JPH06103371A - Color image processing method/device - Google Patents

Abstract

PURPOSE:To prevent the generation of the useless areas by integrating a relevant area with its adjacent area in response to a decided fact that the change of the picture quality cannot be visually recognized basing on the size of the area and the color difference between the area and its adjacent area. CONSTITUTION:The color image to be processed is fetched (SP1) and plural representative colors (having higher appearing frequency in a color image) are selected (SP2). The color which is not selected as a key color is replaced with a representative color having the highest approximation (SP3). Then, a partial image area is obtained at every representative color (SP4). Then, it is discriminated whether a partial image area where the change of the picture quality cannot be visually identified is existed or not (SP5). If so, the relevant partial image area where the change of picture quality cannot be visually recognized is integrated with another partial image area (SP6). Then, the preceding discrimination is carried out again. If it is discriminated that the partial image is not existed, each partial image area is approximated with a polygon.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color image processing method and apparatus thereof, and more particularly to a color image processing method and apparatus suitable for color image compression processing.

[0002]

2. Description of the Related Art With the recent development of image processing technology and the widespread use of large-capacity storage media represented by optical discs, research on image database creation and image search using the created image database is under way. . By the way, since color images generally have a very large amount of data,
If the database is created as it is, not only can a sufficient number of color images be stored even if a large-capacity storage medium is used, the time required for data transfer when performing an image search will become extremely long, making it an extremely practical image search. The system cannot be built.

In consideration of these points, color image compression and expansion techniques have been studied in order to significantly reduce the data amount of color images and to obtain a practical image retrieval system. Among them, color image region segmentation technology, which significantly affects the compression rate, has been studied. Further, when applied to an image recognition system, a color image area dividing technique has been studied because it has a great influence on the image recognition accuracy and the like.

The color image area dividing method is a method of dividing a color image into partial images having uniform local features. For example, brightness and color are adopted as the local features. Specifically, by selecting a plurality of feature values with a high appearance frequency in the original color image and representing the region approximate to the selected feature value with the corresponding feature value,
This is a method of creating a region for each feature value, and the original color image can be divided into partial regions based on a relatively small number of types of feature values. Since it is only assigned, the amount of color image data can be significantly reduced as a whole, and regions having different characteristic values indicate different objects and the like. Image recognition can be achieved easily and accurately.

Further, the image is raster-scanned to search for a pixel which does not belong to any area, and the difference between the gray level of the corresponding pixel and the gray level of a pixel which does not yet belong to any area in the vicinity thereof. When the value is less than or equal to a predetermined threshold value, the area is integrated as one area, the above-mentioned processing is performed by paying attention to the newly integrated pixel, and the above processing is repeated until the area cannot be further expanded. The iterative region expansion method in which the value of each pixel is replaced by the average value in the region and a point for increasing the threshold value is added to the simple region expansion method (19
An image analysis handbook, pages 689 to 690, published on January 17, 1991 by the University of Tokyo Press) has been proposed.
If these methods are adopted, it is possible to expand the region by sequentially integrating the pixels, provided that the difference in gray level between the pixels or the difference between the gray level in the pixel and the average gray level in the region is less than or equal to the threshold value. The image can be divided into a plurality of areas corresponding to.

[0006]

However, since the original color image is considerably superposed with random noise due to variations in the sensor for capturing the color image,
Since such random noise may be significantly different from the surrounding brightness, color, etc., when trying to divide a range that is the same as or close to the selected feature value as a partial image area, random noise There is a high possibility that the part will be left as it is. As a result, in addition to a relatively large area based on the selected feature value, a large number of minute areas due to random noise or the like will be formed. As a result, not only can the processing time for the color image as a whole not be shortened (processing time becomes significantly longer than when there is no random noise, etc.),
As a result of the color image compression processing as the post-processing of the region division, a very good result cannot be obtained as the color image recognition result.

When the original color image is out of focus, an elongated area having a characteristic value different from that of any of the areas is generated at the boundary between the divided partial areas {FIG. 10 (A)]. In addition, when random noise or the like is superimposed on the boundary between the divided partial images, the boundary between the divided partial areas becomes uneven. )reference}. It has been proposed to apply filtering that emphasizes edges to such a color image, but overshoot and undershoot will occur in the edge portion due to the filtering, so overshoot,
There is a new inconvenience that the undershoot affects the color image compression processing result and the color image recognition processing result.

