JPH0773325A - Specified image detecting method - Google Patents

Abstract

PURPOSE:To accurately and easily detect an arbitrary number of specified images contained in a color image. CONSTITUTION:A histogram 103 for the chromaticity of R is prepared by calculating the R chromaticity from the feature amount of an input image such as RGB data, for example, and matching is performed with the respective constant-fold values (0-fold, 1-fold, 2-fold, 3-fold and 4-fold values) of the R chromaticity histogram in a dictionary 104. The multiple of the R chromaticity, of which degree of difference is minimum, (=the number of specified images) and that degree of difference are outputted. When the multiple to be outputted is '0', no specified image is contained in the input image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for detecting a specific image in a color image input by a color scanner or the like.

[0002]

2. Description of the Related Art As a method of detecting a specific color image, a method of checking whether or not a pixel of a previously specified color appears (Japanese Patent Laid-Open No. 4-180348), and how many pixels of a previously specified color as a whole appear How to check whether or not it has been done
7465) is known.

[0003]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
Like the method of No. 348, regarding the color distribution in the spread of the color space indicated by the dictionary of the feature amount of the specific image to be detected, a method that does not pose a problem in image identification is Images with the same spread cannot be identified even if they have different color distributions. That is, if the spread of the color space is the same, an image having a different color distribution in the spread may be erroneously detected as a specific image.

On the other hand, as in the method of the above-mentioned Japanese Patent Laid-Open No. 4-207465, in the method of observing the degree of coincidence of the spread in the entire color space (for example, FIGS. 4 and 5 of the same publication), the hue is It is impossible to identify the difference between. Further, it is not possible to deal with the case where an arbitrary number of specific images are mixed in the images to be processed (for example, when an arbitrary number of originals of a specific image are simultaneously read by a color scanner or the like).

Accordingly, one object of the present invention is to provide a method for detecting a specific color image with higher accuracy.

Another object of the present invention is to provide a method for highly accurately detecting an arbitrary number of specific images existing in an input image.

Another object of the present invention is to provide a method for detecting a specific image existing in an input image including its number.

Another object of the present invention is to provide a method for detecting a specific image in a color image input from an image reading device such as a color scanner with high accuracy.

[0009]

According to the present invention, the feature amount obtained from the input image data is matched with the feature amount which is a constant multiple of the feature amount prepared in advance as a dictionary for the specific image. . In addition, a multiple of the feature amount that best matches is set as the number of specific images.

A chromaticity histogram is used as the characteristic amount, and preferably, a chromaticity histogram reflecting the level of lightness of pixels is used.

The matching process between the feature quantities is performed by excluding the dimension of which the value of the feature quantity as a dictionary is 0. In addition, when the feature amount of the input image data is larger than the feature amount as a dictionary, a membership function that sets the membership value to 0 or more is used to calculate the similarity for each dimension when matching between the feature amounts. .

[0012]

In the case where the input image includes only one specific image by matching each feature amount obtained by multiplying the feature amount prepared as a dictionary with the feature amount obtained from the input image data. Just by preparing a dictionary that assumes, the specification of input image data that contains an arbitrary number of specific images, such as image data obtained by simultaneously reading two or more originals of specific images with a color scanner, etc. The image can be detected. In addition, the number of specific images included in the input image can be easily determined as a multiple of the feature amount that is best matched (maximum similarity or minimum dissimilarity) among the constant-multiplied feature amounts.

By using the chromaticity histogram as the feature quantity, images having different color distributions (hue) can be identified even if the spreads in the color space are overlapped. Therefore, it is possible to check only the appearance of a specific color. Compared with, it is possible to detect a specific image having a fine texture with high accuracy. Further, when an image is read by a color scanner or the like and input, the chromaticity histogram is better for the colors (R, G,
Compared with the histogram of B), there is an advantage that it is less affected by the change in the light amount of the image reading light source.

When the chromaticity histogram is used as the feature quantity, the influence of noise in the input image data can be reduced by calculating the similarity for each dimension between the feature quantities using the membership function as described above. it can.

In a high lightness region of an image, a CCD image sensor, which is a general image reading element, has a large sensitivity variation, and the hue is easily influenced by the hue of the background during reading. According to the present invention, by using a chromaticity histogram that focuses on pixels with low brightness and reflects the brightness of the pixels, it is possible to reduce the above sensitivity variations and the influence of the background. It is possible to improve the detection accuracy of the specific image in the read and input image.

[0016]

【Example】

<Example 1> In this example, a chromaticity histogram of R, which is a one-dimensional feature amount, is used for image identification. A system configuration according to this embodiment is shown in FIG.

In FIG. 1, a color image input device (color scanner) 100 inputs a color image as RGB data. An example of the input image is schematically shown in FIG. In this input image, A1 and A2 are specific image portions to be detected.

