A visual inspection method for belt deviation that eliminates straight line interference

Abstract

The present application belongs to the technical field of visual detection of belt deviation, and specifically provides a visual detection method for belt deviation that can eliminate straight line interference, including: obtaining multiple straight lines based on the Hough straight line detection method, and adopting an area calculation method to eliminate the interference of the coal straight line on the belt when identifying the belt edge straight line, accurately lock the belt edge straight line, accurately identify the roller boundary through the machine learning clustering algorithm, and calculate the exposed area of ​​the roller. If the area changes, it is determined that the belt is deviating, and timely alarm is issued if there is belt deviation during the operation of the belt conveyor, so as to improve the reliability of its safe operation. At the same time, it can also reduce the loss of equipment and accessories, improve the service life of the equipment, and increase its transportation capacity.

Description

Technical Field

[0001] The present application belongs to the technical field of visual detection of belt deviation, and specifically relates to a visual detection method for belt deviation that can eliminate linear interference. Background Art

[0002] Belt conveyors are widely used in the coal industry due to their unique advantages. However, belt deviation often occurs during use, causing the belt to be squeezed and worn, shortening its service life. When the belt deviates seriously, it can also cause serious accidents such as material rollover and belt tearing, which not only affects production efficiency but also poses a safety hazard. It will cause huge economic losses to the company and even cause casualties in severe cases. Therefore, timely and accurate detection of belt deviation and taking effective measures are of great significance to safe production.

[0003] Currently, the mainstream belt deviation detection methods at home and abroad are divided into hardware detection and software detection. Hardware detection typically uses ultrasonic distance detection to obtain data on belt speed and belt deviation, but hardware detection methods are too expensive to implement on a large scale. Software detection works by placing a fixed camera on the belt to monitor the belt status. Currently, this is mainly detected through camera detection, which has large errors. This is especially true when there is a large amount of coal flowing on the belt. Since coal mining and transportation often occur underground in dim light and the coal is dark in color, the coal flow on the belt will interfere with and affect linear detection, making high-precision detection difficult.

[0004] In summary, this field urgently needs a new type of visual detection method for belt linear deviation, which can automatically eliminate the linear interference of coal flow on the belt, accurately lock the straight line of the belt edge, and promptly alarm when the belt deviates during the operation of the belt conveyor, so as to improve the reliability of its safe operation. At the same time, it can also reduce the loss of equipment and accessories, increase the service life of the equipment, and increase its transportation capacity. Summary of the Invention

[0005] In view of this, the present application provides a method for visually detecting belt deviation that can eliminate linear interference, effectively solving or at least alleviating one or more of the above-mentioned problems and other problems existing in the prior art.

[0006] In the first aspect, the present application provides a visual detection method for belt deviation that can eliminate straight line interference, which is characterized in that the detection method includes the following steps: Step 1, capturing a belt image through a camera and selecting an image acquisition area; Step 2, using the Hough line detection method to obtain multiple straight lines in the selected image acquisition area, the multiple straight lines include the edges on both sides of the belt and the material interference straight lines, taking any two straight lines m and n from the multiple straight lines, and setting the coordinates of the two end points of the straight line m to Set the coordinates of the two endpoints of line n to According to the formula:

[0007] Calculate the area s enclosed by any two straight lines m and n. When the area s is the maximum value, extract the next two straight lines as the two side edge lines of the belt; Step 3, use the extracted two side edge lines to calculate the high brightness area of ​​each side edge respectively, and determine it as the side roller area; Step 4, extract a series of feature points in the side roller area, perform clustering processing through machine learning to obtain the boundary points of the side roller area, and call the library function to calculate the side roller area S based on the boundary points of the side roller area 托辊 ; Step 5, when S 托辊 If the area value changes by more than 10%, it is determined that the belt is deviating. Step 6: Based on the belt deviation information obtained in step 5, a deviation warning is issued and the belt is adjusted through the control end.

[0008] In the preferred technical solution of the above test method, any two straight lines m and n in step 2 may cross or not cross, and the line connecting the endpoints of the two straight lines is The area of ​​the quadrilateral formed by the two straight lines m and n is the area s enclosed by the straight lines m and n.