Further, in order to remove a ghost that occurs at the boundary between partial images, a method has been proposed in which a possible image pattern is stored in advance and the ghost is removed based on the stored pattern. (Tokuhei 4-220
69 publication). However, if this method is adopted, for example, when the number of colors is large, the number of combination patterns to be stored becomes remarkably large, and since all these combination patterns must be stored, the memory capacity becomes remarkably large. I will end up. Further, if the number of combination patterns is remarkably large, it takes a very long time to collate the color image to be processed with the stored pattern, and the processing time is remarkably lengthened as a whole.

[0009]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and it is possible to achieve area division of a color image while surely preventing generation of unnecessary areas due to random noise, out-of-focus, and the like. An object of the present invention is to provide an image processing method and its apparatus.

[0010]

A color image processing method according to claim 1 for achieving the above object is to select a predetermined number of representative colors on the basis of a color image, and to select a color on the basis of the selected representative colors. The image is divided into multiple areas, and based on the size of the area obtained as a result of the division and the color difference between adjacent areas, it is determined whether the image quality change can be visually recognized, and the image quality change can be visually recognized. In this method, the corresponding area is integrated with the adjacent area in response to the determination that it is not.

According to a second aspect of the present invention, there is provided a color image processing apparatus, wherein representative color selecting means for selecting a predetermined number of representative colors based on the color image, and an area for dividing the color image into a plurality of areas based on the selected representative color. The dividing means, the area discriminating means for discriminating whether or not the image quality change can be visually recognized based on the size of the area obtained as a result of the division and the color difference with the adjacent area, and the image quality change not visually recognizable Area integration means for integrating the corresponding area with the adjacent area in response to the determination result of the area determination means.

According to a third aspect of the color image processing method, a predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and the result of the division is compared. In response to the existence of a pixel area that is allowed to belong to one area at the boundary between the relatively large areas, the corresponding pixel area is relatively large based on the width of the relatively large areas in the predetermined range. This is a method of integrating into any area.

According to another aspect of the color image processing apparatus of the present invention, a representative color selecting means for selecting a predetermined number of representative colors based on the color image and a region for dividing the color image into a plurality of regions based on the selected representative color. There is a dividing unit, a pixel region determining unit that determines whether or not a pixel region allowed to belong to one region exists at the boundary between relatively large regions as a result of the division, and the pixel region exists. In response to the discrimination result of the pixel region discriminating means, the pixel region consolidating unit consolidates the corresponding pixel region into one of the relatively large regions based on the width of the relatively large regions in the predetermined range. I'm out.

According to a fifth aspect of the color image processing method, a predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and the result of the division is compared. Method for merging a corresponding elongated area into one of the relatively large areas based on the color difference between the corresponding elongated area and the relatively large area in response to the existence of the elongated area at the boundary between the relatively large areas Is.

According to another aspect of the color image processing apparatus of the present invention, a representative color selecting means for selecting a predetermined number of representative colors based on the color image and an area for dividing the color image into a plurality of areas based on the selected representative color. The dividing means, the elongated area discriminating means for discriminating whether or not there is an elongated area at the boundary between relatively large areas as a result of the division, and the discrimination result of the elongated area discriminating means for indicating the existence of the elongated area. In response to
And an elongated region integrating means for integrating the corresponding elongated region into any of the relatively large regions based on the color difference between the corresponding elongated region and the relatively large region.

[0016]

According to the color image processing method of claim 1, a predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and the result of the division is determined. Based on the size of the obtained area and the color difference with the adjacent area, it is determined whether or not the image quality change is visually recognizable, and it is applicable in response to the determination that the image quality change is visually unrecognizable. Since the area is integrated with the adjacent area, it is possible to reliably prevent the area where the image quality change cannot be visually recognized due to random noise etc. from remaining independently, and to divide it into the minimum number of partial areas. it can. Further, since the areas to be integrated are the areas in which the change in the image quality cannot be visually recognized, there is no inconvenience that the image quality is deteriorated due to the area integration.

