JPH04308979A - Method for identifying color - Google Patents

Abstract

PURPOSE:To provide the color identifying method which can easily identify colors in a short time and can be practically used for a production line or a check line. CONSTITUTION:The luminance of an RGB component contained in the mark colors of defect marks M1-M3 is set beforehand as a judgement threshold value for binarizing the RGB component of image data, and the respective component of R, G and B in the image data of image signals supplied from a color camera are binarized by the respective judgement threshold values for the RGB components of the respective mark colors as mentioned above. Then, three pairs of binary data for the respective mark colors obtained as the result are ANDed and based on the AND, the respective mark colors are identified.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color identification method, and more particularly to a color identification method that allows a color to be detected to be easily and quickly identified.

0002

[Prior Art] In recent years, in order to automate work in product assembly and inspection, in addition to technology that automatically identifies the shape of parts and products, or the shapes of characters and marks marked on them, Development of technology is underway to automatically identify colors such as colors or marks marked on them as detection target colors. Conventionally, the method for identifying colors uses color image data obtained by capturing an image of a color with a color CCD camera and A/D converting the image signal consisting of RGB components, and various types of color image data are used for each pixel. Perform color distance conversion calculations and gradation calculations, determine the brightness, hue, and saturation for each pixel from the results of these calculations, identify the color for each pixel, and repeat the same calculations for all pixels to determine the color. A method of identifying was used.

0003

[Problems to be Solved by the Invention] The conventional method for identifying colors requires complex and enormous arithmetic processing for each pixel, and this arithmetic processing is performed on, for example, color image data of approximately 20,000 pixels. Since it takes a long time to identify colors, there is a problem in that conventional identification methods cannot be put to practical use on production lines or inspection lines. In addition, the software becomes complex and enormous, which is disadvantageous in terms of software cost.

An object of the present invention is to provide a method for identifying colors that can be easily and quickly identified and that can be put to practical use on production lines and inspection lines.

[0005]

[Means for Solving the Problems] A color identification method according to the present invention includes detecting one or more detection target colors using a color CCD.
A camera captures an image signal consisting of RGB components.
In a method for identifying a detection target color using /D-converted color image data, the three luminances of RGB components included in the image data of each detection target color are
It is a judgment threshold for binarizing each B component, and is set in advance as a judgment threshold that is set to "1" when it is equal to or higher than the judgment threshold. Each of the RGB components of the image data of the image signal is binarized using each determination threshold value of the RGB component of each detection target color, and the resulting three sets of binarization are obtained for each detection target color. The logical product of the data is calculated, and each detection target color is identified based on the logical product.

[0006]

[Operation] In the color identification method according to the present invention,
When identifying one or more detection target colors, first, each detection target color is imaged with a color CCD camera, and its RG
A determination threshold for converting the three luminances of RGB components included in color image data obtained by A/D conversion of an image signal consisting of B components into binarization of each of the RGB components of the image data. In the above cases, a determination threshold value of "1" is set in advance. Next, the RGB components of the image data of the image signal for each imaging screen supplied from the CCD camera are binarized using the respective determination thresholds for the RGB components of each detection target color. At this time, for example, if the image signal contains one detection target color out of a plurality of detection target colors, the RBG components of the image data of this image signal are set to the respective determination thresholds of the RGB components of each detection target color. Among the binarized data that has been binarized by value, RG has been binarized by each judgment threshold of the RGB components of the detection target color included in the image signal.
Only the portions of the B component binarized data that correspond to the detection target color are binarized as “1”, and the portions other than the portions corresponding to the detection target color are binarized as “1” or “0”. Ru. Next, the logical product of the three sets of binarized data for each detection target color that has been binarized using each determination threshold value of the RGB components of each detection target color is determined. At this time, in the AND of the three sets of binarized data that have been binarized using each judgment threshold for the detection target color included in the image signal, only the portion corresponding to the detection target color is "
1”. Next, each detection target color is identified based on the logical product for each detection target color.

