JP2512808B2 - Specific color extraction method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a specific color extraction method for extracting a portion having a specific color of an imaged object from a color image.

[Prior Art] Recently, in order to make a robot recognize a predetermined position of an article by a visual sensor or to automatically perform a visual inspection of a product, a recognition target object is imaged by a color television camera and the object is detected. The specific color part of is extracted.

As one method of extracting the specific color, a range of the specific color to be extracted is set for each primary color coordinate axis in the three primary color vector space, and the specific color part is determined depending on whether the measured value is within this range. There is a way to extract.

However, in this method, since the brightness is greatly affected, the accuracy of distinguishing colors is low, and only objects with clear color differences can be targeted.

The following method is available as a method that is not affected by such a change in brightness. In this method, the ratio of the three primary color components forming the specific color to be extracted in the three primary color vector space is obtained, and the measured value is specified by whether it is in the area of the same ratio or an approximate ratio. The part is extracted.

[Problems to be Solved by the Invention] However, in the above method, for example, even if the brightness of the three primary color components of the measured value is uniformly high, if the ratio is the same, it is extracted as a specific color, and thus a subtle color There was a problem that they could not be distinguished.

The present invention has been made in view of the above points, and an object of the present invention is to provide a specific color extraction method that can accurately perform delicate color discrimination.

[Means for Solving the Problems] In order to achieve the above object, the present invention uses three primary color signals obtained by imaging a recognition target object with a color television camera,
The coefficients obtained from the differences between the three primary color components of the reference color and the three primary color components of the specific color to be extracted are multiplied for each primary color and then added, and this product-sum operation is performed on at least three specific colors to obtain A specific color is extracted by using the calculated product-sum operation values as a set.

It should be noted that the recognition target object was obtained by imaging with a color television camera according to the degree of correlation with the specific color of the set of product-sum operations obtained by performing the above-described product-sum operation for colors whose relationship with the specific color is already known. A specific color may be extracted by converting a set of product-sum calculated values of the three primary color signals into correlation data with the specific color.

[Operation] According to the present invention, the difference between the three primary color components of the reference color and the three primary color components of the specific color to be extracted is included in the three primary color signals obtained by imaging the recognition target object with the color television camera as described above. By multiplying the coefficients obtained by each primary color and then adding them, the three primary color signals are mapped to a vector space consisting of the vector of the difference between the reference vector and the specific color vector in the three primary color space, and this product-sum operation is performed by at least 3 By performing the operation on three or more specific colors, an arbitrary point in the three primary color vector space is uniquely expressed by a set of product-sum operation values, and the color difference is expressed in the space that maximizes the difference between the reference color and the specific color. It is possible to discriminate even the subtle difference in color.

Further, the recognition target object is obtained by imaging with a color television camera according to the degree of correlation with the specific color of the set of the product-sum calculation obtained by performing the above-mentioned product-sum calculation for the color whose relationship with the specific color is already known. By converting the set of product-sum calculation values of the three primary color signals into correlation data with the specific color, it is possible to express how close the picked-up color is to the specific color, and it is possible to perform identification with higher accuracy.

[Embodiment] FIGS. 1 to 4 show an embodiment of the present invention. In this embodiment, the specific color extracting method of the present invention is applied to an apparatus for displaying an image in which a specific color portion of a recognition target object is extracted on a display. Further, in this embodiment, as shown in FIG. ₁ , the specific color portion is extracted with respect to the object X in which the _three specific color portions x _{1 to} x ₃ of different colors exist. Note that the object X
The surface has a uniform color, and the specific color portions x _{1 to} x ₃ are present on the surface.In the following description, each specific color is assigned to each specific color portion x _{1 to} x ₃ . Correspondingly, they are called specific colors 1 to 3. Further, in this device, after extracting the specific color portions x _{1 to} x ₃ by the specific color extracting method of the present invention which will be described in detail later, the specific color portions _{1 to 3} are binarized by using the correlation threshold value, The specific color portions x _{1 to} x ₃ are individually displayed on the displays 17 to 19.