According to the color image processing apparatus of claim 2,
The representative color selecting means selects a predetermined number of representative colors based on the color image, and the area dividing means divides the color image into a plurality of areas based on the selected representative colors. And
Based on the size of the area obtained as a result of the division and the color difference with the adjacent area, the area determination means determines whether or not the image quality change can be visually recognized in each area, and the image quality change cannot be visually recognized. In response to the discrimination result of the region discriminating means, the corresponding region is integrated with the adjacent region by the region consolidating means, so that the region where the image quality change cannot be visually recognized due to random noise or the like is singular. It can be surely prevented from remaining, and can be divided into a necessary minimum number of partial areas. Further, since the areas to be integrated are the areas in which the change in the image quality cannot be visually recognized, there is no inconvenience that the image quality is deteriorated due to the area integration.

According to the color image processing method of claim 3,
A predetermined number of representative colors is selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and as a result of the division, one of the areas belongs to the boundary between relatively large areas. In response to the existence of a pixel area in which the above is allowed, the corresponding pixel area is integrated into one of the relatively large areas based on the width of the relatively large areas in the predetermined range. Even if the boundary between relatively large areas due to the above is an uneven shape due to the influence of the pixel area, it is possible to eliminate the unevenness by integrating the uneven portion into any relatively large area, Region division can be achieved with simple boundaries. Therefore, it is possible to reduce the processing load such as color image compression and color image recognition that are likely to be performed as post-processing, reduce the processing time, and improve the processing accuracy.

According to the color image processing apparatus of claim 4,
The representative color selecting means selects a predetermined number of representative colors based on the color image, and the area dividing means divides the color image into a plurality of areas based on the selected representative colors. And
As a result of the division, the pixel area determining unit determines whether or not there is a pixel area allowed to belong to one area at the boundary between relatively large areas, and a pixel indicating that the pixel area exists. In response to the determination result of the area determination means, the pixel area integration means integrates the corresponding pixel area into one of the relatively large areas based on the width of the relatively large areas in the predetermined range. Even if the boundary between relatively large areas due to the above is an uneven shape due to the influence of the pixel area, it is possible to eliminate the unevenness by integrating the uneven portion into any relatively large area, Region division can be achieved with simple boundaries. Therefore, it is possible to reduce the processing load such as color image compression and color image recognition that are likely to be performed as post-processing, reduce the processing time, and improve the processing accuracy.

According to the color image processing method of claim 5,
A predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and as a result of the division, elongated areas are present at the boundaries between relatively large areas. In response to this, the corresponding elongated area is integrated into one of the relatively large areas based on the color difference between the corresponding elongated area and the relatively large area. Even if a long and narrow region having a feature value different from that of any of the regions is formed between them, the relatively large regions can be adjacent to each other by integrating the long and thin regions into one of the relatively large regions. Region division can be achieved in a small number. Therefore, it is possible to reduce the processing load such as color image compression and color image recognition that are likely to be performed as post-processing, reduce the processing time, and improve the processing accuracy.

According to another aspect of the color image processing apparatus of the present invention, the representative color selecting means selects a predetermined number of representative colors based on the color image, and the area dividing means selects a plurality of color images based on the selected representative colors. Split into. Then, as a result of the division, whether or not there is an elongated area at the boundary between relatively large areas is determined by the elongated area determination means, and responds to the determination result of the elongated area determination means indicating that the elongated area exists. Then, based on the color difference between the corresponding elongated area and the relatively large area, the elongated area integrating means integrates the corresponding elongated area into one of the relatively large areas. Even if a long and narrow area having a feature value different from that of any of the large areas is formed between the large areas, the relatively large areas can be adjacent to each other by integrating the long and narrow areas into any of the relatively large areas. Region division can be achieved in a small number. Therefore, it is possible to reduce the processing load such as color image compression and color image recognition that are likely to be performed as post-processing, reduce the processing time, and improve the processing accuracy.

[0022]

Embodiments will be described in detail below with reference to the accompanying drawings showing embodiments. FIG. 1 is a flow chart for explaining an embodiment of the color image processing method of the present invention, and step SP1
In step SP2, a plurality of representative colors (colors that appear frequently in the color image) are selected, and a color that is not selected as a representative color in step SP3 is a representative color that can be most approximated. Substitution is performed and partial image areas are obtained for each representative color in step SP4. Then, in step SP5, it is determined whether or not there is a partial image area whose image quality change cannot be visually identified. If there is a partial image area whose image quality change cannot be visually identified, then in step SP6 The partial image area whose change in image quality cannot be identified is integrated with any partial image area, and the determination in step SP5 is performed again. When it is determined in step SP5 that there is no partial image area whose image quality change cannot be visually identified, each partial image area is approximated by a polygon in step SP7, and a series of processing is terminated.

