A coal washing plant belt deviation monitoring and protection system

Abstract

The present invention relates to the field of image processing technology, and in particular to a coal washing plant belt deviation monitoring and protection system, comprising: collecting belt transport images; obtaining an initial segmentation threshold according to the distribution of gray values ​​of each local area in the belt transport image; calculating the feature sensitivity of each local area under each segmentation threshold, and obtaining the optimal segmentation threshold of each local area according to the feature sensitivity; obtaining a binary image of the belt transport image according to the optimal segmentation threshold of each local area, obtaining a coal distribution edge baseline according to the binary image, and judging whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline. The present invention can remove the influence of small coal blocks and belt scratches by adaptively obtaining the optimal segmentation threshold of each local area, accurately obtain the distribution of coal carried on the belt, and promptly discover the deviation of the belt caused by uneven coal distribution.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a coal washing plant belt deviation monitoring and protection system. Background Art

[0002] The deviation of the belt in the coal washing plant will cause material spillage, belt edge damage, shortened belt life and other problems. Therefore, it is necessary to monitor the status of the belt in the coal washing plant. There are many reasons for the deviation of the belt, among which the most serious is the uneven distribution of the coal carried on the belt, which causes uneven force on the belt and causes the belt to deviate. Therefore, it is necessary to monitor whether the distribution of the coal carried on the belt is uniform.

[0003] The distribution of coal carried on the monitoring belt is mainly judged based on the distribution edge of the coal. When obtaining the distribution edge of the coal, small coal pieces and scratches on the belt will affect the accuracy of the distribution edge of the coal, which will lead to the inability to accurately obtain the distribution of the coal carried on the belt, and the inability to timely discover the deviation of the belt caused by uneven coal distribution. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a coal washing plant belt deviation monitoring and protection system, the system comprising:

[0005] An image acquisition module, used for acquiring belt transport images;

[0006] An initial segmentation threshold acquisition module is used to obtain an initial segmentation threshold according to the distribution of grayscale values ​​of each local area in the belt transport image;

[0007] An optimal segmentation threshold acquisition module is used to calculate the feature sensitivity of each local area under each segmentation threshold, and obtain the optimal segmentation threshold of each local area according to the feature sensitivity;

[0008] The coal distribution judgment module is used to obtain a binary image of the belt transport image according to the optimal segmentation threshold of each local area, obtain a coal distribution edge baseline according to the binary image, and judge whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline.

[0009] Furthermore, the calculation formula of the initial segmentation threshold is as follows: In the formula, represents the initial segmentation threshold, h represents the gray value, N represents the number of local areas in the belt transport image, represents the importance of the i-th local area, represents the frequency of gray value h in the i-th local area, Represents the frequency of the adjusted gray value h in the i-th local area.

[0010] Furthermore, the method for obtaining the importance of the local area is specifically as follows:

[0011] Pixels whose gradient amplitude is greater than the preset threshold Y are regarded as highlighted pixels; the result of linear normalization of the ratio of the mean grayscale value of all highlighted pixels in each local area to the information entropy is recorded as the importance of each local area.

[0012] Furthermore, the calculation formula of the feature sensitivity of each local area under each segmentation threshold is as follows:

[0013] In the formula, Represents the feature sensitivity of the local area under the segmentation threshold, represents the number of connected domains in the local area, represents the feature weight of the jth connected domain, represents the main direction of the jth connected domain, d represents the empirical angle value, Represents a logarithmic function with a natural constant as base.

[0014] Furthermore, the calculation formula of the feature weight of the connected domain is as follows:

[0015] In the formula, It represents the angle between the line connecting the geometric center of the j-th connected domain and the geometric center of the first neighboring connected domain of the j-th connected domain and the horizontal direction. It represents the angle between the line connecting the geometric center of the j-th connected domain and the geometric center of the second neighboring connected domain of the j-th connected domain and the horizontal direction. The first neighboring connected domain and the second neighboring connected domain of the j-th connected domain are the two connected domains farthest from the j-th connected domain.

