JP2000067240A - Area dividing method for color image, and program storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten processing time and to precisely area-divided a color an image regardlessly of how much the color of the image is changed. SOLUTION: Based on a color difference between the pixels of a color image, the width of division in each axial direction of hue, lightness and saturation(HLS) is determined by a color separating number preparing means 2a of an area dividing part 2, the color space of HLS is divided based on that determined division width, and a color separating number is set for each color space dividing area of that divided HLS color space. Based on the said color separating number, each pixel of the color image is labeled by an image area dividing means 2b of the area dividing part 2, the area of continuous pixels with the same color separating number in each labeled pixel is discriminated as the same color area and the color image is divided for each of the same color area.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color image area dividing method and a program storage medium used in shape extraction such as image retrieval.

[0002]

2. Description of the Related Art Conventionally, in a color image search, a region division process has been performed as a pre-process for extracting an object shape of an image as a search condition. However, a color image captured from a digital camera or a scanner may have a single color at first glance, but may not be a single color at a pixel value level, or may have a similar color of an object to the background. In such a color image region dividing process, in order to accurately remove the background with a simple fixed color classification or to divide the image into a plurality of regions, the color is finely classified and the region is subdivided. Thereafter, it is necessary to re-integrate the areas segmented by some kind of determination calculation, which causes a problem that the processing time becomes longer.

It is therefore an object of the present invention to provide a color image area dividing method and a program storage medium capable of shortening the processing time and accurately dividing an area regardless of the magnitude of color change of the image.

[0004]

According to a first aspect of the present invention, there is provided a color image area dividing method, comprising the steps of obtaining a color difference between pixels of a color image, and a color space based on the color difference between the pixels. Determining a division width in each axis direction, setting a color classification number for each color space division region of the color space divided according to the division width in each axis direction, and for each color space division region Step of labeling each pixel of the color image based on the set color classification number, and in each of the labeled pixels,
Determining an area in which pixels of the same color classification number are continuous as the same color area, and dividing the color image for each of the same color areas.

According to the color image area dividing method of the first aspect, the color difference between the pixels of the color image is obtained, and the division width in each axial direction of the color space is determined based on the obtained color difference between the pixels. . Next, a color classification number is set for each color space division area of the color space divided by the division width in each axis direction.
Each pixel of the color image is labeled based on the color classification number thus set, a region where pixels of the same color classification number are continuous is determined to be the same color region, and the color image is divided for each of the same color regions. I do. As described above, since the color image is divided for each color region corresponding to the color space divided region of the color space, it is necessary to classify the color finely and subdivide the region, or to integrate the subdivided region again. There is no. Therefore, the number of times of area division processing is reduced, and it is not necessary to perform area integration, so that processing time can be reduced. Further, the division width in each axis direction of the color space is determined based on the color difference between the pixels of the color image, and the division width in each axis direction of the color space becomes large according to the magnitude of the color change of the color image. By doing so, the region can be accurately divided regardless of the magnitude of the color change of the color image.

According to a second aspect of the present invention, there is provided the color image area dividing method according to the first aspect.
The color space is a color space represented by an HLS value, and in the step of determining a division width in each axis direction of the color space, an average value of a hue difference, a lightness difference, and a saturation difference between the pixels is calculated. On the basis of this, the width of the color space represented by the HLS value in each axial direction is determined.

According to the color image area dividing method of the present invention, the color space represented by the HLS value is determined based on the average value of the hue difference, lightness difference and saturation difference between the pixels of the color image. By determining the division width in each axial direction, the division width in each axial direction of the color space can be accurately determined according to the color change of the entire color image.

According to a third aspect of the present invention, there is provided the color image area dividing method according to the first aspect.
The color space is a color space represented by an HLS value. In the step of determining a division width in each axis direction of the color space, the HLS is determined based on an average value and a standard deviation value of hue differences between pixels. The division width in the hue axis direction of the color space represented by the value is determined, and the division width in the lightness axis direction of the color space represented by the HLS value is determined based on the average value and the standard deviation value of the lightness difference between the pixels. Is determined, and the division width of the color space represented by the HLS value in the direction of the saturation axis is determined based on the average value and the standard deviation value of the saturation differences between the pixels.

