JPH11259636A - Image conversion method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the new image converting method which is suitable for the data compression of an artificial image such as a CG image. SOLUTION: To each area where pixels having the same attribute value in an image are connected, 2-bit codes (00, 01, 10, and 11) which do not overlap with other areas are assigned and a pair of 2-bit coded data representing the pixels in each area with the assigned two-bit codes and data of the attribute values made to correspond to respective areas one to one are generated as conversion data. High compressibility is obtained for a color image of CG, etc., having a small number of areas. Area information of the source image is saved in the 2-bit coded data 203 in directly usable form.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to image processing,
In particular, the present invention relates to image conversion that is effective for data compression or memory compression of an image in which the attribute value of each pixel is represented by 3 bits or more.

[0002]

2. Description of the Related Art Generally, image data has a large data size, especially in a color image. Normally, a color scanner converts an original into 8-bit signals (1 bit) for each of RGB primary color components.
(24 bits per pixel), so if the reading resolution is 400 dpi, the image data of the A4 size document becomes very large at 48 MB, and when processing it, the necessary memory capacity including the work area is required. Is several times the size of 48 MB. When storing or transmitting such image data, it is highly necessary to perform data compression.

[0003] Conventionally, as a method for compressing image data,
Various methods such as JPEG, sub-band coding, and wavelet transform based on discrete cosine transform have been devised and put into practical use. These methods basically use the correlation of pixels to achieve data compression.
Each has its pros and cons. For example, it has been pointed out that JPEG is suitable for natural images, but has a low compression ratio for artificial images (CG images are animation images, etc.). In general, if a high compression ratio is to be obtained, lossy compression is performed, and it is not possible to restore some information of the original image.

[0004]

SUMMARY OF THE INVENTION An object of the present invention is to provide a novel image conversion method for compressing image data from a completely different viewpoint from the conventional image compression method. It is another object of the present invention to provide an image conversion method capable of compressing an artificial image such as a CG image at a high compression ratio without loss. It is another object of the present invention to provide an image conversion method capable of obtaining compressed data in which image area information can be directly used.

[0005]

The gist of the present invention for achieving the above object is to allocate a 2-bit code which does not overlap with other areas adjacent to each area where pixels having the same attribute value on an image are connected. , Generates data in which each pixel in each area is represented by an assigned 2-bit code,
Generate attribute value data corresponding to each area on a one-to-one basis,
To convert an image into a set of 2-bit coded data and attribute value data.

[0006]

FIG. 1 is a flowchart showing a processing flow and related data according to an embodiment of the present invention. First, in step 100, image data 200 to be processed is input. In the next step 101, an area division process is performed on the input image data 200, and the labeling data 201 and the area data 20 are input.
2 is created. In the next step 102, 2-bit coded data 203 is created with reference to the labeling data 201 and the area data 202. Finally, in step 103, the set of attribute value data and 2-bit coded data 203 in the area data 202 is
02 is output as the converted data.

FIG. 2 schematically shows an example of the image data 200. In FIG. 2, each grid corresponds to one pixel, and characters “A” to “F” described in each grid represent pixel attribute values, and the same character means the same attribute value. The attribute value is a value representing a characteristic or property of a pixel, such as density, luminance, and color. As an example, the attribute value of each pixel of the color image is expressed by 8 bits for each RGB primary color component, for a total of 24 bits. Generally, a data compression effect is obtained when the present invention is applied to image data in which the attribute value of each pixel is expressed by 3 bits or more.

In the area dividing process in step 101, a connected (four connected) area of pixels having the same attribute value is extracted from the image data 200, an ID number is assigned to each area, and an ID number is assigned to each area. Labeling data 201 is created in which the same label as the ID number of the same area is written at the pixel position. FIG. 3 shows labeling data 201 for the image data 200 illustrated in FIG. In FIG. 3, each grid has a one-to-one correspondence with each pixel of the image data 200, and the numbers in the grid are labels (= ID numbers). FIG.
As can be understood from the above, regions having the same attribute value are sequentially extracted from the upper left to the lower right of the image data 200, and are assigned ID numbers in the extraction order. Since such a region division (labeling) process may be performed by a known method, a detailed description is omitted. However, the boundary between the regions is determined by four connections.