[0016] Furthermore, the method for acquiring the connected domain and the main direction of the connected domain is specifically as follows:

[0017] Perform threshold segmentation on the local area according to the initial segmentation threshold H to obtain a binary image, perform connected domain analysis on all pixels marked as 1, and obtain several connected domains in each local area;

[0018] The geometric center of each connected domain is obtained, and the angle between the line connecting each highlighted pixel and the geometric center and the horizontal direction is taken as the direction of each highlighted pixel, and the average value of the directions of all highlighted pixels is taken as the main direction of the connected domain.

[0019] Furthermore, the step of obtaining the optimal segmentation threshold for each local area includes the following specific steps:

[0020] Calculate the feature sensitivity of the local area under the initial segmentation threshold H: If the feature sensitivity of the local area under the initial segmentation threshold is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the initial segmentation threshold meets the high sensitivity requirement, and the initial segmentation threshold is used as the optimal segmentation threshold of the local area;

[0021] If the feature sensitivity of the local area under the initial segmentation threshold is less than or equal to the preset sensitivity threshold B, calculate the feature sensitivity of the local area under the segmentation threshold H-1 (H represents the initial segmentation threshold). If the feature sensitivity of the local area under the segmentation threshold H-1 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-1 meets the high sensitivity requirement. At this time, the segmentation threshold H-1 is used as the optimal segmentation threshold of the local area.

[0022] If the feature sensitivity of the local area under the segmentation threshold H-1 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+1 is calculated. If the feature sensitivity of the local area under the segmentation threshold H+1 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+1 meets the high sensitivity requirement. At this time, the segmentation threshold H+1 is used as the optimal segmentation threshold of the local area.

[0023] If the feature sensitivity of the local area under the segmentation threshold H+1 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-2 is calculated. If the feature sensitivity of the local area under the segmentation threshold H-2 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-2 meets the high sensitivity requirement. At this time, the segmentation threshold H-2 is used as the optimal segmentation threshold of the local area.

[0024] If the feature sensitivity of the local area under the segmentation threshold H-2 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+2 is calculated. If the feature sensitivity of the local area under the segmentation threshold H+2 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+2 meets the high sensitivity requirement. At this time, the segmentation threshold H+2 is used as the optimal segmentation threshold of the local area.

[0025] And so on, until the optimal segmentation threshold of the local area is obtained.

[0026] Furthermore, the step of obtaining the binary image of the belt transport image according to the optimal segmentation threshold of each local area includes the following specific steps:

[0027] Threshold segmentation is performed on each local area according to the optimal segmentation threshold of each local area to obtain binary images of all local areas. The binary images of all local areas constitute the binary image of the belt conveyor image.

[0028] Furthermore, the step of obtaining the coal distribution edge baseline according to the binary image includes the following specific steps:

[0029] Hough line detection is performed on the binary image of the belt transport image. Among all the lines with angles in the range of [d-5, d+5], the line with the largest voting rate is used as the coal distribution edge baseline, and d represents the empirical value of the angle.

[0030] Furthermore, judging whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline includes the following specific steps:

[0031] Connected domain analysis is performed on the binary image of the belt transport image to obtain all connected domains; the pixel points that belong to the same connected domain as the pixel points on the coal distribution edge baseline are taken as coal edge pixels, and the mean distance from all coal edge pixels to the coal distribution edge baseline is calculated. If it is greater than the preset distance threshold X, the distribution of coal in the belt transport image is uneven.

[0032] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the optimal segmentation threshold of each local area through adaptive means, obtains a binary image of the belt transport image according to the optimal segmentation threshold of each local area, obtains a coal distribution edge baseline according to the binary image, and judges whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline. The influence of small coal blocks and belt scratches can be removed, the distribution of coal carried on the belt can be accurately obtained, and the deviation of the belt caused by uneven coal distribution can be discovered in time. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 A system block diagram of a coal washing plant belt deviation monitoring and protection system of the present invention;

[0035] Figure 2 Transport images for belts;

[0036] Figure 3 It is the gradient image of the belt transport image;

[0037] Figure 4 It is a binary image of the belt transport image under a segmentation threshold;

[0038] Figure 5 It is a binary image of the belt conveyor image under another segmentation threshold. DETAILED DESCRIPTION

[0039] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a coal washing plant belt deviation monitoring and protection system proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0041] The distribution of coal carried on the monitoring belt is mainly judged based on the distribution edge of the coal. If the distribution of the coal is uniform, the distribution edge of the coal is a straight line. The more uneven the distribution of the coal is, the more chaotic the distribution edge of the coal is. The distribution edge of the coal can be obtained by threshold segmentation of the belt transport image, but small coal blocks and scratches on the belt will affect the accuracy of the distribution edge of the coal. Therefore, the present invention accurately obtains the distribution of the coal carried on the belt by removing the influence of small coal blocks and scratches on the belt, and promptly discovers the deviation of the belt caused by uneven distribution of coal.