In FIG. 1, in the R chromaticity calculation unit 101,
From the RGB data of the input image, for example, the chromaticity r of R of each pixel is calculated by the following equation r = 255 × R / (R + G + B), and the calculated R chromaticity data is sequentially input to the matching unit 102.

In the matching unit 102, a histogram (one-dimensional feature amount) 103 of the input R chromaticity is created. When the R chromaticity calculation unit 101 finishes the calculation process of all R chromaticity of the input image and the matching unit 102 finishes the creation process of the R chromaticity histogram 103, the matching unit 102 then starts the matching process. To do. Here, it is assumed that the R chromaticity histogram as shown in FIG. 3A is obtained.

In the matching process, first, the feature amount (R chromaticity histogram) in the dictionary 104 of the specific image to be detected.
The value of each dimension of is multiplied by a constant (here, 0 times, 1 times, 2 times, 3
2 times, 4 times) feature quantities are created. Here, the dictionary 10
If the feature quantity in 4 is as shown in (c) of FIG. 3, the feature quantities as shown in (b) to (f) of FIG. 3 are created.

Then, a constant multiple (0 times, 1 times, 2 times, 3
For each of the feature amounts multiplied by 4 and 4, the difference from the R chromaticity histogram 103 of the input image is calculated, and the difference is used as the degree of difference from the feature amount of each multiple. In the example shown in FIG. 3, the dissimilarity is as shown in Table 1.

[0022]

[Table 1]

Finally, among the feature amounts from 0 to 4 times, the multiple corresponding to the feature amount having the smallest difference from the R chromaticity histogram of the input image and the difference thereof are set as the output data 105. Output. In the example of FIG. 3, as can be seen from <Table 1>, the degree of difference 20 with the feature amount having a multiple of 2 is the minimum. This is because the specific image you want to detect is 2
This means that there are individual pieces (A1 and A2 in FIG. 2).

When the degree of difference from the 0-fold feature amount is the minimum, the specific image to be detected does not exist in the input image.

<Embodiment 2> Since the system configuration of this embodiment may be the same as that of the above-mentioned Embodiment 1, it will be described with reference to FIG.
Similar to the first embodiment, an image is input by the color image input device 100, the R chromaticity of each pixel of the input image is calculated by the R chromaticity calculation unit 101, and the histogram of the R chromaticity (feature amount) is calculated by the matching unit 102. ) 103 and creates a dictionary 104
The feature amount obtained by multiplying the respective dimension values of the inside feature amount by 0 times, 1 time, 2 times, 3 times, and 4 times is calculated.

The difference from the first embodiment is that in the matching process in the matching unit 102, the similarity (difference) between the feature amount of the input image and the feature amount which is a constant multiple (0 to 4 times) of the dictionary feature amount. Is calculated for each dimension using a membership function as shown in FIG. 4, and the average value of the similarity of all dimensions is calculated as the final similarity.

If the feature quantity of the input image and the feature quantity in the dictionary are a multiple of the feature quantity as shown in FIGS. 3A to 3F, the similarity for each dimension and its average value are <Table 2>
As shown in.

[0028]

[Table 2]

The degree of similarity between the feature amount obtained by multiplying the feature amount in the dictionary by 1 and the feature amount of the input image in each dimension will be briefly described.
In the first dimension, the value on the dictionary side is 5, and the value on the input image side is 10.
Therefore, the degree of similarity = 1− (10−5) /10=0.5
Is. In the second dimension, the value on the dictionary side is 10 and the value on the input image side is 20, so the degree of similarity = 1− (20−10) / 1.
0 = 0. In the third dimension, the value on the dictionary side is 20 and the value on the input image side is 50, so the similarity = 1- (50-2
Although 0) / 10, the membership function is determined so that the membership value does not take a negative value when the value on the input side is larger than the value on the dictionary side, so the similarity is 0.
In the fourth dimension, the similarity is 1- (10-0) / 10 = 0.

A case of matching the feature amount obtained by multiplying the feature amount in the dictionary by four with the feature amount of the input image will be described. In the first dimension, the value on the dictionary side is 20, and the value on the input image side is 10.
Therefore, the similarity becomes 1− (20−10) / 10 = 0. In the second dimension, similarity = 1− (40−20) / 1
0 = -1.0. Third-dimensional similarity = 1− (80−
50) /10=−2.0. Fourth-dimensional similarity = 1
-(0-10) / 10 = 0.

The multiple of the dictionary feature value having the maximum average similarity calculated as described above and the average similarity are output as the matching result. In the example shown in FIG. 3, <Table 2>
As is clear from the above, a multiple of 2 (meaning that there are two specific images to be detected in the input image) and a similarity of 0.5 is output.

