CN104165696A - Material surface color feature on-line automatic detection method - Google Patents

Abstract

The invention relates to an online automatic detection method for material surface color characteristics, belonging to the technical field of automatic online detection of material surface quality. First, select the material standard sample according to the real-time environment and material type, and take several original images of the surface of the standard sample of the material to be tested by using several high-precision cameras that can cover the entire surface of the material standard sample to be tested, and obtain the standard color after image processing Classification threshold; the material to be detected is taken by several high-precision cameras that can cover the entire surface of the material to be detected, and several original images of the surface of the material to be detected are taken. After image processing, the standard color classification threshold obtained is used to calculate the material to be detected Detect the surface color characteristics of the material; compare the surface color characteristics of the material to be detected with the evaluation standard, and obtain the surface color quality rating of the online automatic detection of the material to be detected. The detection method has high application value, and the method is simple and feasible.

Description

A method for online automatic detection of material surface color characteristics

technical field

The invention relates to an online automatic detection method for material surface color characteristics, belonging to the technical field of automatic online detection of material surface quality.

Background technique

Materials are the material basis for human survival and development. In the 1970s, people regarded information, materials and energy as the three pillars of contemporary civilization. The new technology revolution represented by the high-tech group in the 1980s also listed new materials, information technology and biotechnology as important symbols of the new technology revolution. This is mainly because materials are closely related to national economic construction, national defense construction and people's lives. With the improvement of people's living standards, people's requirements for material quality have also increased to an unprecedented height.

Due to the complexity of the production process of most materials, the unevenness of the surface color of the material is unavoidable. Although there are many performance indicators of materials, surface quality indicators are the most important category, mainly including surface color, surface defects and size. At present, most of the online surface quality inspections of materials are still in the stage of manual visual classification. The materials are fed by the conveyor belt of the production line, and workers compare them with standard materials, and then classify them according to their own judgment results. This detection method is not only labor-intensive, but also has low detection accuracy and efficiency, and the detection results are easily affected by subjective factors such as the technical quality, experience, human eye resolution ability and visual fatigue of the inspectors, lack of accuracy and standardization, and are prone to missed detection and The phenomenon of inaccurate determination of the quality level, this backward detection method can no longer meet the requirements of the development of the modern material industry. Therefore, it is necessary to realize automatic online detection of material surface quality.

For material manufacturers, the same type of material generally has several "color numbers". The ingredients and texture of different color numbers are almost the same, but the color is visually different. This is mainly due to fluctuations in the process and furnace temperature. There are many reasons. Since this difference is very small, it is required that the detection system can accurately distinguish this difference, and then classify the material surface quality according to these differences, that is to say, the classification algorithm should be quite accurate. At the same time, because the detection system will be applied to the on-line detection in the industrial field, the algorithm is required to be real-time. Therefore, the accuracy and real-time performance have become the main goals of the research.

In summary, the study of high-precision, high-efficiency, and stable online automatic detection methods and systems for material surface color characteristics has strong practical significance for saving labor, reducing labor intensity, and improving detection efficiency and consistency of detection results.

Contents of the invention

Aiming at the problems and deficiencies in the above-mentioned prior art, the present invention provides an online automatic detection method for material surface color characteristics. The detection method has high application value, and the method is simple and easy to implement. The present invention is realized through the following technical solutions.

An online automatic detection method for material surface color characteristics, the specific steps are as follows:

(1) First select the material standard sample according to the real-time environment and material type, and take several original images of the surface of the standard sample of the material to be tested by using several high-precision cameras that can cover the entire surface of the material standard sample to be tested, and after image processing Obtain the standard color classification threshold;

(2) The material to be tested is taken by several high-precision cameras that can cover the entire surface of the material to be tested, and several original images of the surface of the material to be tested are taken. After image processing, the standard color classification obtained in step (1) is used Threshold calculation of the surface color characteristics of the material to be detected;

(3) Comparing the surface color characteristics of the material to be tested obtained in step (2) with the evaluation standard, and obtaining the surface color quality rating of the material to be tested for online automatic detection.