According to the present invention, the three primary color signals obtained by imaging the object X with the color television camera 1 include the three primary color components of the reference color and the three primary color components of the specific color (specific color 1 to 3) to be extracted. Coefficients obtained from the respective differences are multiplied for each primary color and then added, this product-sum operation is performed for the three specific colors 1 to 3, and the specific colors 1 to 3 are extracted using the obtained product-sum operation values as a set. . That is, in the three primary color signals obtained by imaging the object X with the color television camera 1 as described above, the coefficients obtained from the respective differences between the three primary color components of the reference color and the three primary color components of the specific color to be extracted are calculated. Performing the sum-of-products operation in which each primary color is multiplied and then added is a mapping of the three primary color signals into a vector space composed of the vector of the difference between the reference color vector and the specific color vector in the three primary color vector space. The space formed by the vector of represents the maximum color difference between the reference color and the specific color. This will be described on the two-dimensional vector space based on FIG. 2 for simplicity. In FIG. 2, when there is an input vector C with respect to the vector E which is the difference between the reference color vector A and the specific color vector B, the above-mentioned product-sum operation is performed on the input vector C by The projection onto the difference vector E is represented by F in the figure. It should be noted that there is no influence of brightness by performing a product-sum operation in which the coefficients obtained from the respective differences between the three primary color components of the reference color and the three primary color components of the specific color to be extracted are multiplied for each primary color and then added. . Performing this product-sum operation for the three specific colors 1 to 3 uniquely represents an arbitrary point in the three primary color vector space with a set of product-sum operation values. In other words, when the colors are identified in the space well expressing the color difference between the reference color and the specific color obtained in this way,
The color identification accuracy is improved, and even subtle color differences can be identified. In addition, here, the recognition target object X is colored by the phase sensitivity with the specific colors 1 to 3 of the set of the product-sum calculation obtained by performing the above-described product-sum calculation on the colors whose relationship with the specific colors 1 to 3 is known in advance. If the set of product-sum operation values of the three primary color signals obtained by imaging with the television camera 1 is converted into correlation data with a specific color, it is possible to express how close the captured color is to the specific color, and further accuracy Highly discernible.

The apparatus shown in FIG. 1 will be described below in detail. First, the color telescopic camera 1 images the recognition target object X,
The captured image data, which is the obtained analog signal, is converted into an A / D converter 2
Convert to digital signal with. A specific color portion is extracted from this conversion data by using the specific color extracting method of the present invention described above. At this time, the surface color that is the background color of the object X is used as a reference color, and its three primary color components are (Rb, Gb, Bb). Then, the three primary color components of the specific color portions x _{1 to} x ₃ are respectively (Rp ₁ , Gp ₁ , B
p ₁ ), (Rp ₂ , Gp ₂ , Bp ₂ ) and (Rp ₃ , Gp ₃ , Bp ₃ ). In this case, the coefficient by which the three primary color signals obtained by picking up the image of the object X with the color television camera 1 are multiplied is obtained as follows. Now, if the coefficients for the red component R are α ₁ , α ₂ , and α ₃ , then α ₁ = Rp ₁ −Rb α ₂ = Rp ₂ −Rb α ₃ = Rp ₃ −Rb. Similarly, when the coefficients β ₁ , β ₂ , β ₃ and γ ₁ , γ ₂ , γ ₃ for the green component G and the blue component B are obtained, β ₁ = Gp ₁ −Gb γ ₁ = Bp ₁ −Bb β ₂ = Gp ₂ −Gb γ ₂ = Bp ₂ −Bb β ₃ = Gp ₃ −Gb γ ₃ = Bp ₃ −Bb. The three primary colors obtained by imaging the object X with the color television camera 1 by the multipliers 4 ₁ to 4 ₃ , 5 _{1 to} 5 ₃ , 6 _{1 to} 6 ₃ using the respective coefficients thus obtained in advance. The signals are multiplied, and the outputs of these multipliers 4 ₁ to 4 ₃ , 5 _{1 to} 5 ₃ , 6 _{1 to} 6 ₃ are added to adders 7 respectively.
Add in ~ 9. Here, the three primary color components of the imaging data obtained by imaging the object X with the color television camera 1 are (R, G,
B), the outputs O _{1 to} O ₃ of the adders 7 to 9 are O ₁ = α ₁ R + β ₁ G + γ ₁ BO ₂ = α ₂ R + β ₂ G + γ ₂ BO ₃ = α ₃ R + β ₃ G + γ ₃ B. Outputs O _{1 to} O ₃ of the adders 7 to 9 are converted into data showing a correlation with specific colors 1 to ₃ by a look-up table (hereinafter referred to as LUT) 10.