[0009] In the preferred technical solution of the above-mentioned test method, step 3 calculates the average grayscale value of all pixels in the image acquisition area, compares the collected grayscale value of each side edge area with the average grayscale value of all pixels, and obtains the area where the grayscale value of each side edge area is higher than the average grayscale value of all pixels by 20%, which is regarded as the high-brightness area of ​​each side edge and is determined to be the roller area of ​​that side.

[0010] In the preferred technical solution of the above test method, step 4 extracts a series of characteristic points σ in the side roller area. i , where i = 1, 2, 3…n, n < 20, 4*4 pixels are selected with each feature point as the center to form a 1*16 column vector, and the feature points whose grayscale values ​​of 4 pixels out of 16 pixels are higher than 20% of the grayscale values ​​of the remaining 12 pixels are screened out through the machine learning clustering algorithm, and the screened feature points are sorted from large to small according to the grayscale value difference of the pixels inside each feature point, and the first four feature points are selected to be defined as the boundary points of the roller area on this side.

[0011] In the preferred technical solution of the above test method, step 4 calls the area library function to calculate the area of ​​the quadrilateral surrounded by the four boundary points, thereby obtaining the area S of the side roller. 托辊 .

[0012] In the preferred technical solution of the above test method, step 4 only requires calculating the area of ​​the roller on one side.

[0013] In the preferred technical solution of the above test method, in step 5, when the area of ​​the S roller decreases by more than 10%, it is determined that the belt is deviating toward that side edge; when the area of ​​the S roller increases by more than 10%, it is determined that the belt is deviating toward the other side edge.

[0014] In the preferred technical solution of the above test method, the material in step 2 is coal.

[0015] The present application provides a method for visually detecting belt deviation that can eliminate straight line interference. The method uses a camera to capture an image of the belt working area, selects an image capture area, determines whether the capture is successful, and extracts the belt edge straight line after the capture is successful. The Hough line detection method is used to capture straight lines in the selected image capture area. The multiple straight lines captured include the edge straight lines on both sides of the belt and the material interference straight lines. In order to eliminate the interference of the material interference straight lines on the belt on the belt edge straight line detection, the present application proposes an area calculation method: take any two straight lines m and n from the multiple straight lines, set the coordinates of the two end points of the straight line m to be The coordinates of the two endpoints of line n are Lines m and n are any two of all the straight lines on the belt obtained by the Hough line detection method. They can be crossed or not. The endpoints of the two straight lines are connected. And two straight lines form a quadrilateral, according to the formula: The area s of the quadrilateral is calculated. When the area s reaches its maximum value, the next two straight lines are extracted as the two side edge lines of the belt. According to the area calculation method proposed in this application, when identifying the belt edge line, the interference of the coal line on the belt can be eliminated, the belt edge line can be accurately locked, and the detection accuracy can be improved.

[0016] After extracting the straight lines on both sides of the belt edge, the brightness value of each side edge area is collected, and the mean gray value of all pixels in the image acquisition area is calculated. The gray value of each side edge area collected is compared with the mean gray value of all pixels. The area where the gray value of each side edge area is higher than the mean gray value of all pixels by 20% is considered as the high brightness area of ​​each side edge and is determined to be the roller area on that side. A series of feature points σ are extracted from the roller area on that side. i , where i = 1, 2, 3…n, n < 20, select 4*4 pixels with each feature point as the center to form a 1*16 column vector, and use the machine learning clustering algorithm to filter out the feature points where the grayscale values ​​of 4 pixels out of 16 pixels are higher than the grayscale values ​​of the remaining 12 pixels by 20%, and sort the filtered feature points from large to small according to the grayscale value difference of the pixels inside each feature point. The first four feature points are selected and defined as the boundary points of the side roller area. Call the area library function to calculate the area of ​​the quadrilateral surrounded by the four boundary points to obtain the area S of the side roller. 托辊 When S 托辊If the area decreases by more than 10%, it is judged that the belt is deviating toward the edge of that side; when S 托辊 If the area increases by more than 10%, it is determined that the belt is deviating to the other edge. 托辊 The belt can be adjusted by the control end based on the area change information, and an early warning of deviation can be issued.