In step 101, area data 202 is created in parallel with the area dividing process. FIG. 4 shows an example of the area data 202 for the image data 200 of FIG.
Shown in Although this example is shown in a table format, the data format may be arbitrary. In FIG. 4, the ID number of each area is set in the “area ID” field, and the ID numbers of other areas adjacent to each area are set in the “adjacent area ID” field. In the "attribute value" field, the attribute value of the pixel in each area is set. The setting of the “attribute value” field is performed, for example, when a new area is detected in the area division processing. By doing so, the labeling data 201 can be written by directly writing the label on the image data 200.
(In this case, the image data 200 itself is lost when the area division processing is completed, but there is no problem since the attribute value of each area is stored in the labeling data 201). Alternatively, the "attribute value" field may be left blank in the area dividing process, and the "attribute value" field may be set in the 2-bit coding process of step 102. The “2-bit code” field is left blank at the stage of the area division processing, and the 2-bit code assigned to the area in the 2-bit coding processing is set.

In step 102, each area divided by the area dividing process is adjacent (four connected).
2-bit coded data 2 in which a 2-bit code (any one of 00, 01, 10, and 11) that does not overlap with other areas is assigned, and the 2-bit code is written at each pixel position in the area.
03 is created. Regardless of the positional relationship of any region and the number of regions, the "four-color theorem" proved by Haken and Appel allows assignment of 2-bit codes without duplication. When a 2-bit code is assigned to each area, the 2-bit code is assigned to area data 2
02 is set in the corresponding "2-bit code" field. FIG. 5 shows an example of 2-bit coded data 203 created for the image data 200 shown in FIG.

The two-bit coding is performed, for example, in the following procedure. First, a 2-bit code that does not overlap with an adjacent area is assigned to each area with reference to the “adjacent area ID” field of the area data 202 (FIG. 4). For example, by assigning a 2-bit code with priority to an area having a large number of adjacent areas, and when duplication cannot be avoided on the way, the procedure returns to the previous stage and the 2-bit code is assigned again by a trial and error procedure. , All areas are assigned 2-bit codes that do not overlap with adjacent areas. In the case of the image data 200 in FIG. 2, for example, as shown in FIG. 4, a 2-bit code is assigned to each area.

When the 2-bit code is assigned to each area in this manner, 2-bit data 203 is generated. This is performed by the following procedure, for example. The labeling data 201 (FIG. 3) is raster-scanned from the upper left to search for unprocessed labels. An unprocessed label 0 is found at the upper left pixel position A. ID of area data 202 =
The 2-bit code “00” set in the “2-bit code” field of the area 0 is written to the area of the label 0 of the 2-bit coded data 203. If the "attribute value" field of the area data 202 is not set in the area division step 101, the attribute value of the area is read from the image data 200 at the stage of the 2-bit coding, and is read as the "attribute value". Set in the field. The raster scanning of the labeling data 201 is restarted from the pixel position next to the pixel position A, and an unprocessed label is searched. Since an unprocessed label 1 is found at the pixel position B, the I
The 2-bit code “01” set in the “2-bit code” field of the area of D = 1 is written in the area of label 1 of the 2-bit coded data 203. Next, the raster scanning is restarted from the pixel position next to the pixel position B, and the pixel position C
, An unprocessed label 2 is found, so that the 2-bit code “11” is stored in the label 2 area of the 2-bit coded data 203.
Write. Thereafter, by repeating the same procedure, 2-bit coded data 203 shown in FIG. 5 is obtained.

In step 103, in the case of the image data 200 shown in FIG. 2, the 2-bit coded data 203 shown in FIG. 5 and the "attribute value" field of the area data 202 shown in FIG. 4 are set. A set of attribute values arranged in the order of the area ID numbers, that is, a set of attribute value data 204 corresponding to each area on a one-to-one basis is output as conversion data.

Assuming that the number of pixels of the image data is n, the number of bits of the pixel attribute value is B, and the number of regions is N, 2 in the converted data
The total number of bits of the bit coded data is 2 × n, and the total number of bits of the attribute value data is B × N. Therefore, the total number of bits of the converted data is 2 × n + B × N. The total number of bits of the original image data is B × n. In an artificial color image such as a CG image, B = 24 bits, n = 10 to the sixth power, and N <10 = 4
The power is a representative value. In this case, according to the present invention, converted data obtained by compressing image data to about 1/12 is obtained. Moreover, there is no loss. If the number of bits of the attribute value of the original image data is further increased, the data compression ratio will be further increased.

As described above, according to the present invention, lossless compression of an image is possible. Further, the area information of the original image is stored in the 2-bit coded data in the converted data in a form that can be directly used. There is also the advantage that there is. For example, when a color document image or the like is compressed according to the present invention, regions such as character lines of the same color can be directly recognized from the compressed data (without decompressing the original image). Memory area and time can be greatly reduced.