[0042] The specific scheme of the coal washing plant belt deviation monitoring and protection system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0043] See also Figure 1 , which shows a coal washing plant belt deviation monitoring and protection system provided by an embodiment of the present invention, the system includes the following modules:

[0044] The image acquisition module 101 is used to acquire belt conveying images.

[0045] Due to the limited viewing angle of the camera, it is necessary to use a dual-view camera to capture two images of the coal being transported by the coal washing plant belt, and perform the same operation on the two images. The following will take one of the images as an example for explanation.

[0046] Specifically, the belt image is captured by a camera, grayed and histogram equalized, and used as the belt transport image. Figure 2 , which shows a belt transport image.

[0047] It should be noted that histogram equalization is used to enhance the image. Grayscale processing and histogram equalization are both well-known technologies and will not be described in detail here. By enhancing the grayscale image, the visual effect of the middle edge can be magnified, making the subsequent segmentation effect better.

[0048] The initial segmentation threshold acquisition module 102 is used to obtain the initial segmentation threshold according to the distribution of gray values ​​in the belt conveyor image.

[0049] It should be noted that the present invention obtains the distribution edge of the coal by performing threshold segmentation on the belt transport image. When setting the segmentation threshold, the grayscale distribution characteristics of the pixel points on the distribution edge of the coal are combined to more accurately obtain the distribution edge of the coal. Therefore, it is necessary to first obtain the pixel points on the image that may belong to the distribution edge of the coal; because the distribution edge of the coal is the position that distinguishes the coal from the conveyor belt, and the difference in grayscale values ​​between the coal and the conveyor belt is large, the gradient amplitude of the pixel points on the distribution edge of the coal is relatively large.

[0050] The preset gradient amplitude threshold is Y, which is used to distinguish highlight pixels from ambient pixels. Implementers can set the threshold according to actual implementation conditions, for example, Y=100.

[0051] Specifically, edge detection is performed on the belt transport image to obtain the gradient amplitude of each pixel. Figure 3 , which shows the gradient image of the belt transport image; the pixels whose gradient amplitude is greater than the preset threshold value Y are taken as highlight pixels, and the highlight pixels are pixels that may belong to the distribution edge of coal, and the pixels whose gradient amplitude is less than or equal to the preset threshold value Y are taken as environmental pixels.

[0052] The preset size is K×K, which is used to divide the belt conveyor image. The implementer can set the size according to the actual implementation situation, for example, K×K=20×20.

[0053] Furthermore, the belt transport image is divided into several local areas, and the size of each local area is equal to the preset size. For the local areas that do not meet the preset size, the size is made equal to the preset size by supplementing the pixels with a gray value of 0; the result of linear normalization of the ratio of the mean gray value of all highlighted pixels in each local area to the information entropy is recorded as the importance of each local area.

[0054] It should be noted that since the grayscale value distribution of the distribution edge of coal is relatively uniform, if the local area belongs to the distribution edge of coal, the smaller the information entropy, the distribution edge of coal is the location where coal is distinguished from the conveyor belt. Therefore, the mean grayscale value is larger than that of the coal area; the smaller the information entropy of all local areas and the larger the mean grayscale value, the greater the importance of the local area, and the more likely the corresponding local area is to be the distribution edge of coal.

[0055] Furthermore, the frequency of each gray value is adjusted according to the importance of the local area, and the gray values ​​are weighted and summed according to the adjusted frequency to obtain the initial segmentation threshold. The specific calculation formula is: In the formula, represents the initial segmentation threshold, h represents the gray value, N represents the number of local areas in the belt transport image, represents the importance of the i-th local area, It represents the frequency increase when the frequency of each gray value is adjusted according to the importance of the i-th local area. It represents the frequency of the gray value h in the i-th local area. The frequency refers to the ratio of the number of highlighted pixels with gray value h in the i-th local area to the number of all highlighted pixels.