The step of obtaining the standard color classification threshold in the step (1) is:

1.1 Read the surface original images of several standard samples of the material to be tested into matlab sequentially using the matlab function imread;

1.2 Then use the matlab function rgb2gray to convert each original surface image in step 1.1 from a color image to a grayscale image;

1.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 1.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending pixel points of each image;

1.4 Segment the graph of ascending pixels according to the number of curve turning points in the graph of ascending pixels in the main color category in the standard sample and step 1.3, and the number of segments corresponds to the main colors of the material in turn;

1.5 Use the matlab function plot to fit the curve corresponding to the number of segments obtained in step 1.4, and obtain the tangent at the midpoint of each curve, and the intersection point of the tangents of adjacent curves is the threshold for the color division of the two sections, A set of thresholds can be obtained for each original image, and the thresholds in the group are arranged in ascending order;

1.6 Process all surface original images through steps 1.1 to 1.5 and calculate the average value of the threshold within the group.

The steps of calculating the surface color characteristics of the material to be detected in the step (2) are:

2.1 Take several original images of the surface of the material to be tested and read them into matlab sequentially using the matlab function imread;

2.2 Then use the matlab function rgb2gray to convert each original surface image in step 2.1 from a color image to a grayscale image;

2.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 2.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending pixel points of each image;

2.4 Divide the graph of the ascending pixel points obtained in step 2.3 into curves of the corresponding number of segments according to the average threshold value in the group obtained in step 1.6, and obtain the corresponding pixel sum of the color. Each picture can obtain a group of main color correspondences of pixels and;

2.5 Process all the original surface images through steps 2.1 to 2.4 and calculate the corresponding pixels and average value of the surface color of the material to be detected.

In the step (3), the surface color quality rating of the online automatic detection of the material to be tested is as follows: firstly, the corresponding pixels and the average value of the surface color of the material to be tested obtained in step 2.5 are divided by the total pixels of the material to be tested respectively Obtain the pixel specific gravity value of the main color; then compare the pixel specific gravity value with the evaluation standard of the material to be tested, and then the grade of the surface color characteristic of the material can be obtained.

As shown in Figure 1, the above-mentioned online automatic detection device for surface color characteristics of materials includes a computer 1, material production control equipment 2, material production equipment 3, material delivery equipment 4 and high-precision camera 5 connected in sequence.

The beneficial effects of the present invention are: (1) Solve the shortcomings of the current material enterprises in mass production due to slow manual detection, high labor intensity, harsh environment, and subjective factors that have a greater impact on the results; (2) the present invention The method and system for real-time detection of material surface color characteristics are simple and easy, and can accurately, quickly, reliably, and real-time detect material surface color characteristics online, and quickly provide material surface quality evaluation; (3) the present invention is simple and has relatively Wide application prospects and higher economic benefits can be obtained; (4) The detection method and system of the present invention can detect the surface quality of most types of materials in the market, and can set the quality level for each quality level according to the manufacturer's own standards. The maximum limit for each color, and these limits can be adjusted online by professionals as needed.

Description of drawings

Fig. 1 is the structural representation in the automatic detection process of the present invention;

Fig. 2 is an example of the original image of material A of Embodiment 1 of the present invention;

Fig. 3 is an example diagram of ascending order arrangement of pixel gray value of A material standard sample according to Embodiment 1 of the present invention;

Fig. 4 is a fitting example diagram of the first segment of the pixel point of the A material standard sample in Example 1 of the present invention, wherein the fitting degree is 0.999;

Fig. 5 is a fitting example diagram of the second segment of the pixel point of the material A standard sample in Example 1 of the present invention, wherein the fitting degree is 0.992;

Fig. 6 is a fitting example diagram of the third paragraph of the pixel point of the material A standard sample in Example 1 of the present invention, wherein the fitting degree is 0.988;

Fig. 7 is a fitting example diagram of the fourth paragraph of the pixel point of the material A standard sample in Example 1 of the present invention, wherein the fitting degree is 0.991;

Fig. 8 is an example diagram of the gray value threshold of the surface color division of material A according to Embodiment 1 of the present invention;