According to the third aspect of the present invention, even if a color image has substantially the same average value of the hue difference, lightness difference, and chroma difference between the pixels, the hue, lightness, and color difference are obtained. The degree of variation of the degree may be different. In such a case, the hue, lightness, and saturation differences are added by adding the standard deviation values of the hue difference, lightness difference, and saturation difference between the pixels to the average value of the hue difference, lightness difference, and saturation difference between the pixels. Is large, the width of division in each axis direction of the color space can be increased to more accurately capture the color change of the entire color image. In this case, a value obtained by simply adding the standard deviation value to each average value of the hue difference, lightness difference, and chroma difference between pixels may be used as the division width in each axis direction, or the hue difference, lightness difference, and chroma difference may be used. A value obtained by adding a value obtained by multiplying a different weighting coefficient for each standard deviation value to each average value may be used as the division width in each axis direction.

The program storage medium according to claim 4 is
Obtaining a color difference between the pixels of the color image; determining a division width in each axis direction of the color space based on the color difference between the pixels; and Setting a color classification number for each color space division area; labeling each pixel of the color image based on the color classification number set for each color space division area; and setting each of the labeled pixels , An information processing program having a step of determining an area in which pixels of the same color classification number are continuous as the same color area and dividing the color image for each of the same color areas is stored.

According to the fourth aspect of the present invention, a color difference between pixels of a color image is obtained, and a division width in each axial direction of a color space is determined based on the obtained color difference between pixels. Next, a color classification number is set for each color space division area of the color space divided by the division width in each axis direction. Each pixel of the color image is labeled based on the color classification number thus set, a region where pixels of the same color classification number are continuous is determined to be the same color region, and the color image is divided for each of the same color regions. I do. As described above, since the color image is divided for each color region corresponding to the color space divided region of the color space, it is necessary to classify the color finely and subdivide the region, or to integrate the subdivided region again. There is no. Therefore,
The number of times of region division processing is reduced, and it is not necessary to perform region integration, so that processing time can be reduced. Further, the division width in each axis direction of the color space is determined based on the color difference between the pixels of the color image, and the division width in each axis direction of the color space becomes large according to the magnitude of the color change of the color image. By doing so, the region can be accurately divided regardless of the magnitude of the color change of the color image.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A color image area dividing method and a program storage medium according to the present invention will be described below in detail with reference to the illustrated embodiments.

FIG. 1 is a functional block diagram of a feature region extracting apparatus using a color image region dividing method according to an embodiment of the present invention. Reference numeral 1 denotes an input unit for inputting image data and various instructions. Reference numeral 3 denotes an area dividing unit for dividing an area of a color image, and reference numeral 3 denotes a feature amount extracting unit for extracting a feature amount of the image data divided by the area dividing unit 2. The area dividing section 2 has a color classification number creating means 2a and an image area dividing means 2b.

FIG. 2 is a diagram showing a hardware configuration for realizing the functions of FIG. As shown in FIG. 2, the characteristic region extraction device includes a control device 11 for controlling the operation of the entire device, and a CRT (cathode, ray, and so on) for displaying images, characters, and the like, and for performing various displays for operations and the like. Tube) 12, a keyboard 13 and a mouse 14 for performing various input operations, instruction operations, and the like, and a floppy disk 1 built in the control device 11 and serving as a data storage medium.
Floppy disk drive 15b for recording and reproducing 5a
A hard disk device 16, a printer 17 for outputting image data or a drawing created by image organization, a scanner 18 for capturing image data,
CD-ROM drive 19b which is built in the control device 11 and reproduces a CD (compact disk) 19a
And a speaker 20 for voice output and a microphone 21 for voice input. The CRT 12, keyboard 13, mouse 14, hard disk drive 16, printer 17, scanner 18, speaker 20
And the microphone 21 are connected to the control device 11.

FIG. 3 shows the C mounted on the control device 11 of FIG.
FIG. 2 is a block diagram mainly showing a PU (central processing unit) 25.

As shown in FIG. 3, the CPU 25 includes:
For example, a product number i80486DX manufactured by Intel Corporation is used. Then, the CPU 25 is provided with a ROM (read only memory) in which a program for controlling the entire apparatus is stored.
A memory) 28, a RAM (random access memory) 29 in which various data and programs are stored,
A display control circuit 30 for displaying images or characters on the CRT 12; a keyboard control circuit 31 for controlling the input from the keyboard 13; a mouse control circuit 32 for controlling the transfer of the input from the mouse 14; A floppy disk drive control circuit 33 for controlling the drive 15b, a hard disk control circuit 34 for controlling the hard disk drive 16, and a printer control circuit 35 for controlling output to the printer 17
And a scanner control circuit 36 for controlling the scanner 18
And a CD-ROM for controlling the CD-ROM drive 19b.
A ROM drive control circuit 37; a speaker control circuit 38 for controlling the speaker 20;
1 is connected to the microphone control circuit 39 via the data bus 26.