Regarding the processing of steps SP2 and SP3, for example, color data (for example, RGB data) of a color image to be processed is converted into chromaticity data (for example, L ^* a ^* b ^*).
Data), the corresponding chromaticity values are selected as the representative colors in descending order of appearance frequency, and the colors not selected as the representative colors are the chromaticity values of the representative color with the shortest distance on the chromaticity diagram. Replace.

The determination in step SP5 is made based on the color difference between adjacent partial image areas and the size of the partial image area to be processed. Specifically, the identification shown in FIG. 2 is based on whether or not the image quality change can be visually identified as a whole when various partial image areas are integrated and removed to other partial image areas having the smallest color difference. An impossible area is created, and determination is made based on whether or not it belongs to the unidentifiable area shown in FIG. In FIG. 2, the horizontal axis represents the area size (the number of pixels × the number of pixels), and the vertical axis represents the color difference ΔEab ′ between the adjacent partial image areas (for example, CIE 1976L ^*).
a ^* b ^* 1/10 of the color difference value of the color system). When the boundary line defining the unidentifiable area is approximated by a function, ΔEab ′ = ∞ when the area size is 6 or less, and when the area size is larger than 6, the identification results by a large number of subjects are obtained. Based on the region size 7, 10, 20,
ΔEab ′ is 93, 83, 70, 6 corresponding to 30, 40, 50, 60, 100, 200, 500 respectively.
The point sequence of 2, 58, 52, 50, 44, 36, 36 corresponds to Y = 64 · exp (−0.026X) +.
It can be approximated by a function of 36. Where X is the area size,
Y is the value of ΔEab ′.

By performing the above-described series of processing, a region whose image quality change cannot be visually identified due to random noise or the like is integrated into any adjacent region (specifically, the region having the smallest color difference). Then, the number of partial image areas finally obtained can be greatly reduced.

[0026]

[Embodiment 2] FIG. 3 is a block diagram showing an embodiment of a color image processing apparatus of the present invention, which represents a color image capturing section 1 and a plurality of colors having a high appearance frequency based on the captured color image. A representative color selection unit 2 for selecting a color,
A replacement unit 3 that replaces a color that has not been selected as a representative color with a representative color that can be most approximated; an image dividing unit 4 that obtains a partial image area for each representative color based on the replacement result by the replacement unit 3.
A region discriminating unit 5 that discriminates whether or not the partial image region obtained by the image dividing unit 4 is a region in which the image quality change can be visually identified, and a region discriminating unit that indicates that the image quality change cannot be visually discerned. In response to the determination result of 5, the area integration unit 6 integrates the corresponding area with the adjacent area having the smallest color difference among the adjacent areas. The operation of each component is the same as the process of the corresponding step of the flowchart of FIG. 1, and thus detailed description will be omitted.

Therefore, by adopting the color image processing apparatus shown in FIG. 3, the area in which the change in image quality is visually unrecognizable due to random noise or the like is integrated into the adjacent area having the smallest color difference, and the final color difference is finally obtained. It is possible to significantly reduce the number of partial image areas obtained in step 1.

[0028]

[Third Embodiment] FIG. 4 is a flow chart for explaining another embodiment of the color image processing method of the present invention.
In P1, a pixel is selected from the image obtained as a result of the area division by the flowchart of FIG. 1, the block diagram of FIG. 3 or other conventionally known methods or apparatuses, and in step SP2, the selected pixel is centered. A rectangular area (for example, an area of L pixels × L pixels) is set, the area occupying the largest area in the rectangular area is extracted in step SP3, and the selected pixel is extracted in step SP4. If it is determined that the selected pixel does not belong to the extracted area, it is determined in step SP5 that the color difference between the selected pixel and the extracted area is predetermined. Is less than or equal to the threshold value (for example, about twice the threshold value set when the area division is performed). Then, when it is determined that the color difference between the selected pixel and the extracted area is less than or equal to the predetermined threshold value, the selected pixel is integrated into the extracted area in step SP6, and step SP7
In step SP8, it is determined whether or not all the pixels have been processed. If it is determined that there is a pixel that has not been processed, the next pixel is selected in step SP8, and the processing in step SP2 is performed again. Repeat the process. If it is determined that the pixel selected in step SP4 belongs to the extracted area, step S
When it is determined in P5 that the color difference between the selected pixel and the extracted area exceeds the predetermined threshold value, the determination in step SP7 is performed as it is. That is, in this case, the integration process is not performed. If it is determined in step SP7 that all the pixels have been processed, the series of processes is terminated.