[0017] In summary, the present application provides a visual detection method for belt deviation that can eliminate linear interference. When identifying the straight line of the belt edge, it can eliminate the linear interference of the material on the belt and accurately lock the straight line of the belt edge. The machine learning clustering algorithm selects the belt edge feature points for calculation to obtain the exposed roller area boundary points, calculates the area enclosed by the roller area boundary points, and judges the belt deviation situation based on the change in the area, thereby improving the detection accuracy and issuing a timely alarm for the existence of belt deviation during the operation of the belt conveyor to improve the reliability of its safe operation. At the same time, it can also reduce the loss of equipment and accessories, increase the service life of the equipment, and increase its transportation capacity. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and / or other aspects and advantages of the present application will become more clearly and more easily understood through the following description in conjunction with the accompanying drawings. The methods, processes, and / or structures in the drawings are further described based on exemplary embodiments. These exemplary embodiments are non-limiting exemplary embodiments. The same or similar elements in the drawings are represented by the same reference numerals.

[0019] Figure 1 A schematic diagram of a belt deviation visual detection system provided according to an embodiment of the present application is shown;

[0020] Figure 2 shows an image acquisition area according to an embodiment of the present application;

[0021] Figure 3 A schematic diagram of a belt straight line obtained by using the Hough straight line detection method according to an embodiment of the present application is shown;

[0022] Figure 4 shows a first interference straight line graph according to an embodiment of the present application;

[0023] Figure 5 shows a second interference straight line graph according to an embodiment of the present application;

[0024] Figure 6 shows the calculated areas of the first and second interference straight line graphs according to the embodiment of the present application;

[0025] Figure 7 A schematic flow chart of a method for visually detecting belt deviation according to an embodiment of the present application is shown;

[0026] List of reference numerals:

[0027] 10. Belt; 20. Coal; 11. Belt edge straight line; 12. Roller; 21. Interference straight line. DETAILED DESCRIPTION

[0028] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments, and specific illustrative embodiments are shown in the drawings by way of diagrams. However, it should be understood that other embodiments may be utilized and that changes in logic, components, and modules may be made. It should be noted that the present invention is not limited to the specific embodiments described below. In addition, for the sake of simplicity, detailed descriptions of known technologies that are not directly related to the present invention have been omitted to prevent confusion in understanding the present invention.

[0029] It should be noted that, in the description of this application, the terms "upper," "lower," "outer," and the like, indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate the description of this application and simplify the description. They are not intended to indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0030] In addition, it should be noted that Hough Line Detection is a commonly used line detection method in this field. Its basic principle is to utilize the duality of points and lines. In the line detection task, there is a one-to-one correspondence between lines in the image space and points in the parameter space, and there is also a one-to-one correspondence between lines in the parameter space and points in the image space. The results are: 1) Each line in the image space corresponds to a single point in the parameter space; 2) Any part of a line segment in the image space corresponds to the same point in the parameter space. Therefore, the Hough line detection algorithm converts the line detection problem in the image space into the point detection problem in the parameter space, and completes the line detection task by finding peaks in the parameter space.

[0031] Figure 1The following is a schematic diagram of a visual inspection system for belt deviation according to an embodiment of the present application. A camera is provided at a certain distance above the belt 10. The camera is preferably a binocular point cloud camera. It should be noted that the "certain distance" here is a common parameter that is adjusted by those skilled in the art according to the belt structure when adjusting the camera, so that the camera can clearly capture images. This application does not impose any special restrictions. Therefore, the "certain distance" here does not cause the present application to be unclear or insufficiently disclosed. Material is also transported on the surface of the belt 10. In this application, the material is coal 20. The camera and the belt 10 are both connected to the control end. The control end receives the visual image signal of the camera, performs calculations and processing, and determines whether the belt is deviating. Based on the visual image processing judgment result, the control end outputs a control signal for adjusting the belt.

[0032] Figure 2 The image acquisition area according to the embodiment of the present application is shown. The belt image is captured by a camera, and the image acquisition area is selected. When the image acquisition is successful, the image acquisition area contains the belt 10, the belt edge 11 and the rollers 12 with a partial area exposed on both sides of the belt edge. In the image acquisition direction, the remaining area of ​​the roller is covered by the belt. Since coal mining and transportation operations often take place in a dim underground working environment, in the image acquisition area, the rollers 12 appear as a highlighted area on the belt edge.

[0033] Figure 3 A schematic diagram of belt lines captured using the Hough line detection method according to an embodiment of the present application is shown. After successful image acquisition, the belt edge lines need to be extracted. Since coal mining and transportation operations often take place underground in dimly lit environments, the dark color of coal transported by the belt can easily form multiple interfering lines 21.