[0056] It should be noted that the distribution edge of coal is the position where the coal is distinguished from the conveyor belt. Since the conveyor belt is lighter in color than the coal blocks, that is, the grayscale value of the pixel points belonging to the conveyor belt is larger than the grayscale value of the pixel points belonging to the coal blocks. All the pixels at the distribution edge of the coal have a larger gradient, and the highlighted pixels represent the pixels with a larger gradient in the belt transport image. These pixels are more likely to be the pixels at the distribution edge of the coal. Therefore, the frequency of each grayscale value is adjusted according to the distribution of the highlighted pixels, and the grayscale values ​​are weighted and summed according to the adjusted frequency to obtain the initial segmentation threshold. This can make the obtained initial segmentation threshold more have the grayscale characteristics of the pixels at the distribution edge of the coal.

[0057] The optimal segmentation threshold acquisition module 103 is used to calculate the feature sensitivity of each local area under each segmentation threshold, and obtain the optimal segmentation threshold of each local area according to the feature sensitivity.

[0058] It should be noted that, under normal circumstances, to adaptively obtain the optimal segmentation threshold of each local area, it is necessary to traverse each grayscale value and judge the segmentation effect of each grayscale value as the segmentation threshold. The present invention obtains the initial segmentation threshold of the grayscale characteristics of the pixel points at the distribution edge of the coal according to the grayscale value distribution of each local area, and calculates the feature sensitivity under each segmentation threshold in turn with the initial segmentation threshold as the starting point. When the feature sensitivity is greater than the preset threshold, it stops. At this time, the optimal segmentation threshold is obtained, which can speed up the acquisition of the optimal segmentation threshold and reduce the amount of calculation.

[0059] It should be noted that after the initial threshold is determined, in order to combine the scene features to obtain the optimal threshold K, a continuous iteration process is required. Iteration refers to how the threshold image will change each time it is added or subtracted based on the initial threshold. In order to achieve this iteration, it is necessary to determine the iteration conditions and stop conditions. The connected domain formed by the coal edge has directional irregularities and some intermittent connections. For the connected domain formed by the scratches between the belt edge and the coal edge and the small coal blocks, their characteristics are more obvious, with isolated connected domains and intermittent straight lines; please refer to Figure 4 and Figure 5 , which shows the binary images of the belt conveyor image under two segmentation thresholds.

[0060] The preset sensitivity threshold is B, which is used to determine whether the feature sensitivity of the local area meets the high sensitivity requirement. The implementer can set the sensitivity threshold according to the actual implementation situation, for example, B=0.8.

[0061] Specifically, for any local area, the process of obtaining the optimal segmentation threshold of the local area is:

[0062] (1) Calculate the feature sensitivity of the local region under the initial segmentation threshold H: If the feature sensitivity of the local region under the initial segmentation threshold is greater than the preset sensitivity threshold B, the feature sensitivity of the local region under the initial segmentation threshold meets the high sensitivity requirement. At this time, the initial segmentation threshold is used as the optimal segmentation threshold of the local region.

[0063] (2) If the feature sensitivity of the local area under the initial segmentation threshold is less than or equal to the preset sensitivity threshold B, calculate the feature sensitivity of the local area under the segmentation threshold H-1 (H represents the initial segmentation threshold). If the feature sensitivity of the local area under the segmentation threshold H-1 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-1 meets the high sensitivity requirement. At this time, the segmentation threshold H-1 is used as the optimal segmentation threshold of the local area.

[0064] (3) If the feature sensitivity of the local area under the segmentation threshold H-1 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+1 is calculated. If the feature sensitivity of the local area under the segmentation threshold H+1 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+1 meets the high sensitivity requirement. At this time, the segmentation threshold H+1 is used as the optimal segmentation threshold of the local area.

[0065] (4) If the feature sensitivity of the local area under the segmentation threshold H+1 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-2 is calculated. If the feature sensitivity of the local area under the segmentation threshold H-2 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-2 meets the high sensitivity requirement. At this time, the segmentation threshold H-2 is used as the optimal segmentation threshold of the local area.