FIG. 5 is a schematic diagram for explaining the pixel area integration processing. When the pixel indicated by P1 in FIG. 5A is the processing target, L pixel × L centered on this pixel P1.
A region of pixels is set, and the region R1 is the widest in this region. Therefore, the pixel P1 is integrated into the region R1 on condition that the color difference between the pixel P1 and the region R1 is equal to or less than the threshold value (see FIG. 5 (B)). Thereafter, by repeating the same processing, as shown in FIG.
The boundary line between the region R2 and the region R2 is changed from the unevenness to the straight line. That is, the shapes of both regions R1 and R2 can be simplified.

[0030]

[Embodiment 4] FIG. 6 is a block diagram showing another embodiment of the color image processing apparatus of the present invention, in which a pixel extraction unit for extracting one pixel as a processing target pixel from a color image subjected to area division processing. 11, a rectangular area setting unit 12 that sets a rectangular area centered on the extracted pixel, an area extraction unit 13 that extracts the widest area in the set rectangular area, and the extracted pixel is extracted. The pixel determination unit 14 that determines whether or not the extracted pixel belongs to the extracted region and the determination result of the pixel determination unit 14 that indicates that the extracted pixel does not belong to the extracted region. A color difference calculating unit 15 for calculating a color difference between the extracted area, a color difference determining unit 16 for determining whether the calculated color difference is less than or equal to a predetermined threshold, and an area where the extracted pixel is extracted. That it does not belong In response to the discrimination result of the pixel discriminating unit 14 and the discrimination result of the color difference discriminating unit 16 indicating that the calculated color difference is equal to or less than a predetermined threshold value, pixel integration for consolidating the extracted pixel into the extracted region. It has a unit 17 and an iterative control unit 18 that iterates the processing by the above-described respective units for all pixels included in the color image.

The operation of each component is the same as the processing of the corresponding steps in the flowchart of FIG. 4, so detailed description will be omitted. Therefore, by adopting the color image processing apparatus of FIG. 6, the boundary line between the adjacent regions is changed from the unevenness to the straight line, and the shapes of both regions can be simplified.

[0032]

[Fifth Embodiment] FIG. 7 is a flow chart for explaining still another embodiment of the color image processing method of the present invention. In step SP1, the flow chart of FIG. 1, the block diagram of FIG. An elongated area having a width of one pixel is extracted from the image resulting from the area division by the apparatus in the scanning direction or in the direction orthogonal to the scanning direction, and in step SP2, the orientation of the elongated area is discriminated. If the orientation is the direction orthogonal to the scanning direction, the color difference between the pixels in the scanning direction and the preceding and succeeding pixels is calculated in step SP3, and the process target pixel in the elongated region is integrated into the region to which the pixel with the small color difference belongs in step SP4. . On the contrary, if it is determined in step SP2 that the direction of the elongated region is the scanning direction, step S
In P5, the color difference between the pixels preceding and succeeding in the direction orthogonal to the scanning direction is calculated, and in step SP6, the process target pixel in the elongated region is integrated into the region to which the pixel having a small color difference belongs. After the processing in step SP4 or step SP6 is performed, the next pixel belonging to the elongated area is selected again in step SP8 until it is determined in step SP7 that all pixels belonging to the elongated area have been processed. The determination in step SP2 is performed. If it is determined in step SP7 that all pixels belonging to the elongated region have been processed, step SP
In step 9, it is determined whether or not the processing has been performed for all the elongated areas. If there is an elongated area that has not been processed, another elongated area is extracted in step SP10, and the determination in step SP2 is performed again. To do. If it is determined in step SP9 that the processing has been performed for all the elongated areas, the series of processing ends.

