized (step 315). The divisor may be a predetermined value or may be input for each image.
Next, a separation / compression process is executed for C (step 316), and the process ends.
As described above, by obtaining a divisor that minimizes the compression size for each input image and performing compression processing, it is possible to compress image data at a high compression rate even when the image to be compressed has a lot of noise components. Thus, the influence on the image analysis processing can be reduced.

FIG. 4 is a flowchart showing the separation and compression processing in the present embodiment.
First, a remainder component separation process is performed on the input separation data C (step 401). Details of this processing will be described with reference to FIG.
Next, the C quotient component array CP is compressed (step 402). The compression method used at this time may be a known one. The compression method to be used may be determined in advance, or an ID for identifying the compression method may be input for each image, and the compression method may be selected based on the ID.
After the array is compressed, the C quotient component compression method ID is set, and the CP array size is updated simultaneously.
Next, it is checked whether the remainder component is discarded (step 403). This may be determined in advance or may be input separately for each image.
If step 403 is Yes, the remainder component CQ is discarded (step 404). At the same time, “0” is assigned to the remainder component array size.
If step 403 is No, the remainder component CQ is compressed (step 405). The compression method used at this time may be a known one as in the case of the quotient component, the compression method to be used may be predetermined, or an ID for identifying the compression method may be input for each image. Good.
After the array is compressed, the C remainder component compression method ID is set, and the CQ array size is simultaneously updated.
Here, when the remainder component CQ is discarded, lossy compression is performed. However, if an appropriate algorithm is selected for compression of the quotient component, the magnitude relationship between pixel values is not reversed, so the influence on the image analysis processing is reduced. be able to.

FIG. 5 is a flowchart of the remainder component separation process in the present embodiment.
Separation data C is input to this process, the original image data is stored in the quotient component, and a divisor necessary for performing the separation process is set.
First, an array pq is prepared (step 501). p is an area for storing data that becomes a quotient component after processing, and q is an area for storing data that becomes a remainder component. For p and q, only the same size as the C quotient component array is secured.
Next, “0” is substituted into the variable i (step 502). Next, while i is less than the number of pixels, the following is executed (step 503). First, the remainder obtained by dividing CP [i] by the divisor CM is substituted into q [i] (step 504). Next, the quotient obtained by dividing CP [i] by the divisor CM is substituted into p [i] (505).
Next, “1” is added to i (step 606), and the process returns to step 503.
In the case of No in step 503, first, the array p is substituted into the quotient component CP (step 507).
Next, the array q is set as the remainder component CQ (step 508), and the process is terminated. At this time, the C remainder component array size is simultaneously updated to the size of the array q.

FIG. 6 is a flowchart of the image restoration process in the present embodiment.
First, the quotient component array CP of the input separated data C is restored by the restoration method corresponding to the method designated by the compression method ID (step 601). If the size of the remainder component CQ is “0”, nothing is done.
Next, it is checked whether the divisor CM is larger than 1 and the size of the remainder component CQ is larger than “0” (602).
If Yes, the remainder component CQ is first restored by the restoration method for the method specified by the compression method ID (step 603). Next, “0” is substituted into the variable i (step 604). Then, while i is less than the number of pixels, the following is executed (step 605).
First, the original pixel value is calculated from the quotient component, remainder component, and divisor at the position i, and the quotient component is updated (step 606). Pixel value is calculated as CP [i] × CM + CQ [i]. Here, CP represents a quotient component array, CQ represents a remainder component array, and CM represents a divisor. Here, when the size of CQ is “0”, CQ [i] is always regarded as “0”.
Next, “1” is added to i (step 607), and the process returns to step 605.
On the other hand, if No in step 602 (when there is no remainder component) or No in step 605, a restored image I is generated from the C header information (206 to 209) and the quotient component CP (step 608), the process is terminated.

DESCRIPTION OF SYMBOLS 101 ... Processing apparatus 102 ... Program 103 ... Separation compression processing part 104 ... Image restoration processing part 105 ... Input image data 106 ... Output separation compression data 107 ... Input separation compression data 108 ... Output image data

Claims (8)

  For the input image, the quotient and the remainder are calculated by dividing the pixel value of each pixel of the image by the divisor using the input by the user or a preset integer of 2 or more as the divisor, and calculating the pixel value of the quotient Component separating means for separating the quotient component data consisting of and residual component data consisting of the remainder pixel values, and data compression for compressing each component separated by the component separating means and outputting compressed data for each component And an image processing apparatus.   The image processing apparatus according to claim 1, further comprising means for searching for the divisor having the highest compression rate for each input image.   The compressed data of each component is restored, the restored quotient component is multiplied by the divisor for each pixel, the restored remainder component is added to the multiplied value, the image before compression is restored, and the restored image is output The image processing apparatus according to claim 1, further comprising a compressed data decompression unit.   The image processing apparatus according to claim 1, wherein the data compression unit discards the remainder component and outputs compressed data including only a quotient component.   For the input image, the quotient and the remainder are calculated by dividing the pixel value of each pixel of the image by the divisor using the input by the user or a preset integer of 2 or more as the divisor, and calculating the pixel value of the quotient Separating the quotient component data and the remainder component data comprising the pixel value of the remainder, and compressing each of the separated components and outputting the compressed data of the respective components. Image processing method.   The image processing method according to claim 5, further comprising a step of searching for the divisor having the highest compression rate for each input image.   The compressed data of each component is restored, the restored quotient component is multiplied by the divisor for each pixel, the restored remainder component is added to the multiplied value to restore the image before compression, and the restored image is The image processing method according to claim 5, further comprising an output step.   The image processing method according to claim 5, wherein the data compression unit includes a step of discarding the remainder component and outputting compressed data of only a quotient component.

Images