A clock 2 for generating a reference clock necessary for operating the entire apparatus is provided to the CPU 25.
7, and an expansion slot 40 for connecting various expansion boards via the data bus 26. Note that an SCSII board may be connected to the expansion slot 40 to connect the floppy disk drive 15b, the hard disk device 16, the scanner 18, the CD-ROM drive 19b, and the like.

In the characteristic region extracting device, the input unit 1 is constituted by the keyboard 13, the mouse 14, the scanner 18, and the CD-ROM drive 19b.
Constitute the region dividing unit 2 and the feature amount extracting unit 3 shown in FIG.

The characteristic region extracting apparatus having the above configuration operates according to the flowcharts of FIGS. 4, 5, 6, 7, and 8.

FIG. 4 is a flowchart showing a main routine of a program of the above-mentioned feature region extracting apparatus. In FIG. 4, first, when the program is started, in step S1, initialization of flags necessary for processing described later, initial screen display, and the like are performed. Next, the process proceeds to step S2, where it is determined whether or not “region division” of the job menu has been selected from the initial screen displayed on the CRT 12, and if it is determined that “region division” has been selected, the process proceeds to step S3. The region dividing unit 2 performs a region dividing process, and proceeds to step S9. On the other hand, if it is determined in step S2 that “region division” has not been selected,
Proceed to step S4. Next, in step S4, it is determined whether or not "extract feature" is selected. If it is determined that "extract feature" is selected, the process proceeds to step S5, where the feature extraction unit 3 executes the feature extraction process. Go to step S9. On the other hand, if it is determined in step S4 that "extract feature" has not been selected, the process proceeds to step S6. Next, it is determined whether or not “other menu” is selected in step S6. If it is determined that “other menu” is selected, the process proceeds to step S7, where other menu processing is performed, and the process proceeds to step S9. move on. On the other hand, if it is determined that "other menu" has not been selected, the process proceeds to step S8. Then, it is determined whether or not “end” has been selected in step S8, and if it is determined that “end” has been selected, this processing ends. On the other hand, if it is determined in step S8 that "end" has not been selected, the process proceeds to step S9. After performing other processes in step S9, the process returns to step S2.

Hereinafter, the area dividing process in step S3 and the feature amount extracting process in step S5 will be described.
Note that other steps are not directly related to the present invention, and thus detailed description is omitted.

[Area Division Processing] FIG. 5 is a flowchart showing an area division processing subroutine of step S3 in the main routine. First, in step S31, image data represented by RGB values for performing a region division process is read. Next, the process proceeds to step S32, where color space division is performed based on the color information of the image data, and a color classification number is created. Next, the process proceeds to step S33, in which each pixel value of the image data is labeled for each classified color, and the image is divided into regions. Then, the process proceeds to step S34, where the area division data representing the data of each of the divided areas is stored, and then the processing returns to the higher-order processing, that is, the main routine shown in FIG. The area division data stored here is used in a feature amount extraction process described later.

FIG. 6 shows a color classification number creating process (step S in FIG. 5) by the color classification number creating means 2a of the area dividing section 2.
It is a flowchart which shows 32). In step S321, the hue value (H), lightness value (L), and chroma value (S) of each pixel of the image data represented by the RGB values are obtained, and the color difference (hue difference, lightness difference, (Saturation difference). Next, the process proceeds to step S322, where a division width in each axis direction of the HLS color space is calculated based on the color difference obtained in step S321 in order to classify the color characteristics of the image in the HLS color space. Then, the process proceeds to step S323, in which HL is determined by the obtained division width.
After dividing the S color space and assigning a color classification number to the color space divided area of the divided HLS color space, the process returns to the higher-level processing.

In the above steps S321 and S322, the following methods (1) to (5) and the like are used to determine the division width in each axis direction of the HLS color space from the color difference between adjacent pixels. (1) The hue difference, lightness difference, and chroma difference between the right pixel and the lower pixel are obtained for each pixel, and the average value of the hue difference, lightness difference, and chroma difference for all pixels is obtained, and the average value is obtained. Is a division width in each axis direction of the HLS color space.