FIG. 8 is a diagram schematically showing the integration processing of the slender areas, and the areas r1, r2 shown in FIGS. 8A and 8B.
r3 and r4 are extracted as elongated regions. FIG. 8 (C),
(D) shows the integration processing when the region r1 is extracted, and in the figure it is assumed that the color difference with the left pixel is small, so the pixels belonging to the elongated region are integrated into the left region. {See FIG. 8D}.

Therefore, for example, even if the pixel areas are integrated by performing the processing shown in FIG. 4 and the elongated areas remain due to the color difference exceeding a predetermined threshold, the processing shown in FIG. By doing so, the elongated area can be integrated into the area having the smaller color difference, the boundary between the partial image areas can be simplified, and the shape of each partial image area can be simplified. Of course, the present invention can be similarly applied to a color image in which an elongated area due to out-of-focus is generated, and the same effect can be achieved.

[0035]

[Sixth Embodiment] FIG. 9 is a block diagram showing still another embodiment of the color image processing apparatus of the present invention, in which one elongated region is extracted as a processing subject elongated region from a color image subjected to region division processing. An elongated region extraction unit 21, an orientation detection unit 22 that detects the orientation of the extracted elongated region, and a color difference calculation unit that computes a color difference between a pixel that is adjacent in the direction orthogonal to the detected orientation and a pixel that belongs to the elongated region. 23, a slender area unifying unit 24 that unifies slender areas with respect to an area to which a pixel having a large calculated color difference belongs, and an iterative control unit 25 that repeats the processing by the above-described constituent units for all slender areas included in a color image. And have.

The operation of each component is the same as the process of the corresponding step of the flowchart of FIG. 7, so detailed description will be omitted. Therefore, by adopting the color image processing device of FIG. 9, it is possible to integrate the elongated region into the region on the smaller color difference side and simplify the boundary between the partial image regions.
As a result, the shape of each partial image area can be simplified.

The present invention is not limited to the above-described embodiment, and for example, a pixel area larger than one pixel can be set as the processing target pixel area in the third and fourth embodiments. It is possible to set a region having a width exceeding one pixel as the elongated regions in the fifth and sixth embodiments, and various design changes can be made within a range not changing the gist of the present invention.

[0038]

As described above, according to the first aspect of the present invention, it is possible to surely prevent a region, in which a change in image quality is visually unrecognizable due to random noise or the like, from remaining alone, and to minimize the necessary minimum. It has a unique effect that it can be divided into a number of partial areas and that the deterioration of the image quality due to the area integration can be surely prevented.

According to the second aspect of the present invention, it is surely prevented that a region in which a change in image quality cannot be visually recognized due to random noise or the like remains alone, and the region is divided into a minimum number of partial regions. This has a unique effect of being able to reliably prevent deterioration of image quality due to area integration. Claim 3
In the invention, even if the boundary between relatively large areas due to random noise or the like has an uneven shape due to the influence of the pixel area, the uneven portion is integrated into any relatively large area to eliminate the unevenness. By doing so, it is possible to achieve area division with a simple boundary line, and it is possible to reduce the processing load such as color image compression and color image recognition that is likely to be performed as post-processing, and to reduce the processing time. It has a unique effect that it can be shortened and the processing accuracy can be improved.

According to the fourth aspect of the present invention, even if the boundary between relatively large areas is uneven due to the influence of the pixel area due to random noise or the like, the uneven portion is formed in any relatively large area. By integrating and eliminating unevenness, it is possible to achieve area division with a simple boundary line, and it is possible to reduce the processing load such as color image compression and color image recognition that is likely to be performed as post-processing. At the same time, it has a unique effect that the processing time can be shortened and the processing accuracy can be improved.

According to the fifth aspect of the present invention, even if a long and narrow region having a feature value different from that of any of the regions is formed between the relatively large regions due to out-of-focus or the like, the long and thin region is defined as any of the relatively large regions. By combining the areas with relatively large areas adjacent to each other, it is possible to achieve area division in a state where the number of areas is small, and as a result, processing load such as color image compression and color image recognition that is likely to be performed as post-processing. And the processing time can be shortened and the processing accuracy can be improved.