In this way, the image signal of each detection target color is
The three luminances of the RGB components included in the /D-converted color image data are set in advance as judgment thresholds for binarizing the RGB components of the image data, and Binarize the RGB components of the image data of the image signal using each determination threshold value for each detection target color, and calculate the logical product of the three sets of binarized data for each detection target color obtained as a result, Since each detection target color is identified based on the logical product, the detection target color can be easily and quickly identified from image data, and this identification method can be put to practical use on production lines and inspection lines. . Further, the complicated and enormous calculation process for identifying the detection target color from the image data can be omitted, and the software for identifying the detection target color can be extremely simplified, so that the cost of the software can be reduced.

[0008]

Effects of the Invention According to the color identification method according to the present invention, as explained in the operation section above, the RG of each detection target color is
The three luminances of the B component are set in advance as judgment thresholds for binarizing the RGB components of the image data, and the RGB components of the image data of the image signal for each imaging screen supplied from the camera are set in advance. Perform binarization processing using each judgment threshold for each detection target color, calculate the logical product of the three sets of binarized data for each detection target color, and calculate each detection target color based on the logical product. , the color to be detected can be easily and quickly identified from image data, this identification method can be put to practical use on production lines and inspection lines, and the color to be detected can be identified from image data. Complex and enormous arithmetic processing for identification can be omitted, and the software for identifying the detection target color can be extremely simplified, resulting in effects such as a reduction in software costs.

[0009]

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings. In this embodiment, the present invention is applied to a method for identifying the color of a defect mark indicating a paint defect marked on a car body in a water polishing line of an automobile manufacturing factory. Note that the description will be made by defining front, rear, left, and right based on the front, rear, left, and right of the vehicle body.

The water polishing line L will be briefly explained. As shown in FIG. 1, the water polishing line L provided downstream of the intermediate coating line (not shown) includes an inspection marking station L1, a mark detection station L2, a polishing station L3, and a water washing station L4 from the upstream side. It is provided. Inspection marking station L
1, inspectors m1 and m2 are arranged, and the vehicle body B coated with black is transported to the inspection marking station L1, where the inspectors m1 and m2 visually inspect the painted surface of the vehicle body B. When inspectors m1 and m2 find paint defects such as pinholes or sagging on the paint surface, they will select the corresponding one of the three color-coded mark stamps that correspond to the polishing grade that should be applied to the paint defect area. Select and attach a defect mark M to the defective part of the coating. As shown in Table 1, areas with significant defects that require heavy polishing are marked with a yellow-green defect mark M1, areas that require medium polishing are marked with an orange defect mark M2, and areas that require light polishing are marked with a yellow-green defect mark M2. Yellow defect marks M3 are attached, and numerical values from "1" to "3" are assigned to each of these defect marks M as data values indicating the polishing grade. Note that the size of the defect mark M is approximately 10 mm in diameter.

[Table 1]

At the mark detection station L2, the vehicle body B
Mark detection devices 10 and 10A are provided to detect defective marks M placed on the left half of the vehicle body B, and the mark detection device 10A detects the defective marks M placed on the left half of the vehicle body B. A defective mark M placed on the right half is detected. The polishing station L3 includes a polishing device 50 for polishing defective parts of the paint.
・50A is provided, and the polishing device 50 is the mark detection device 1
The polishing device 50A receives the detection data from the mark detection device 10A and polishes the defective portion on the left half of the vehicle body B. It has become. The washing station L4 is provided with a plurality of washing showers 1 and washing brushes 2 and 3, and is designed to wash the car body B after polishing.

