JPH05165953A - Color identification method - Google Patents

Abstract

(57) [Abstract] [Purpose] To enable accurate color identification by RGB signals from an image recognition device. [Structure] An identification object is painted in an appropriate color, the plurality of painted identification objects are read by an image processing device, and the color composition of the identification object is input to the fuzzy controller as RGB signals. It is judged whether it is color or not, and it is inputted to the voice input device, then the words inputted to the voice input device are coded and inputted to the fuzzy controller, and the RGB intensity and the language code inputted to the fuzzy controller are stored. Data is generated as a color identification file to determine an identification object.

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 for identifying colors learned in advance by fuzzy inference.

[0002]

2. Description of the Related Art In the conventional color identification method, the RGB values of the object are registered in advance, and when the color of the object is identified by the R, G, B color data input from the image recognition device, the image recognition is performed. The data input from the device and the color data registered in advance are matched with each other to determine whether or not the object.

[0003]

The conventional color identification method is configured to identify the object color by identifying the color of the object from the RGB values of the object registered in advance as described above, and the RGB value from the image recognition device is Since it slightly changes depending on the illuminance near the object and the angle of the camera, it does not always have the same value as the registered color data, but rather has a different value each time. Therefore, the RGB values registered in advance do not substantially match the input data from the image recognition device, and the RGB values of each object
The need arises for the values to have an acceptable range. Therefore, there is a problem that the color identification cannot be determined when there is an object whose RGB values are registered in advance.

The present invention has been made in order to solve the above problems, and an object of the present invention is to obtain a color identification method capable of performing accurate color identification by an RGB signal from an image recognition device.

[0005]

According to the color identification method of the present invention, a plurality of identification objects painted in appropriate colors are read by an image processing apparatus, and the fuzzy controller inputs the color composition of the identification objects as RGB signals. , The color of the identification object is input to the voice input device, the words input to the voice input device are encoded and input to the fuzzy controller, and the fuzzy controller inputs the RGB intensity. And the language code is stored to generate data as a color identification file, and fuzzy labels and membership functions of R, G, and B are generated from the generated data to determine the identification object.

[0006]

The color identification method according to the present invention is equipped with a learning function for colors, and a designated color to be inputted is inputted as a language code, and fuzzy matching is performed to identify the object of the designated color among a plurality of objects. The color recognition is performed by.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a flow chart showing an embodiment of the present invention. The outline of this system is to perform "image understanding" and "object recognition by fuzzy matching" as a combination thereof.
Use unique identifications painted in appropriate colors.

In this embodiment, a voice input device is used for color designation. First, the size and color of the identification object are designated by voice (step ST1), and a plurality of sizes and colors can be designated. Finally, the designation end is input (step ST2). In the voice output response to this, in order to confirm the instruction by the voice input, "Yes, I understand. What size, what color, what size, what color do you carry?" It is determined whether the voice output response matches the voice input content while repeating all the input of the input (step ST
3). Then, if the voice output response matches the voice input content, the operator inputs that they match in step ST4. If it is wrong, the fact that it is wrong is input in step ST5, and all the identification objects are input again. This is repeated until they match. After that, if there is no specified identification object, for example, a servomotor, "What size is what color, What size is what color,
I can't find any distinguishing objects such as colors in any size. Please instruct again. Then, the processing for the purpose of re-instruction is performed (step ST6) and the instruction is received again. In this case, the instruction receives all the identification object instructions from the beginning again (all the previous instructions are canceled), and the confirmation operation is repeated from the beginning.

By carrying out such a processing, the contents instructed by the operator are correctly recognized, and when all of them are present, the conveyance start is output and the conveyance is started (step ST7). In this case, the shape of the identification object can be determined in advance, but it is difficult to prepare the fuzzy inference rule in advance because the color of the identification object is appropriately painted by spraying or the like. Therefore, this system will be equipped with a learning function for colors.