Fig. 9 is an example of the original image of B material in Embodiment 2 of the present invention;

Fig. 10 is an example diagram of the ascending order arrangement of pixel gray value of B material standard sample according to Embodiment 2 of the present invention;

Fig. 11 is a fitting example diagram of the first section of the pixel points of the B material standard sample in Example 2 of the present invention, wherein the fitting degree is 0.999;

Fig. 12 is a fitting example diagram of the second paragraph of the pixel points of the material B standard sample in Example 2 of the present invention, wherein the fitting degree is 0.999;

Fig. 13 is a fitting example diagram of the third section of the pixel point of the material B standard sample in Example 2 of the present invention, wherein the fitting degree is 0.999;

Fig. 14 is a fitting example diagram of the fourth segment of the pixel points of the material B standard sample in Example 2 of the present invention, wherein the fitting degree is 0.997;

Fig. 15 is a fitting example diagram of the fifth segment of the pixel point of the material B standard sample in Example 2 of the present invention, wherein the fitting degree is 0.998;

Fig. 16 is an example diagram of the threshold value of the gray value of the surface color division of material B according to the second embodiment of the present invention.

In the figure: 1-computer, 2-material production control equipment, 3-material production equipment, 4-material transmission equipment, 5-high precision camera.

Detailed ways

The present invention will be further described below in combination with the accompanying drawings and specific embodiments.

Example 1

As shown in Figure 2, the A material has four colors: black, dark gray, light gray and white, wherein black and white are impurities, the less the content, the better. The evaluation criteria for the A material are: (1) Under normal circumstances (without impurities), the proportion of dark gray is 30%, the proportion of light gray is 70%, and the acceptable error is (-1%, 1%) If it meets this standard, it is qualified, otherwise it is unqualified; (2) If it contains impurities, the content of impurities (black and white) is less than 1%, and the proportion of black and white is less than 0.5%, otherwise it is unqualified, only the above two items When both are qualified, the surface quality of the material is qualified, otherwise it is unqualified.

The online automatic detection method of the surface color characteristics of the material, the specific steps are as follows:

(1) First select material standard samples according to the real-time environment and material types, and take 50 original surface images of material A standard samples to be tested by using 6 high-precision cameras that can cover the entire surface of material A standard samples to be tested. Obtain the standard color classification threshold after processing; the steps to obtain the standard color classification threshold are:

1.1 Read the surface original images of 50 standard samples of materials to be tested into matlab sequentially using the matlab function imread;

1.2 Then use the matlab function rgb2gray to convert each original surface image in step 1.1 from a color image to a grayscale image;

1.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 1.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending order pixels of each image, as shown in Figure 3;

1.4 Segment the curve of ascending pixels according to the number of curve turning points in the graph of ascending pixels obtained in the main color category in the standard sample and step 1.3, and the number of segments corresponds to the main colors of the material in turn. A material has four colors: black, dark gray, light gray and white. The curve in Figure 3 also corresponds to 3 turning points, so the curve is divided into four segments, and the main colors correspond to different curve segments, black and dark gray The pixel values corresponding to , light gray and white increase in turn;

1.5 Use the matlab function plot to fit the curve corresponding to the number of segments obtained in step 1.4, as shown in Figures 4 to 7, and find the tangent at the midpoint of each curve, and the intersection point of the tangents of adjacent curves is The thresholds divided by the two colors, each original image can obtain a set of thresholds, and the thresholds in the group are arranged in order from small to large;

1.6 Process all the surface original images through steps 1.1 to 1.5 and calculate the average threshold value within the group, as shown in Figure 8, the obtained average threshold value is [gray value 120; gray value 225; gray value 240], That is, the cut-off point of the gray value between black and dark gray is 120; the cut-off point of the gray value between dark gray and light gray is 225, and the cut-off point of the gray value between light gray and white is 240;

(2) Take 50 original images of the surface of the material to be tested by using 6 high-precision cameras that can cover the entire surface of the material to be tested, and after image processing, use the standard color classification obtained in step (1) The threshold value calculates the surface color feature of the material to be detected; the steps of calculating the surface color feature of the material to be detected are:

2.1 Take 50 original images of the surface of the material to be tested and read them into matlab sequentially using the matlab function imread;

2.2 Then use the matlab function rgb2gray to convert each original surface image in step 2.1 from a color image to a grayscale image;

2.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 2.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending pixel points of each image;

2.4 Divide the graph of the ascending pixel points obtained in step 2.3 into curves of the corresponding number of segments according to the average threshold value in the group obtained in step 1.6, and obtain the corresponding pixel sum of the color. Each picture can obtain a group of main colors The corresponding pixels and;

2.5 Process all the surface original images through steps 2.1 to 2.4 and obtain the corresponding pixels and average values of the surface color of the material to be tested. The pixels and average values obtained for the main colors of the material are shown in Table 1.

(3) Compare the surface color characteristics of the material to be tested obtained in step (2) with the evaluation standard to obtain the surface color quality rating of the online automatic detection of the material to be tested; the surface color quality rating of the online automatic detection of the material to be tested It is: first divide the corresponding pixel and the average value of the surface color of the material to be tested obtained in step 2.5 by the total pixels of the material to be tested to obtain the pixel specific gravity value of the main color; the pixel specific gravity value of the main color is shown in Table 1 , then as shown in Table 1, it can be seen that the proportion of black and white impurities is 0.61% and 0.77% respectively, both higher than 0.5%, so the impurities do not meet the quality standard and are unqualified; even if the proportion of light gray is 70.83%, It is within the error range of 70% (-1%, 1%), but the proportion of dark gray is 27.79%, which is not within the error range of 30% (-1%, 1%), so it does not meet the quality standard and is unqualified . In summary, the surface quality of the block A material is unqualified.

Table 1

Example 2

As shown in Figure 9, the B material is in five colors: black, dark gray, gray, light gray and white, and the evaluation standard for the B material is: the proportions of black, dark gray, gray, light gray and white are respectively : 6%, 24%, 6%, 61% and 3%, the acceptable error is (-0.5%, 0.5%), if this standard is met, it is qualified, otherwise it is unqualified.

The online automatic detection method for the surface color characteristics of the B material, its specific steps are as follows:

(1) First select material standard samples according to the real-time environment and material types, and take several original images of the surface of the standard samples of materials to be tested by using 6 high-precision cameras that can cover the entire surface of the material standard samples to be tested. After image processing Obtain the standard color classification threshold; the steps for the standard color classification threshold are:

1.1 Read the surface original images of 50 standard samples of materials to be tested into matlab sequentially using the matlab function imread;

1.2 Then use the matlab function rgb2gray to convert each original surface image in step 1.1 from a color image to a grayscale image;

1.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 1.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combining the graph of the ascending order pixels of each image, the graph is shown in Figure 10;

1.4 Segment the curve of ascending pixels according to the number of curve turning points in the graph of ascending pixels obtained in the main color category in the standard sample and step 1.3, and the number of segments corresponds to the main colors of the material in turn. Material B has five colors: black, dark gray, gray, light gray and white. The curves in Figure 10 are divided into the first, second, third, fourth and fifth corresponding to the above colors. part;

1.5 Use the matlab function plot to fit the curve corresponding to the number of segments obtained in step 1.4, as shown in Figures 11 to 15, and find the tangent at the midpoint of each curve, and the intersection point of the tangents of adjacent curves is The thresholds divided by the two colors, each original image can obtain a set of thresholds, and the thresholds in the group are arranged in order from small to large;

1.6 Process all the surface original images through steps 1.1 to 1.5 and calculate the average threshold value within the group, as shown in Figure 16, the obtained average threshold value is [gray value 100; gray value 130; gray value 190; gray value degree value 215], that is, the cut-off point of the gray value between black and dark gray colors is 100; the cut-off point of the gray value between dark gray and gray colors is 130; the gray value between gray and light gray colors The cut-off point of the value is 190, and the cut-off point of the gray value of light gray and white is 215;

(2) Take 50 original images of the surface of the material to be tested by using 6 high-precision cameras that can cover the entire surface of the material to be tested, and after image processing, use the standard color classification obtained in step (1) The threshold value calculates the surface color feature of the material to be detected; the steps of calculating the surface color feature of the material to be detected are:

2.1 Take several original images of the surface of the material to be tested and read them into matlab sequentially using the matlab function imread;

2.2 Then use the matlab function rgb2gray to convert each original surface image in step 2.1 from a color image to a grayscale image;

2.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 2.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending pixel points of each image;

2.4 Divide the graph of the ascending pixel points obtained in step 2.3 into curves of the corresponding number of segments according to the average threshold value in the group obtained in step 1.6, and obtain the corresponding pixel sum of the color. Each picture can obtain a group of main color correspondences of pixels and;

2.5 Process all the original surface images through steps 2.1 to 2.4 and calculate the corresponding pixels and average values of the surface color of the material to be tested. The corresponding pixels and average values of the surface color are shown in Table 2.

(3) Comparing the surface color characteristics of the material to be tested obtained in step (2) with the evaluation standard, and obtaining the surface color quality rating of the material to be tested for online automatic detection.

The surface color quality rating of the online automatic detection of the material to be tested is as follows: first, divide the corresponding pixels and the average value of the surface color of the material to be tested obtained in step 2.5 by the total pixels of the material to be tested to obtain the pixel proportion of the main color respectively Value, the pixel proportion value of the main color is shown in Table 2; then according to the proportion of black, dark gray, gray, light gray and white: 6.15%, 23.97%, 5.56%, 61.42% and 2.90%, each color The specific gravity is within the acceptable error range, so the surface quality of the block B material is qualified.

Table 2

Example 3

The online automatic detection method of the surface color characteristics of the material, the specific steps are as follows:

(1) First select material standard samples according to the real-time environment and material types, and take 100 original images of the surface of the material standard samples to be tested by using 10 high-precision cameras that can cover the entire surface of the material standard samples to be tested. After image processing Obtain the standard color classification threshold; the steps for the standard color classification threshold are:

1.1 Read the surface original images of 100 standard samples of the material to be tested into matlab sequentially using the matlab function imread;

1.2 Then use the matlab function rgb2gray to convert each original surface image in step 1.1 from a color image to a grayscale image;

1.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 1.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending pixel points of each image;

1.4 Segment the graph of ascending pixels according to the number of curve turning points in the graph of ascending pixels in the main color category in the standard sample and step 1.3, and the number of segments corresponds to the main colors of the material in turn;

1.5 Use the matlab function plot to fit the curve corresponding to the number of segments obtained in step 1.4, and obtain the tangent at the midpoint of each curve, and the intersection point of the tangents of adjacent curves is the threshold for the color division of the two sections, A set of thresholds can be obtained for each original image, and the thresholds in the group are arranged in ascending order;

1.6 Process all surface original images through steps 1.1 to 1.5 and calculate the average value of the threshold within the group.

(2) Take 100 original images of the surface of the material to be tested by using 10 high-precision cameras that can cover the entire surface of the material to be tested, and after image processing, use the standard color classification obtained in step (1) The threshold value calculates the surface color feature of the material to be detected; the steps of calculating the surface color feature of the material to be detected are:

2.1 Take 100 original images of the surface of the material to be tested and read them into matlab sequentially using the matlab function imread;

2.2 Then use the matlab function rgb2gray to convert each original surface image in step 2.1 from a color image to a grayscale image;

2.3 Use the matlab function sort to arrange the pixels of each grayscale image obtained in step 2.2 in ascending order, that is, arrange the pixels in ascending order of grayscale values; Combine the graph of the ascending pixel points of each image;

2.4 Divide the graph of the ascending pixel points obtained in step 2.3 into curves of the corresponding number of segments according to the average threshold value in the group obtained in step 1.6, and obtain the corresponding pixel sum of the color. Each picture can obtain a group of main color correspondences of pixels and;

2.5 Process all the original surface images through steps 2.1 to 2.4 and calculate the corresponding pixels and average value of the surface color of the material to be detected.