Next, the mark detection devices 10 and 10A provided at the mark detection station L2 will be explained. However, since the mark detection devices 10 and 10A have the same configuration, only the mark detection device 10 will be described. As shown in FIG. 2, the mark detection device 10 includes an articulated mark detection robot 11 and a hand 12 of the mark detection robot 11. an imaging device 20 for capturing an image, and a robot 11
The control device 30 is configured to drive and control the image pickup device 20 and to perform image processing on the image signal from the imaging device 20. The imaging device 20 includes a support member 21 attached to the hand 12 of the robot 11, and a light projector 22 installed at a predetermined angle on the support portion 21a of the support member 21 for projecting irradiation light onto the small inspection area S. , and a color CCD camera 23 that is installed at a predetermined angle on the support portion 21b and captures an image of the small inspection area S illuminated by the projector 22. The control device 30 includes a main control unit 31, a robot control unit 32, a floodlight control unit 33, and an image processing unit 34. When the vehicle body B is transported to a predetermined position of the mark detection station L2, the main control unit 31 A control signal is output to the projector control unit 33 to start the projector 22, and at the same time, a start command signal is output from the main control unit 31 to the robot control unit 32, and the robot control unit 32 operates based on a preset control program. The robot 11 is driven and controlled, and the imaging device 20 sequentially moves through the hand 12 of the robot 11 to predetermined positions where each small inspection area S can be imaged, and the CCD camera 23 images each small inspection area S. . The image signal of the small inspection area S captured by the CCD camera 23 is input to the image processing unit 34, where the defect mark M is detected from the image signal and its mark color (polishing grade) and position are memorized. is,
After the inspection of the vehicle body B is completed, the detection data of the mark colors and positions of all detected defective marks M is sent to the main control unit 3.
1, and the polishing device 5 is input from the main control unit 31.
0. In addition, the polishing device 50
50A each consists of a polishing robot and a control device.

Next, the image processing unit 34 will be explained. As shown in FIG. 3, the image processing unit 34
a signal processing unit 35 that performs analog processing such as amplification on an image signal composed of RGB components inputted from the CD camera 23;
An A/D conversion unit 36 that converts the RGB components of the image signal processed by the signal processing unit 35 into image data of 256 levels of digital values at 8-bit gradation brightness, and an image memory unit that stores the image data. 37, a program memory section 38 storing a control program for defective mark detection control to be described later, an input/output interface section 39 for inputting and outputting data to and from the main control unit 31, and a microprocessor section 40. Each part 35-40 is connected by a bus as shown.

Next, a description will be given of the mark color determination thresholds for each defective mark M, which are preset in the control program for the defective mark detection control. The judgment threshold is C
This threshold value is used to binarize the RGB components of the image data of the image signal supplied from the CD camera 23 as "1" when each is equal to or higher than the determination threshold value, and is determined experimentally. be. That is, each defect mark M is marked on the car body B, which is coated in black, and is imaged by the CCD camera 23.
R included in the image data obtained by A/D converting the image signal
The average value of the three brightness distributions of the GB components is determined for each defective mark M, and these are set as the mark color determination threshold for each defective mark M as shown in Table 1. The values are preset in the table as data.

Next, regarding defect mark detection control, FIG.
This will be explained based on the flowchart in FIG. However, S in the diagram
i (i=1, 2, . . . ) indicates each step. Note that this control is the same for the mark detection devices 10 and 10A, so the control in the mark detection device 10 will be explained. When the control starts, initialization such as clearing the counter and memory is performed (S1), and then the vehicle body B is transported to a predetermined position of the mark detection station L2, and the imaging device 20 is When moving to the imaging position of the small inspection area S to be imaged first and imaging becomes possible (S2
:Yes), then the small inspection area S is imaged by the CCD camera 23, and the image signal for one screen is input (S3). Next, the RGB components of the input image signal are A/D converted for each pixel (S4), and the image data of the A/D converted RGB components are stored in the image memory section 37 (S4).
5). That is, as shown in FIG. 6 as an example, if a yellow-green defect mark M1, an orange defect mark M2, and a yellow defect mark M3 are attached to the imaged small inspection area S, then the small inspection area S RGB of image data about
In the portion corresponding to the pixel of each defective mark M1 to M3 of the component, RGB of the mark color of the defective mark M1 to M3, respectively.
The brightness values of the components are included as image data. Next, the R component of the image data is read from the image memory unit 37 (S6), then a determination threshold value for the R component of the mark color of the defective mark M1 is set (S7), and then this determination threshold value is set (S7). The R component of the image data is binarized (S8
), and then the binarized data subjected to the binarization process is stored in the image memory section 37 (S9). In the binarization process of S7, the defect mark M1 portion of the R component of the image data is
As shown by the solid circle in Fig. 7, it is binarized at its own judgment threshold, so it is binarized as "1", and the R component of the defect mark M2 part is lower than the judgment threshold. Since it is large, it is similarly binarized as "1", and since the R component of the defect mark M3 part is smaller than the determination threshold, it is binarized as "0" as shown by the double-dashed line circle in FIG. Binarized. Incidentally, since the portions other than the defect marks M1 to M3 are black, they are binarized as "0".