[0066] (5) If the feature sensitivity of the local area under the segmentation threshold H-2 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+2 is calculated. If the feature sensitivity of the local area under the segmentation threshold H+2 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+2 meets the high sensitivity requirement. At this time, the segmentation threshold H+2 is used as the optimal segmentation threshold of the local area.

[0067] (6) And so on, until the optimal segmentation threshold of the local area is obtained.

[0068] The method for calculating the feature sensitivity of the local area under the segmentation threshold is the same. The following is an example of calculating the feature sensitivity of the local area under the initial segmentation threshold H, as follows:

[0069] According to the initial segmentation threshold H, the local area is threshold segmented to obtain a binary image, and the connected domain analysis is performed on all the pixels marked as 1 to obtain several connected domains in each local area; the geometric center of each connected domain is obtained, and the angle between the line connecting each highlighted pixel and the geometric center and the horizontal direction is taken as the direction of each highlighted pixel, and the average value of the directions of all highlighted pixels is taken as the main direction of the connected domain; the feature sensitivity of the local area under the segmentation threshold is calculated, and the specific calculation formula is: In the formula, Represents the feature sensitivity of the local area under the segmentation threshold, represents the number of connected domains in the local area, represents the feature weight of the jth connected domain, represents the main direction of the jth connected domain, d represents the empirical angle value, represents a logarithmic function with a natural constant as the base, where the feature weight of the jth connected domain , It represents the angle between the line connecting the geometric center of the j-th connected domain and the geometric center of the first neighboring connected domain of the j-th connected domain and the horizontal direction. It represents the angle between the line connecting the geometric center of the j-th connected domain and the geometric center of the second neighboring connected domain of the j-th connected domain and the horizontal direction. The first neighboring connected domain and the second neighboring connected domain of the j-th connected domain are the two connected domains farthest from the j-th connected domain.

[0070] Since the position and angle of the camera and the position of the belt are fixed, when the coal is evenly distributed, the angle between the distribution edge of the coal on the image and the horizontal direction is fixed, and this angle is recorded as the empirical angle value d.

[0071] It should be noted that the angle empirical value is the angle between the distribution edge of the coal on the image and the horizontal direction when the coal is evenly distributed. The difference between the main direction of the connected domain and the angle empirical value The smaller it is, the more likely that the connected domain is the edge of the coal distribution, and the greater the feature sensitivity of the local area under the segmentation threshold; the more consistent the angle between the geometric centers of the connected domain and its two adjacent connected domains (neighborhood connected domains), the more likely that the connected domain and its two adjacent connected domains are small connected domains separated on the edge of the coal distribution, and the more likely that the connected domain is on the edge of the coal distribution, the greater the weight of the connected domain.

[0072] The coal distribution judgment module 104 is used to obtain a binary image of the belt transport image according to the optimal segmentation threshold of each local area, obtain a coal distribution edge baseline according to the binary image, and judge whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline.

[0073] The preset distance threshold is X, which is used to determine whether the distribution of coal is uniform. The implementers can set the distance threshold according to the actual implementation situation. For example, X=0.1L, where L represents the width of the conveyor belt in the belt transport image.

[0074] Specifically, the belt transport images under the left camera perspective and the right camera perspective are respectively segmented according to the optimal segmentation threshold of each local area to obtain binary images of all local areas, and the binary images of all local areas constitute the binary image of the belt transport image; the connected domain analysis is performed on the binary image of the belt transport image to obtain all connected domains; Hough line detection is performed on the two images, and the straight line with the largest voting rate among all straight lines with angles in the range of [d-5, d+5] is used as the coal distribution edge baseline, d represents the empirical value of the angle, and the pixel points belonging to the same connected domain as the pixel points on the coal distribution edge baseline are used as coal edge pixels. The mean distance from all coal edge pixels to the coal distribution edge baseline is calculated. If it is greater than the preset distance threshold X, it means that the distribution of coal in the belt transport image is uneven, and the belt may deviate if it continues to run. Manual intervention is used to make the coal on the belt evenly distributed.