In order to restore the original image data from the converted data according to the present invention, a procedure reverse to the above-described conversion procedure may be used. That is, for example, the 2-bit coded data 203 in FIG. 5 is raster-scanned from the upper left. The two-bit code “00” at the first pixel position A and the same four-connected code are converted into the first attribute value “A” of the attribute value data 204. When the raster scanning is restarted from the pixel position next to the pixel position A and the 2-bit code “01” at the unprocessed pixel position B is detected, the same code and the same code that is connected to the same 4 times in the attribute value data 204 To the attribute value "i". By repeating such a procedure, the same image data as the original image data 200 can be restored.

When a loss is allowed, the present invention may be applied after the attribute value of an image is quantized to a smaller number of gradations or colors to reduce the number of regions. Thus, a high compression ratio can be achieved even for an image whose attribute value changes in a complicated manner. Further, the present invention can be applied to a bit plane image. In other words, even in an image in which the attribute values change in a complicated manner, if the image is divided into bit planes, the value frequently changes in the lower bit plane, and in the upper bit plane, there are many portions where the same value is continuous. A method of compressing several higher-order bit planes together and compressing the lower-order bit planes by another appropriate method is also an effective data compression method.

The image conversion method of the present invention can be easily combined with other conventional image compression methods. For example, the amount of information can be further reduced by applying a conventional appropriate encoding method to the converted image data according to the present invention. The image conversion of the present invention can be applied to conversion data such as discrete cosine conversion data of an image.

The above-described image conversion processing of the present invention can be realized by dedicated hardware. For example, the CPU 300 and the memory 30 as shown in FIG.
1. An input interface 302, an output interface 303, a drive 304 of an information recording medium 305 such as a floppy disk or a CD-ROM, and the like can be realized by software processing on a general computer connected via a bus 306. It is. In this case, the image conversion program 310 for executing the processing procedure of each of the processing steps 100 to 103 shown in FIG. 1 is read, for example, by the drive 304 from the information recording medium 305 on which it is recorded and loaded into the memory 301. Is executed. Image data 200, labeling data 201, area data 202, 2-bit coded data 2
As the storage area 03, a specific area of the memory 301 is used. The image data 200 to be converted is input from an external scanner or the like via the input interface 302, for example. The converted data is output to, for example, an external auxiliary storage device or a printer via an output interface 303, or written to an information recording medium 305 such as a floppy disk via a drive 304.

[0020]

As is apparent from the above description, claim 1
Alternatively, the image conversion method according to the invention described in 2 is based on a completely different viewpoint from the conventional image compression method, enables lossless image data compression, and particularly, a CG image having a relatively small number of regions. For an artificial image such as, lossless compression with a high compression ratio is possible. Further, since the area information of the original image is stored in the 2-bit coded data in the converted data in a form that can be directly used, for example, when a color document image is converted, the original image is directly converted from the converted data. Regions such as character lines of the same color can be recognized (without restoration), and the memory area and processing time for such processing can be significantly reduced. Further, original image data can be restored from the converted data by this image conversion method by a very simple procedure. Also, this image conversion method can be easily applied in combination with a conventional image compression method. According to the information recording medium of the third aspect of the present invention, it is possible to easily execute the above-described image conversion using a general computer.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a processing flow and related data according to an embodiment of the present invention.

FIG. 2 is a diagram schematically illustrating an example of input image data.

FIG. 3 is a diagram illustrating an example of labeling data.

FIG. 4 is a diagram showing an example of area data.

FIG. 5 is a diagram showing an example of 2-bit coded data.

FIG. 6 is a block diagram showing an example of implementing the present invention by software processing using a computer.

[Explanation of symbols]

 Reference Signs List 200 image data 201 labeling data 202 area data 203 2-bit coded data 204 attribute value data 300 CPU 301 memory 302 input interface 303 output interface 304 drive 305 information recording medium 310 image conversion program

Claims (3)

[Claims] 1. A two-bit code that does not overlap with another region adjacent to each region where pixels having the same attribute value on an image are connected, and each pixel in each region is represented by the allocated two-bit code. An image conversion method, comprising: converting an image into a set of data and attribute value data corresponding to each area on a one-to-one basis. 2. A method for converting an image in which the attribute value of each pixel is represented by 3 bits or more, comprising: dividing the image into regions in which pixels having the same attribute value are connected; Generating data represented by a 2-bit code that does not overlap with other adjacent areas, and generating attribute value data corresponding to each of the divided areas on a one-to-one basis. Characteristic image conversion method. 3. A machine-readable information recording medium on which a program for causing a computer to execute the processing of each step of the image conversion method according to claim 2 is recorded.