Next, a counter is incremented by one each time the judgment thresholds corresponding to the three types of mark colors are set (S10), and then the value of the counter becomes "
3" (S11), and in this case, since it is No, the process moves to S7. In S7, the judgment threshold of the R component of the mark color of the defective mark M2 is set.
8 to S11 are executed, and then in S11 the value of the counter is determined. Since the result is No in this case as well, the process moves to S7, and in S7, a determination threshold value for the R component of the mark color of the defective mark M3 is set. Hereinafter, S8 to S11 as above
is executed. In S8, when the R component of the image data is binarized using the judgment threshold for the R component of the defect mark M2, as shown in FIG. , the R component is smaller than the determination threshold, so it is binarized as "0" and the defect mark M
The 2 part is binarized using its own judgment threshold, so it is ``1''.
", and the defect mark M3 is binarized as
Since the component is smaller than the determination threshold, it is binarized as "0". Further, in S8, when the R component of the image data is binarized using the determination threshold for the R component of the defect mark M3, the R component is determined for each of the defect marks M1 to M3 of the image data. Since the threshold value is exceeded, all of those parts are binarized as "1" as shown in FIG. 15. Next, in S11, it is determined whether the counter value is "3" or not, and since it is YES, the process moves to S12.

In S12, the G component of the image data is read from the image memory unit 37, and the G component of the defect marks M1 to M3 is determined based on the determination threshold of the G component of the defect marks M1 to M3.
Processes S13 to S17 similar to steps S7 to S11 are executed. As a result, binarized data as shown in FIGS. 8, 12, and 16 are obtained, and these binarized data are stored in the image memory section 37. Next, in S18, the B component of the image data is read from the image memory section 37, and the B component of the defect marks M1 to M3 is determined in steps S19 to S2, which are similar to S7 to S11, based on the determination threshold of the B component of the defect marks M1 to M3.
processing is executed. As a result, binarized data as shown in FIGS. 9, 13, and 17 are obtained, and these binarized data are stored in the image memory section 37.

Next, three sets of binarized data for each defect mark color are read from the image memory section 37 (S24). ), and then the logical product of the three sets of binarized data is calculated (S25). As a result, as shown in FIG. 10, the logical product of the three sets of binarized data (FIGS. 7 to 9) that have been binarized using the respective judgment thresholds for the RGB components of the defective mark M1 is Only the part becomes the image data of the defective mark M1 expressed as "1", and the RGB of the defective mark M2
As shown in FIG. 14, the logical product of three sets of binarized data (FIGS. 11 to 13) that have been binarized using each determination threshold value for each component shows that only the defect mark M2 is "1". The image data of the represented defect mark M2 is obtained, and three sets of 2
The logical product of the valued data (FIGS. 15 to 17) becomes image data of the defective mark M3 in which only the portion of the defective mark M3 is expressed as "1", as shown in FIG. 18.

Next, each image data obtained by calculating the logical product is subjected to labeling processing (S26), and then the centroid positions of the defect marks M1 to M3 are calculated for each image data subjected to the labeling processing. (S27)
, these center of gravity positions and the numerical values of the polishing data indicated by the defect marks M1 to M2 are stored in association with each other (S28). In this case, the center of gravity of each of the defect marks M1 to M3 is
x and y of the center of gravity for the x and y coordinate system set in the captured image
Coordinate values are calculated and stored. Next, it is determined whether mark detection has been completed for all the small inspection areas S (S29
), in this case, it has not finished, so the process returns to S2, and steps S2 to S29 are repeated for each small detection area S, and when mark detection is finished for all small detection areas S (S29: Y
es), as shown in Table 2, the detection data of the defect mark M stored in S28 is transferred to the control device of the polishing apparatus 50 (S30). Note that mark detection by the mark detection device 10A is also performed in parallel with mark detection by the mark detection device 10, and the mark detection device 10A performs mark detection by the polishing device 50.
The detected data is similarly transferred to A, and the paint defective portion of the vehicle body B is polished by the polishing devices 50 and 50A based on these detected data.