For simplification, an example of how to determine the value indicating the degree of correlation in the LUT 10 will be described with reference to FIG. 3 for a two-dimensional case. The horizontal axis l ₁ in FIG. 3 indicates the direction of the vector of the difference between the reference color and the specific color 1, and the vertical axis l ₂ indicates the direction of the vector of the difference between the reference color and the specific color 2. Then, the above-mentioned product-sum operation is performed on the color having a known relationship with the characteristic color in advance, and the product-sum operation result is plotted on the plane determined by the axes l ₁ and l ₂ , and the degree of correlation with each specific color 1 or 2 is plotted. Determine the area.
The region shown in FIG. 3 _{a 1 ~a 4, b 1 ~b} 4 respectively show a correlation of 100%, 90%, 70%, 50% of the area. Then, in order to facilitate subsequent processing, values are given as shown in the table below.

In addition, when it is outside the area, it is set to 0. For example, if the set of measured product-sum operation values is a point plotted with an X mark in FIG. 3, the output value of the LUT 10 is 7.

The set (O ₁ , O ₂ , O ₃ ) of product-sum operation values is input to the LUT 10 and converted into correlation data with the specific colors 1 to 3 as described above, and then stored in the memories 14 to 16. Comparing the threshold value indicated by the degree of correlation for each of the specific colors 1 to 3 and the above conversion output with the comparator 11 to
13 to obtain a portion having a correlation value of a certain degree or more for each of the specific colors 1 to 3, and convert this data into analog signals by the D / A converters 17 to 19 Output to displays 20-22. Thus, the image of the portion x ₁ ~x ₃ which extracts the specific color 1-3 In each display 20-22 can be displayed respectively to. The display 20 shown in FIG.
Shown next to each of the ~ 22 is the image.

Although the final processing result is only displayed on the display in the above description, an existing image processing device may be connected instead of the display to automatically measure the number or area of specific color regions. Well, in this case, it is an automatic visual inspection system that automatically performs visual inspection. Further, although the case where the number of the specific colors is three has been described in the present embodiment, the specific color can be accurately extracted in the same manner even when the number of the specific colors is three or more.

EFFECTS OF THE INVENTION As described above, according to the present invention, in the three primary color signals obtained by imaging the recognition target object with the color television camera, the three primary color components of the reference color and the three primary color components of the specific color to be extracted are respectively provided. By multiplying each primary color by a coefficient obtained from the difference between the primary colors and adding them, the three primary color signals are mapped to a vector space consisting of the vector of the difference between the reference vector and the specific color vector in the three primary color space, and this product-sum operation is performed. By performing at least three or more specific colors, an arbitrary point in the three-primary-color vector space is uniquely represented by a set of product-sum operation values, and the maximum difference between the reference color and the specific color is represented. The colors can be discriminated in and even subtle differences in color can be discriminated.

Further, the recognition target object is obtained by imaging with a color television camera according to the degree of correlation with the specific color of the set of the product-sum calculation obtained by performing the above-mentioned product-sum calculation for the color whose relationship with the specific color is already known. By converting the set of sum-of-products calculated values of the three primary color signals into the correlation data with the characteristic color, it is possible to express how close the picked-up color is to the specific color, and it is possible to perform identification with higher accuracy.

[Brief description of drawings]

FIG. 1 is a block diagram of an apparatus using the specific color extracting method of the present invention, FIG. 2 is an explanatory view of the specific color extracting method of the present invention, and FIG. 3 is a correlation degree data of a product sum operation value with a specific color. FIG. 4 is an explanatory diagram showing an example of how to determine the degree of correlation in the case of conversion into, and FIG. 1 is a TV camera, 4 ₁ to 4 ₃ , 5 _{1 to} 5 ₃ , 6 _{1 to} 6 ₃ are multipliers, 7
~ 9 is an adder, 10 is a lookup table, X is an object,
x _{1 to} x ₂ are specific color portions.

Claims (2)

(57) [Claims] 1. A coefficient obtained from a difference between a three-primary-color component of a reference color and a three-primary-color component of a specific color to be extracted is added to each of the three-primary-color signals obtained by imaging a recognition target object with a color television camera. A specific color extraction method in which each product is multiplied and then added, this product-sum operation is performed on at least three or more specific colors, and the specific colors are extracted using the obtained product-sum operation values as a set. 2. A coefficient obtained from the difference between the three primary color components of a reference color and the three primary color components of a specific color to be extracted is added to each primary color of a three primary color signal obtained by imaging a recognition target object with a color television camera. Each product is multiplied and then added, and this product-sum operation is performed on at least three or more specific colors, and the product-sum operation is performed on the color whose relationship with the specific color is known in advance. According to the degree of correlation with a specific color, a set of product-sum operation values of three primary color signals obtained by imaging a recognition target object with a color television camera is converted into correlation data with the specific color to extract the specific color. Specific color extraction method.