As shown in FIG. 2, the method of learning the paint color is as follows. First, a discriminating object is identified by an image processing device (Meiscop).
The reading is performed in step e) (step ST8), and the fuzzy controller is made to input the color composition of the identification object as an RGB signal (step ST9). Next, in step ST10, the narrator judges the color by looking at the identification object and inputs it to the voice input device. Then, the voice input device codes the word and inputs it to the fuzzy controller (step ST11). In step ST12, the fuzzy controller stores the language code including the pair of data input in steps ST9 and ST11, that is, the RGB intensity and the language code. In step ST13, steps ST8 to ST
Repeat the operations up to 12 to create a color identification file as shown in Table 1.
The data shown in is generated.

[0011]

[Table 1]

Next, in step ST14, step S
Fuzzy labels and membership functions (hereinafter abbreviated as MSF) of R, G, and B are generated from the data generated in T13 by the following method. Now, step S
When the intensity of the color C generated in T13 is rearranged in ascending order to be Cs ₁ , Cs ₂ , Cs ₃ , Cs ₄ , ..., Cs _n , the MSF is determined as shown in FIG. The fuzzy labels are Cs ₁ , Cs ₂ , Cs ₃ , Cs ₄ , ..., C
_Let s _n . In this way, the paint color is learned.

Next, a method of identifying an identification object by voice instruction performed by size identification and color identification will be described. Since there are only two sizes in this system, the sizes are identified by comparison with the threshold value instead of fuzzy inference.
In addition, color identification is performed by fuzzy inference. That is, the color information input from the voice input device is input as the language code (1 to N). Now, in the method of identifying an identification object when the input language code is i, first, a fuzzy rule with the MSF of RGB color intensities corresponding to the language code i as the AND condition is generated as follows.

If the fuzzy label Rs is Rs ₁ and G
If s is Gs ₁ and Bs is Bs ₁ , then color C is i. However, C is a variable meaning a color, and i is a color name corresponding to a language code. Next, the color information of the identification object is input as RGB intensities from the image processing apparatus. Now, the color information of each inputted identification object is represented by Rs _j , Gs _j , Bs _j (j = 1.
.About.n), the degree of establishment of the fuzzy rule (.mu.j, j = 1 to n) based on the color information from each conveyed product is calculated, and .mu.
The maximum μj that satisfies j ≧ λ is obtained, and the j-th identification object is selected. However, λ is a threshold value for color determination. A color matching the color input by voice input is searched from among the colors of the identification objects determined in this way. Then, the one that matches both the size identification and the color identification is determined as an identification object to be mounted on the automatic guided vehicle (AGV), voice output, and the transfer operation is started by a start command. If there is no match for both, output voice to receive instructions of the conveyed item again,
The operation of deciding the identification that matches both the size identification and the color identification by voice input as the identification object to be placed on the AGV is repeated.

[0015]

As described above, according to the present invention, a plurality of identification objects painted in appropriate colors are read by the image processing device,
The fuzzy controller inputs the color composition of the identification object as an RGB signal, determines the color of the identification object and inputs the color composition to the voice input device, and encodes the words input to the voice input device to generate the fuzzy logic. This fuzzy controller inputs R
The GB intensity and language code are stored to generate data as a color identification file, and fuzzy labels and membership functions of R, G, and B are generated from the generated data to determine the identification object. An effect is that accurate color identification can be performed by the RGB signal from the image recognition device.

[Brief description of drawings]

FIG. 1 is a flowchart showing an embodiment of a color identification method according to the present invention.

FIG. 2 is a flowchart showing a method for learning a paint color.

FIG. 3 is an explanatory diagram showing a membership function determination method.

Claims (1)

[Claims] 1. A plurality of identification objects painted in appropriate colors are read by an image processing apparatus and a fuzzy controller is made to input the color composition of the identification objects as RGB signals, and the color of the identification objects is judged. The fuzzy controller inputs the data to the voice input device, codes the words input to the voice input device, and inputs the words to the fuzzy controller. The fuzzy controller stores the input RGB intensity and language code and stores the data as a color identification file. A color identification method for generating a fuzzy label of R, G, and B and a membership function from the generated data, and determining the identification object by voice input.