[Table 2]

In this way, R contained in the color image data obtained by A/D converting the image signals of each defect mark M1 to M3.
The three luminances of the GB components are set in advance as judgment thresholds for binarizing the RGB components of the image data, respectively, and the RGB of the image data of each image signal supplied from the camera 23 is set in advance.
The components are binarized using each determination threshold value for each defect mark M1 to M3, and three sets of 2 for each defect mark color are obtained as a result.
Since the logical product of the value data is calculated and each defective mark M1 to M3 is identified based on the logical product, the defective mark M can be easily and quickly identified from the image data. Moreover, the complicated and enormous calculation process for identifying the defective mark M from image data can be omitted, and the software for identifying the defective mark M can be extremely simplified, so that the cost of the software can be reduced. Note that the polishing grade is not limited to three stages, but may be set to two or more various stages, and in that case, the mark color of the defect mark M may be differentiated according to the polishing grade stage. Further, as the mark color of the defect mark M, it is also possible to use various colors other than the above-mentioned mark color. Further, the mark color judgment threshold of the defect mark M is not limited to the above value, but may be set in various ways within the range of the luminance distribution of the RGB components of each mark color, adapting to the imaging conditions of the small inspection area S. good. It should be noted that the color identification method described above is not limited to the identification of defect marks of paint defects, but can of course be applied to the identification of the colors of various parts and products or the colors of marks marked on parts and products. .

[Brief explanation of drawings]

FIG. 1 is a plan view of a water polishing line.

FIG. 2 is an overall configuration diagram of a mark detection device.

FIG. 3 is a block diagram of an image processing unit.

FIG. 4 is a part of a flowchart of a mark detection control routine.

FIG. 5 is a part of a flowchart of a mark detection control routine.

FIG. 6 is an explanatory diagram showing an example of an image of a small inspection area.

FIG. 7 is an explanatory diagram of binarized data obtained by binarizing the R component of image data using a determination threshold value for the R component of a yellow-green mark color.

FIG. 8 is an explanatory diagram of binarized data obtained by binarizing the G component of image data using a determination threshold value for the G component of a yellow-green mark color.

FIG. 9 is an explanatory diagram of binarized data obtained by binarizing the B component of image data using a determination threshold value for the B component of a yellow-green mark color.

FIG. 10 is an explanatory diagram of logical product of the binarized data shown in FIGS. 7 to 9;

FIG. 11 is an explanatory diagram of binarized data obtained by binarizing the R component of image data using a determination threshold value for the R component of an orange mark color.

FIG. 12 is an explanatory diagram of binarized data obtained by binarizing the G component of image data using a determination threshold for the G component of the orange mark color.

FIG. 13 is an explanatory diagram of binarized data obtained by binarizing the B component of image data using a determination threshold for the B component of an orange mark color.

FIG. 14 is an explanatory diagram of logical product of the binarized data shown in FIGS. 11 to 13;

FIG. 15 is an explanatory diagram of binarized data obtained by binarizing the R component of image data using the determination threshold value of the R component of the yellow mark color.

FIG. 16 is an explanatory diagram of binarized data obtained by binarizing the G component of image data using a determination threshold value for the G component of the yellow mark color.

FIG. 17 is an explanatory diagram of binarized data obtained by binarizing the B component of image data using a determination threshold value for the B component of a yellow mark color.

18 is an explanatory diagram of logical product of the binarized data shown in FIGS. 15 to 17. FIG.

[Explanation of symbols]

M1~M3 Defect mark 10/10A mark detection device 23 CCD camera 34 Image processing unit

Claims (1)

[Claims] [Claim 1] One or more detection target colors are color C.
In a method for identifying a detection target color using color image data obtained by capturing an image with a CD camera and A/D converting an image signal consisting of RGB components, three of the RGB components included in the image data of each detection target color are used. The brightness is set in advance as a determination threshold value for binarizing each of the RGB components of image data, which is set to "1" when the brightness is equal to or higher than the determination threshold value, and the luminance is supplied from the camera. The RGB components of the image data of the image signal for each imaging screen are binarized using the determination thresholds for the RGB components of each detection target color, and the resulting A color identification method, characterized in that the logical product of three sets of binary data is calculated, and each detection target color is identified based on the logical product.