[0034] Figure 4 shows a first interference straight line graph according to an embodiment of the present application; Figure 5 The second interference straight line graph according to the embodiment of the present application is shown. The interference straight line 21 can be as follows: Figure 4 The non-intersecting straight lines shown can also be Figure 5 Take any two interfering lines (m, n) and set the coordinates of the two endpoints A and B of line m to be and Assume that the coordinates of the two endpoints C and D of line n are and Connect the two endpoints AB of line m and CD of line n respectively to form a circle like Figure 6 The calculated area S of the first and second interference straight line graphs according to the embodiment of the present application is shown. 四边形ABCD , the area calculation formula of quadrilateral ABCD is S 面积ABCD =(S ΔABC +SΔABD +S ΔACD +S ΔBCD ) / 2, when we substitute the coordinates of the two endpoints AB and CD of the two straight lines m and n respectively When , we can deduce that the calculation formula for the area s enclosed by any two straight lines m and n is: Lines m and n can be any arbitrary lines, including the two belt edge lines and multiple interfering lines. Different lines are selected, resulting in different calculated areas. When area s reaches its maximum value, the next two lines are extracted as the belt edge lines. Thus, this embodiment proposes an area calculation formula that eliminates the influence of material interfering lines when extracting the belt edge lines, accurately identifying the belt edge lines and improving detection accuracy.

[0035] Figure 7 The following is a flow chart of the visual detection method for belt deviation according to the embodiment of the present application. Step 1: Capture the belt image through the camera, select the image acquisition area, and determine whether the image acquisition is successful. If successful, proceed to the next step. If not, return to re-acquisition. Step 2: Extract the belt edge straight line: Use the Hough line detection method to obtain multiple straight lines for the selected image acquisition area. The multiple straight lines include the edges of the belt on both sides and the material interference line. Take any two straight lines m and n from the multiple straight lines, and set the coordinates of the two end points of line m to Set the coordinates of the two endpoints of line n to According to the formula: Calculate the area s enclosed by any two straight lines m and n. When the area s is the maximum value, extract the next two straight lines as the edge straight lines on both sides of the belt; Step 3, collect the brightness values ​​of the edge areas of the extracted edge straight lines on both sides, calculate the average grayscale values ​​of all pixels in the image acquisition area, compare the collected grayscale values ​​of each side edge area with the average grayscale values ​​of all pixels in the image acquisition area, obtain the area where the grayscale value of each side edge area is 20%-25% higher than the average grayscale value of all pixels, preferably 20%, regard this area as the high-brightness area of ​​each side edge, and determine it as the roller area of ​​this side.

[0036] Figure 7 It also shows step 4, extracting a series of feature points σ in the side roller area i, where i = 1, 2, 3…n, n < 20. The value of n can be appropriately adjusted to 15 according to the size of the side roller area. 4*4 pixels are selected with each feature point as the center to form a 1*16 column vector. The machine learning clustering algorithm is used to filter out feature points where the grayscale values ​​of 4 pixels out of 16 pixels are higher than the grayscale values ​​of the remaining 12 pixels by 20%-25%, preferably 20%. The filtered feature points are sorted from large to small according to the grayscale value difference of the pixels inside each feature point. The first four feature points are selected as the boundary points of the side roller area. The area library function is called to calculate the area of ​​the quadrilateral formed by the four boundary points to obtain the area S of the side roller. 托辊 . It should be noted that the area library function here is a conventional library function for calculating the area of ​​quadrilaterals in this field. The machine learning clustering algorithm can be KNN clustering or other conventional clustering algorithms, and this application does not make special limitations. In addition, alternatively, 5*5 pixels can be selected with each feature point as the center to form a 1*25 column vector. Those skilled in the art can make adaptive choices based on the size of the side roller area, and the embodiments can be expanded without departing from the spirit and principles of the present invention. Step 5, based on the side roller area S obtained in step 4 托辊 , when S 托辊 If the area decreases by more than 10%, it is judged that the belt is deviating toward the edge of that side; when S 托辊 If the area increases by more than 10%, it is determined that the belt is deviating toward the other side edge; Step 6: Based on the belt deviation information obtained in Step 5, the control end issues a deviation warning and adjusts the belt.