(Equation 1) (2) The hue difference, lightness difference, and chroma difference between the right pixel and the lower pixel are obtained for each pixel, and the average value of the hue difference, the lightness difference, and the chroma difference for all pixels and the hue difference, the lightness difference, A method in which each standard deviation value of the saturation difference is obtained, and a value obtained by adding the standard deviation value to each average value is used as a division width in each axis direction of the HLS color space. (3) In the above (2), a value calculated on a different basis for each axial direction of the HLS color space in consideration of the properties of hue, lightness and saturation (for example, a standard deviation value is multiplied by a different weighting coefficient for each axial direction). Is added to the average value) as a division width in each axis direction of the HLS color space. (4) In the above (1) to (3), not only the hue difference, the brightness difference and the saturation difference from the right pixel and the lower pixel are obtained for each pixel, but also the upper right pixel and the lower right pixel for each pixel. And calculating a hue difference, a lightness difference, and a chroma difference between all pixels, based on the hue difference, the lightness difference, and the chroma difference in all the pixels, as a division width in each axis direction of the HLS color space. (5) In the above (1) to (4), the hue difference, lightness difference and saturation difference are obtained not only between adjacent pixels but also for pixels within a certain designated distance, and the hue difference and lightness difference in all pixels are obtained. And a method in which a result calculated based on the saturation difference is set as a division width in each axis direction of the HLS color space.

FIG. 7 is a flowchart showing the image area dividing process in step S33 in FIG. 5 by the image area dividing means 2b of the area dividing section 2. First, in step S331, each pixel is labeled with a color classification number according to the color of each pixel. Next, the process proceeds to step S332, in which a pixel group having the same color classification number continuously is determined as one region, and the process returns to a higher-level process.

[Feature Extraction Processing] FIG. 8 is a flowchart showing the feature extraction processing of step S5 in the main routine of FIG. First, in step S51, image data to be subjected to feature amount extraction processing is read. Next, the process proceeds to step S52, where the area division data corresponding to the image data is read. Next, the process proceeds to step S53, and the feature amount of the image is extracted from the image data and the area division data. Then, the process proceeds to step S54, where the extracted feature amount is stored, and then the process returns to a higher-level process.

The feature quantity extracted from the image data and the area division data in step S53 is, for example,
There are the following (i) and (ii). (i) Color information and texture information of the entire image (ii) Size, color information, contour shape, and texture information of the divided area The feature amounts obtained in this manner are used for image recognition, image retrieval, and the like. You.

As described above, since a color image is divided into the same color regions corresponding to the color space divided regions of the divided HLS color space, the regions are subdivided by finely classifying colors.
There ’s no need to reintegrate the subdivided areas,
The processing time for area division can be reduced. Also, the color difference between pixels of a color image (hue difference, lightness difference and saturation difference)
, The division width of the HLS color space is determined, and the division width of the color space is made large or small according to the size of the color change of the color image. Thus, regardless of the size of the color change of the color image, Region division can be performed with high accuracy.

Further, the division width in each axis direction of the HLS color space is determined based on each average value of the hue difference, lightness difference and saturation difference between the pixels of the color image, so that the H
The color change (hue difference, lightness difference, and saturation difference) in the LS color space is used to accurately capture the color change of the entire color image, and the division width in each axis direction of the color space according to the color change of the color image is determined. be able to.

Further, based on the average value and standard deviation value of the hue difference, lightness difference and saturation difference between the pixels of the color image, H
By determining the division width in each axis direction of the LS color space, when the variation in hue, lightness and saturation is large,
By increasing the division width in each axis direction of the HLS color space, the color change of the entire color image can be more accurately captured.

In the above embodiment, the image data storage medium is constituted by the floppy disk 15a and the hard disk device 16, but may be another information storage device such as a magneto-optical disk device. In addition, although the scanner 18 and the CD-ROM drive 19b are used as the input unit 1, other input devices such as a still camera and a digital camera may be used. Although the printer 17 is used as an output device, another output device such as a digital copying machine may be used.

In the above embodiment, the division width of each axis direction of the HLS color space is determined, and the HLS color space is divided based on the division width. However, the color space is not limited to this, and L * a * b
* Needless to say, a color space (JIS Z8729- (1980)) or the like may be used.

In the above embodiment, the control device 11
Although the characteristic region extracting device in which a program for controlling the whole is stored in the ROM 28 has been described, a part or all of the program according to the present invention is stored in a program storage medium such as a floppy disk, and the The program may be read into an information processing device such as a personal computer and executed.

[0034]

As is apparent from the above description, the color image area dividing method according to the first aspect of the present invention provides a method for dividing a color image in accordance with a division width in each axial direction of a color space determined based on a color difference between pixels of the color image. After dividing the color space, each pixel of the color image is labeled based on the color classification number set for each color space division region of the divided color space, and the same color classification is performed in each of the labeled pixels. A color image is divided for each of the same color regions, with the region in which the pixels of the numbers are consecutive being the same color region.

Therefore, according to the color image area dividing method of the first aspect of the present invention, the color image is divided into the same color areas corresponding to the color space divided areas in the color space, and the areas are classified by finely classifying the colors. Since there is no need to subdivide or re-integrate the subdivided regions, it is possible to reduce the processing time of the region division, and based on the color difference between the pixels of the color image, in each axial direction of the color space. By determining the division width, regardless of the magnitude of the color change of the color image,
Region division can be performed with high accuracy.