When the feature amount of the input image, such as the membership function shown in FIG. 4, is larger than the feature amount obtained by multiplying each multiple of the feature amount in the dictionary, the membership value (similarity of each dimension) is obtained. If a specific image to be detected exists in the same color space, the number of pixels in the mixed part in the color space is disturbed, and the membership function that does not become negative is used. The benefit is that the decline in (If the feature amount of the input image is less than the feature amount of the dictionary, it is clear that the image to be detected does not exist, so the membership value may take a negative value.) <Example 3 > The system configuration according to the present embodiment may be the same as in the first and second embodiments. The process from image input to the creation of the R chromaticity histogram and the creation of the feature amount that is a constant multiple of the feature amount in the dictionary in the matching process are the same as in the second embodiment.

In the matching process, the membership function shown in FIG. 4 is used to calculate the similarity between the feature amount of the input image and the feature amount that is a constant multiple of the dictionary, as in the second embodiment. However, the difference from the second embodiment is that the dimension of which the value of the feature amount in the dictionary is 0 is ignored in the similarity calculation. For example, the matching process is not performed for the fourth dimension of the feature amount shown in FIGS. 3B to 3F. Therefore, the feature amount of the input image shown in FIG. 3A and the feature amount which is a constant multiple of the feature amount in the dictionary shown in FIGS. 3B to 3F are similar to those shown in Table 3. You get a degree. It is natural that the ignored dimension (the fourth dimension in this example) is also ignored when calculating the average similarity.

[0034]

[Table 3]

In this example, a multiple of 2 (the number of detected pixels) having the maximum average of the similarity in all dimensions and a similarity of 0.667 are output as the output data 105.

<Fourth Embodiment> FIG. 5 shows a system configuration according to the present embodiment. In FIG. 5, a color image input device (color scanner) 500 inputs a color image as RGB data to the effective image extraction unit 501.

In the effective pixel extraction unit 501, the input RG
From the B data, the brightness L of each pixel is calculated according to the following formula. L = (R + G + B) / 3 Then, the effective pixel extraction unit 501 determines a pixel whose brightness L is smaller than a predetermined threshold value Lth as an effective pixel, and determines the RGB data of the pixel as the R chromaticity calculation unit 502 and the G color. Input to the degree calculation unit 503. However, for a pixel whose brightness is equal to or higher than the threshold value Lth, the RGB data of the pixel is not input to the R and G color calculation unit 502, and 0 is set to the chromaticity buffer 504 as the values of R chromaticity and G chromaticity. Output.

The R chromaticity calculation unit 502 has an effective pixel extraction unit 5
The chromaticity r of R is calculated from the RGB data of the effective pixel input by 01 by the following equation, and the calculated chromaticity r is
Output to. r = 255R / (R + G + B)
.

The G chromaticity calculation unit 503 has an effective pixel extraction unit 5
The chromaticity g of G is calculated by the following equation from the RGB data of the effective pixel input from 01 and is output to the chromaticity buffer 504. g = 255G / (R + G + B) Therefore, R and G chromaticity data for each pixel of the input image is obtained in the chromaticity buffer 504. However, the R and G chromaticities of pixels other than the effective pixel are all 0.

The peripheral effective pixel number calculation unit 505 creates a chromaticity histogram for each effective pixel 506 by referring to the chromaticity data in the chromaticity buffer 504. The effective pixel chromaticity histogram 506 includes an effective pixel chromaticity histogram 506a and an effective pixel colorlessness histogram 506.
It consists of two of b. Each histogram is a two-dimensional histogram as shown in FIG.

From the chromaticity data in the chromaticity buffer 504, the peripheral effective pixel number calculation unit 505 detects a pixel in which at least one of R chromaticity and G chromaticity is not 0 (called a target pixel), Count the number of pixels in which at least one of R chromaticity and G chromaticity is not 0 on the circumference of a circle centered on the pixel of interest, and pay attention to that number on the effective pixel chromaticity histogram 506a. The values of the r and g coordinates corresponding to the R and G chromaticities of the pixel are incremented. In addition, among the pixels on the circumference of the circle centered on the pixel of interest, the number of pixels in which both the R chromaticity and the G chromaticity are 0 is counted, and the number of pixels is counted on the effective pixel colorlessness histogram 506b. The values of the r and g coordinates corresponding to the R and G chromaticities of the target pixel of are increased.