According to the sixth aspect of the present invention, even if elongated regions having different feature values are formed between the relatively large regions due to out-of-focus or the like, the elongated regions are defined as any relatively large regions. By combining the areas with relatively large areas adjacent to each other, it is possible to achieve area division in a state where the number of areas is small, and as a result, processing load such as color image compression and color image recognition that is likely to be performed as post-processing. And the processing time can be shortened and the processing accuracy can be improved.

[Brief description of drawings]

FIG. 1 is a flowchart illustrating an embodiment of a color image processing method of the present invention.

FIG. 2 is a diagram showing an unidentifiable area.

FIG. 3 is a block diagram showing an embodiment of a color image processing device of the present invention.

FIG. 4 is a flowchart illustrating another embodiment of the color image processing method of the present invention.

FIG. 5 is a schematic diagram illustrating pixel area integration processing.

FIG. 6 is a block diagram showing another embodiment of the color image processing apparatus of the present invention.

FIG. 7 is a flow chart for explaining still another embodiment of the color image processing method of the present invention.

FIG. 8 is a diagram schematically showing an integration process of elongated regions.

FIG. 9 is a block diagram showing still another embodiment of the color image processing apparatus of the present invention.

FIG. 10 is a diagram schematically illustrating an inconvenience caused by conventional image division.

[Explanation of symbols]

2 representative color selection unit 4 image division unit 5 region discrimination unit 6 region integration unit 14 pixel discrimination unit 15 pigment calculation unit 16 pigment discrimination unit 17 pixel integration unit 21 elongated region extraction unit 22 orientation detection unit 23 color difference calculation unit

Claims (6)

[Claims] 1. A predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and the size of the area obtained as a result of the division and the adjacent area are divided. The feature is that it determines whether or not the image quality change is visually recognizable based on the color difference of, and integrates the corresponding region with the adjacent region in response to the determination that the image quality change is not visually recognizable. And color image processing method. 2. A representative color selecting means (2) for selecting a predetermined number of representative colors based on the color image, and an area dividing means (4) for dividing the color image into a plurality of areas based on the selected representative color. And an area discriminating means (5) for discriminating whether or not the image quality change can be visually recognized based on the size of the area obtained as a result of the division and the color difference with the adjacent area, and the image quality change can be visually recognized. A color image processing apparatus comprising: an area unifying means (6) for unifying the corresponding area with an adjacent area in response to a discrimination result of the area discriminating means (5) indicating that it is not. 3. A predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative color, and as a result of the division, a boundary portion between relatively large areas is formed. In response to the existence of a pixel area that is allowed to belong to one area, the corresponding pixel area is integrated into one of the relatively large areas based on the width of the relatively large areas in a predetermined range. And a color image processing method. 4. A representative color selecting means (2) for selecting a predetermined number of representative colors based on the color image, and an area dividing means (4) for dividing the color image into a plurality of areas based on the selected representative color. (5) and (6), and as a result of the division, a pixel area determination unit (14) (that determines whether or not there is a pixel area that is allowed to belong to one area at the boundary between relatively large areas. 15) and in response to the discrimination result of the pixel region discriminating means (14) (15) indicating that the pixel region exists, the corresponding pixel regions are compared based on the width of the relatively large regions in the predetermined range. A color image processing apparatus, comprising: a pixel area integrating means (16) (17) for integrating into any area that is relatively large. 5. A predetermined number of representative colors are selected based on the color image, the color image is divided into a plurality of areas based on the selected representative colors, and as a result of the division, a boundary portion between relatively large areas is formed. A color image processing method, characterized in that, in response to the presence of an elongated region, the corresponding elongated region is integrated into a region having a small color difference based on a color difference between the elongated region and a relatively large region. 6. A representative color selecting means (2) for selecting a predetermined number of representative colors based on the color image, and an area dividing means (4) for dividing the color image into a plurality of areas based on the selected representative color. (5) (6) and, as a result of the division, an elongated area discriminating means (21) for discriminating whether or not an elongated area exists at the boundary between relatively large areas, and showing that the elongated area exists. Responsive to the discrimination result of the slender region discriminating means (21), the slender region unifying means (22) () for consolidating the corresponding slender region into the region having the small color difference based on the color difference between the slender region and the relatively large region. 23) and a color image processing device.