According to a second aspect of the present invention, there is provided the color image area dividing method according to the first aspect, wherein the color image area dividing method is based on the average value of the hue difference, lightness difference, and chroma difference between the pixels. Thus, since the division width in each axis direction of the HLS color space is determined, the division width in each axis direction of the color space can be determined according to the color change of the entire color image.

According to a third aspect of the present invention, there is provided the color image area dividing method according to the first aspect, wherein the HLS color space is determined based on an average value and a standard deviation value of the hue differences between the pixels. Is determined in the hue axis direction, and the HLS is determined based on the average value and the standard deviation value of the brightness difference between the pixels.
The division width of the color space in the lightness axis direction is determined, and the division width of the HLS color space in the saturation axis direction is determined based on the average value and the standard deviation value of the saturation differences between the pixels. Even when the average values of the phase difference, the lightness difference, and the saturation difference are different in the hue, the lightness, and the saturation in the same color image, the color change of the entire color image can be more accurately captured.

Further, the program storage medium of the invention according to claim 4 includes a step of obtaining a color difference between pixels of a color image,
Determining a division width in each axis direction of the color space based on the color difference between the pixels; and dividing the color space according to the division width in each axis direction, and performing color classification for each of the divided color space division regions. Setting a number, labeling each pixel of the color image based on the color classification number set for each of the color space divided regions, and, in each of the labeled pixels, pixels having the same color classification number Is determined as the same color region, and a color image is divided for each of the same color regions.

Therefore, the information processing program stored in the program storage medium according to the fourth aspect of the present invention is read into an information processing apparatus such as a personal computer and executed, whereby the same color corresponding to the color space divided area of the color space is obtained. Since the color image is divided for each area, it is not necessary to subdivide the area by classifying the colors finely, and it is not necessary to integrate the subdivided areas again. By determining the division width in each axis direction of the color space based on the color difference between the pixels of the image, it is possible to accurately perform the region division regardless of the magnitude of the color change of the color image.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a feature region extracting apparatus using a color image region dividing method according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a hardware configuration for realizing the functions of FIG. 1;

FIG. 3 shows a CP mounted on the control device shown in FIG. 2;
It is a block diagram centering on U.

FIG. 4 is a flowchart illustrating a basic processing operation of a CPU in FIG. 3;

FIG. 5 is a flowchart showing an area dividing process in FIG. 4;

FIG. 6 is a flowchart showing a color classification number creating process of FIG.

FIG. 7 is a flowchart showing an image area dividing process of FIG. 5;

FIG. 8 is a flowchart showing a feature amount extraction process of FIG. 4;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... input part, 2 ... area division part, 2a ... color classification number preparation means, 2b ... image area division means, 3 ... feature-value extraction part.

Claims (4)

[Claims] A step of determining a color difference between pixels of a color image; a step of determining a division width in each axis direction of a color space based on the color difference between the pixels; Setting a color classification number for each color space divided region of the color space; labeling each pixel of the color image based on the color classification number set for each color space region; Determining a region in which pixels of the same color classification number are continuous as the same color region, and dividing the color image for each of the same color regions. . 2. The color image area dividing method according to claim 1, wherein the color space is a color space represented by an HLS value, and the step of determining a division width in each axis direction of the color space. Determining a division width in each axial direction of the color space represented by the HLS value based on each average value of the hue difference, the lightness difference, and the saturation difference between the pixels. Split method. 3. The color image area dividing method according to claim 1, wherein the color space is a color space represented by an HLS value, and the step of determining a division width of each axis direction of the color space. Determining a division width in the hue axis direction of the color space represented by the HLS value based on the average value and the standard deviation value of the hue differences between the pixels, and calculating the average value and the standard deviation value of the lightness differences between the pixels; Is determined in the lightness axis direction of the color space represented by the HLS value based on the HLS value, and based on the average value and the standard deviation value of the saturation difference between the pixels, the color space represented by the HLS value is determined. A method for dividing an area of a color image, comprising determining a division width in a direction of a saturation axis. 4. A step of obtaining a color difference between pixels of a color image; a step of determining a division width in each axis direction of a color space based on the color differences between pixels; Setting a color classification number for each color space divided area of the color space; and labeling each pixel of the color image based on the color classification number set for each color space divided area; In each of the labeled pixels, an area in which pixels of the same color classification number are consecutive is determined to be the same color area, and the color image is divided for each of the same color areas. Program storage medium.