An example of creating such a histogram will be described with reference to FIG. FIG. 6A schematically shows an example of the contents of the chromaticity buffer 504. The black (line) part corresponds to the effective pixel, and the white part is r = 0, g = other than the effective pixel.
Corresponds to 0 pixel. X is the pixel of interest, where r = 9
5, g = 88. There are 84 pixels on the circumference Y of the circle centered on the pixel of interest X. Of these 84 pixels, 15 pixels are pixels in which r or g is not 0, and 69 pixels are pixels in which both r and g are 0. Suppose In this case, the effective pixel chromaticity histogram 506 as shown in FIG.
Add 15 to the coordinate value of r = 95, g = 88 on a,
As shown in FIG. 6C, 69 is added to the coordinate values of r = 95 and g = 88 of the effective pixel colorlessness histogram 506b.

The detected image dictionary 508 stores, as a characteristic amount, a characteristic amount (chromaticity histogram for each effective pixel) created by the same method for a specific image to be detected. In the matching unit 507, the feature amount that is a constant multiple (0 times, 1 time, 2 times, 3 times, 4 times) of the feature amount in the dictionary 508, and the feature amount of the input image (chromaticity histogram for each effective pixel 506) Match between. In this matching process, the same method as in the first, second or third embodiment is used (the difference between the two-dimensional feature amount and the one-dimensional feature amount in the present embodiment and the first, second or third embodiment). However, the dissimilarity or similarity is calculated for each corresponding value, and the multiple of the feature amount whose average is the minimum (in the case of dissimilarity) or the minimum (similarity) and its average dissimilarity or average similarity. The degree is output as output data 509.

According to the feature amount extraction method as described above in this embodiment, the color feature amount and the texture feature amount can be extracted with a small number of dimensions. Moreover, since it is basically a chromaticity histogram of pixels with low lightness, variations in sensitivity of the CCD image sensor at the time of image reading and influence of the background can be suppressed. Therefore, it is possible to detect the specific image in the image input by the image scanner or the like with extremely high accuracy.

[0045]

As is apparent from the above description, according to the present invention, it is possible to detect a specific color image with high accuracy. Even if the number of specific images included in the input image is arbitrary, it is possible to easily detect the existence of the specific images and further the number thereof. It is possible to accurately detect a specific image in a color image input from an image reading device such as a color scanner. It is possible to achieve such effects.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a system configuration according to Examples 1, 2, and 3.

FIG. 2 is a diagram showing an example of an input image for explaining Embodiments 1, 2, and 3.

FIGS. 3A to 3F are diagrams showing an example of a feature amount of an input image and an example of a feature amount obtained by multiplying a feature amount in a dictionary by a constant, for explaining the first, second, and third embodiments.

FIG. 4 is a diagram showing an example of a membership function for calculating the degree of similarity in the second and third embodiments.

FIG. 5 is a block diagram showing an example of a system configuration according to a fourth embodiment.

6A to 6C are diagrams for explaining extraction of a feature amount of an input image according to the fourth embodiment.

[Explanation of symbols]

 100 color image input device 101 R chromaticity calculation unit 102 matching unit 103 R chromaticity histogram 104 detected image dictionary 105 output data 500 color image input device 501 effective pixel extraction unit 502 R chromaticity calculation unit 503 G chromaticity calculation unit 504 colors Degree buffer 505 peripheral effective pixel number calculation unit 506 chromaticity histogram for each effective pixel 506a effective pixel chromaticity histogram 506b effective pixel colorlessness histogram 507 matching unit 508 detected image dictionary 509 output data

Claims (8)

[Claims] 1. A method for detecting a specific image from input image data, comprising: a feature amount obtained from the input image data;
A specific image detecting method characterized in that matching is performed with a specific image obtained by multiplying a specific amount of a feature value prepared in advance as a dictionary with a specific image. 2. The specific image detection according to claim 1, wherein among the feature amounts obtained by multiplying the feature amount as a dictionary by each constant, a multiple of the feature amount that is best matched is specified as the number of the specific images. Method. 3. A dimension having a feature value of 0 as a dictionary is
3. The specific image detection method according to claim 1, wherein the specific image detection is excluded from the target of matching with the feature amount of the input image data. 4. A membership in which a membership value is set to 0 or more when the feature amount of input image data is larger than the feature amount as a dictionary in calculating the similarity for each dimension when matching between feature amounts. The specific image detecting method according to claim 1, 2 or 3, wherein a function is used. 5. The specific image detection method according to claim 1, wherein a chromaticity histogram is used as the characteristic amount. 6. The specific image detection method according to claim 1, wherein a chromaticity histogram reflecting the brightness of pixels is used as the characteristic amount. 7. A method for detecting a specific image from input image data, which comprises matching a chromaticity histogram obtained from the input image data with a chromaticity histogram prepared in advance as a dictionary for the specific image. A method for detecting a specific image, which is characterized by performing. 8. The method for detecting a specific image according to claim 7, wherein a chromaticity histogram reflecting the brightness of pixels is used.