(3) Comparing the surface color characteristics of the material to be tested obtained in step (2) with the evaluation standard, and obtaining the surface color quality rating of the material to be tested for online automatic detection. The surface color quality rating of the online automatic detection of the material to be tested is as follows: first, divide the corresponding pixels and the average value of the surface color of the material to be tested obtained in step 2.5 by the total pixels of the material to be tested to obtain the pixel proportion of the main color respectively value; then compare the pixel specific gravity value with the evaluation standard of the material to be tested, and then the grade of the surface color characteristic of the material can be obtained.

Claims (4)

1. a material surface color characteristic online automatic detection method, is characterized in that concrete steps are as follows:

(1) first according to real time environment and material category chosen material master sample, by adopting some the high precision video cameras that can cover whole material standard sample surface to be detected, take the surperficial original image of some parts of material standard samples to be detected, after image is processed, obtain Standard Colors classification thresholds;

(2) material to be detected is passed through to adopt some the high precision video cameras that can cover whole material surface to be detected, take the surperficial original image of some parts of materials to be detected, after image is processed, then the Standard Colors classification thresholds that adopts step (1) to obtain calculates the surface color feature of this material to be detected;

(3) the surface color feature and the evaluation criterion that step (2) are obtained to this material to be detected contrast, and obtain the surface color quality grading of this material online automatic detection to be detected.

2. material surface color characteristic online automatic detection method according to claim 1, is characterized in that: the step that obtains Standard Colors classification thresholds in described step (1) is:

1.1 adopt matlab function imread to read in matlab the surperficial original image of the some parts of material standard samples to be detected of having taken successively;

Then 1.2 adopt matlab function rgb2gray that the surperficial original image in each step 1.1 is transformed into gray-scale map by cromogram;

1.3 adopt matlab function sort to arrange by ascending order its pixel each gray-scale map obtaining in step 1.2, arrange pixel according to gray-scale value order from small to large; And continue to adopt matlab function plot to simulate the curve map of the ascending order pixel of each image according to collating sequence point;

1.4 carry out segmentation according to the number of curve break in the curve map of the ascending order pixel obtaining in main color category and step 1.3 in master sample by the curve map of ascending order pixel, and the hop count of segmentation is corresponding with the main color of material successively;

1.5 curves that step 1.4 is obtained to corresponding hop count adopt matlab function plot to simulate corresponding curve map, and obtain the midpoint tangent line of every section of curve, the tangent line joining of adjacent curve is the threshold value that these two sections of colors are divided, each original image can obtain one group of threshold value, and in group, threshold value is arranged by order from small to large;

1.6 by all surfaces original image through step 1.1 to 1.5 processing and obtain group in threshold value mean value.

3. arbitrary material surface color characteristic online automatic detection method according to claim 2, is characterized in that: the step that described step (2) is calculated the surface color feature of this material to be detected is:

2.1 adopt matlab function imread to read in matlab the surperficial original image of taking some parts of materials to be detected successively;

Then 2.2 adopt matlab function rgb2gray that the surperficial original image in each step 2.1 is transformed into gray-scale map by cromogram;

2.3 adopt matlab function sort to arrange by ascending order its pixel each gray-scale map obtaining in step 2.2, arrange pixel according to gray-scale value order from small to large; And continue to adopt matlab function plot to simulate the curve map of the ascending order pixel of each image according to collating sequence point;

In the group that the curve map of the 2.4 ascending order pixels that step 2.3 is obtained obtains according to step 1.6, threshold value mean value is divided into the curve of corresponding hop count, obtain this color corresponding pixel and, each pictures can obtain pixel corresponding to one group of main color and;

2.5 by all surfaces original image through step 2.1 to 2.4 processing and obtain surface color respective pixel and the mean value of this material to be detected.

4. arbitrary material surface color characteristic online automatic detection method according to claim 3, is characterized in that: the surface color quality grading of this material online automatic detection to be detected of described step (3) be: the surface color respective pixel of this material to be detected of first step 2.5 being tried to achieve and mean value obtain respectively the pixel rate of specific gravity of main color divided by total pixel of this detected materials; Then compare according to pixel rate of specific gravity and this evaluation of material standard to be detected, can obtain the affiliated grade of this kind of material surface color characteristic.