[0037] For those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of the present application. For some technical feature terms that are not explained, those skilled in the art are fully capable of making reasonable and unambiguous deductions based on the logical relationship between the preceding and following contexts, thereby clearly and completely implementing the above technical solutions. When the functional modules are actually implemented, they can all be implemented in the form of software called by processing elements; they can also be implemented in the form of hardware; some modules can be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. In addition, all or part of these modules can be integrated together or implemented independently. The control end described here can be an integrated circuit with signal processing capabilities. In the implementation process, some or all of the steps of the above method can be completed by the hardware integrated logic circuit in the processor element or by instructions in the form of software.

[0038] Systems and methods have been generally described herein in detail to facilitate understanding of the present invention. In addition, various specific details have been given to provide an overall understanding of embodiments of the present invention. However, those skilled in the relevant art will recognize that embodiments of the present invention may be practiced without one or more of these specific details, or may be practiced using other devices, systems, methods, components, parts, etc. In other cases, well-known structures, materials, and / or operations are not specifically shown or described in detail to avoid obscuring various aspects of embodiments of the present invention.

[0039] In summary, the technical solutions of the present application have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is readily understood by those skilled in the art that the scope of protection of the present application is clearly not limited to these specific embodiments. Without departing from the principles of the present application, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present application.

Claims

1. A visual detection method for belt deviation that can eliminate linear interference, characterized in that The detection method comprises the following steps: Step 1: Use the camera to capture the belt image and select the image capture area; Step 2: Use the Hough line detection method to obtain multiple straight lines in the selected image acquisition area. The multiple straight lines include the edges of the belt and the material interference line. Take any two straight lines m and n from the multiple straight lines, and set the coordinates of the two end points of line m as Set the coordinates of the two endpoints of line n to According to the formula: Calculate the area s enclosed by any two straight lines m and n. When the area s is the maximum, extract the next two straight lines as the two side edge lines of the belt. Step 3: Using the extracted edge lines on both sides, calculate the high-brightness area of ​​each side edge and determine it as the side roller area; Step 4: Extract a series of feature points in the side roller area, perform clustering processing through machine learning to obtain the boundary points of the side roller area, and call the library function to calculate the side roller area S based on the boundary points of the side roller area. 托辊 ; Step 5: When S 托辊 If the area value changes by more than 10%, it is judged that the belt is deviating; Step 6: Based on the belt deviation information obtained in step 5, a deviation warning is issued, and the belt is adjusted through the control end.

2. The detection method according to claim 1, wherein In step 2, any two straight lines m and n may cross or not cross, and the line connecting the endpoints of the two straight lines is The area of ​​the quadrilateral formed by the two straight lines m and n is the area s enclosed by the straight lines m and n.

3. The detection method according to claim 1, wherein The step 3 calculates the average grayscale value of all pixels in the image acquisition area, compares the grayscale value of each side edge area collected with the average grayscale value of all pixels, and obtains the area where the grayscale value of each side edge area is 20% higher than the average grayscale value of all pixels, which is regarded as the high-brightness area of ​​each side edge and determined to be the roller area of ​​that side.

4. The detection method according to claim 1, wherein The step 4 extracts a series of characteristic points σ in the side roller area i , where i = 1, 2, 3…n, n < 20. 4*4 pixels are selected with each feature point as the center to form a 1*16 column vector. Feature points whose grayscale values ​​of 4 pixels out of 16 pixels are higher than 20% of the grayscale values ​​of the remaining 12 pixels are screened out through machine learning clustering algorithm. The screened feature points are sorted from large to small according to the grayscale value difference of the pixels inside each feature point. The first four feature points are selected and defined as the boundary points of the side roller area.

5. The detection method according to claim 4, characterized in that The step 4 calls the area library function to calculate the area of ​​the quadrilateral surrounded by the four boundary points, thereby obtaining the area S of the side roller. 托辊 .

6. The detection method according to claim 1, characterized in that The step 4 only calculates the area of ​​the roller on one side.

7. The detection method according to claim 1, characterized in that In step 5, when S 托辊 If the area decreases by more than 10%, it is judged that the belt is deviating toward the edge of that side; when S 托辊 If the area increases by more than 10%, it is determined that the belt is deviating to the other side edge.

8. The detection method according to claim 1, wherein The material in step 2 is coal.

Images