[0075] The system of the present invention comprises an image acquisition module, an initial segmentation threshold acquisition module, an optimal segmentation threshold acquisition module and a coal distribution judgment module. The present invention obtains the optimal segmentation threshold of each local area by adaptively obtaining a binary image of the belt transport image according to the optimal segmentation threshold of each local area, obtains a coal distribution edge baseline according to the binary image, and judges whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline, thereby removing the influence of small coal blocks and belt scratches, accurately obtaining the distribution of coal carried on the belt, and timely discovering the deviation of the belt caused by uneven coal distribution.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A coal washing plant belt deviation monitoring and protection system, characterized in that: The system comprises: An image acquisition module, used for acquiring belt transport images; An initial segmentation threshold acquisition module is used to obtain an initial segmentation threshold according to the distribution of grayscale values ​​of each local area in the belt transport image; An optimal segmentation threshold acquisition module is used to calculate the feature sensitivity of each local area under each segmentation threshold, and obtain the optimal segmentation threshold of each local area according to the feature sensitivity; A coal distribution judgment module is used to obtain a binary image of the belt transport image according to the optimal segmentation threshold of each local area, obtain a coal distribution edge baseline according to the binary image, and judge whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline; The calculation formula of the initial segmentation threshold is as follows: In the formula, represents the initial segmentation threshold, h represents the gray value, N represents the number of local areas in the belt transport image, represents the importance of the i-th local area, represents the frequency of gray value h in the i-th local area, represents the frequency of the gray value h adjusted in the i-th local area; The calculation formula of the feature sensitivity of each local area under each segmentation threshold is as follows: In the formula, Represents the feature sensitivity of the local area under the segmentation threshold, represents the number of connected domains in the local area, represents the feature weight of the jth connected domain, represents the main direction of the jth connected domain, d represents the empirical angle value, represents a logarithmic function with a natural constant as base; The specific steps of obtaining the optimal segmentation threshold for each local area are as follows: Calculate the feature sensitivity of the local area under the initial segmentation threshold H: If the feature sensitivity of the local area under the initial segmentation threshold is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the initial segmentation threshold meets the high sensitivity requirement, and the initial segmentation threshold is used as the optimal segmentation threshold of the local area; If the feature sensitivity of the local area under the initial segmentation threshold is less than or equal to the preset sensitivity threshold B, calculate the feature sensitivity of the local area under the segmentation threshold H-1 (H represents the initial segmentation threshold). If the feature sensitivity of the local area under the segmentation threshold H-1 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-1 meets the high sensitivity requirement. At this time, the segmentation threshold H-1 is used as the optimal segmentation threshold of the local area. If the feature sensitivity of the local area under the segmentation threshold H-1 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+1 is calculated. If the feature sensitivity of the local area under the segmentation threshold H+1 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+1 meets the high sensitivity requirement. At this time, the segmentation threshold H+1 is used as the optimal segmentation threshold of the local area. If the feature sensitivity of the local area under the segmentation threshold H+1 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-2 is calculated. If the feature sensitivity of the local area under the segmentation threshold H-2 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H-2 meets the high sensitivity requirement. At this time, the segmentation threshold H-2 is used as the optimal segmentation threshold of the local area. If the feature sensitivity of the local area under the segmentation threshold H-2 is less than or equal to the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+2 is calculated. If the feature sensitivity of the local area under the segmentation threshold H+2 is greater than the preset sensitivity threshold B, the feature sensitivity of the local area under the segmentation threshold H+2 meets the high sensitivity requirement. At this time, the segmentation threshold H+2 is used as the optimal segmentation threshold of the local area. And so on, until the optimal segmentation threshold of the local area is obtained.

2. A coal washing plant belt deviation monitoring and protection system according to claim 1, characterized in that: The method for obtaining the importance of the local area is specifically as follows: Pixels whose gradient amplitude is greater than the preset threshold Y are regarded as highlighted pixels; the result of linear normalization of the ratio of the mean grayscale value of all highlighted pixels in each local area to the information entropy is recorded as the importance of each local area.

3. A coal washing plant belt deviation monitoring and protection system according to claim 1, characterized in that: The calculation formula of the feature weight of the connected domain is as follows: In the formula, It represents the angle between the line connecting the geometric center of the j-th connected domain and the geometric center of the first neighboring connected domain of the j-th connected domain and the horizontal direction. It represents the angle between the line connecting the geometric center of the j-th connected domain and the geometric center of the second neighboring connected domain of the j-th connected domain and the horizontal direction. The first neighboring connected domain and the second neighboring connected domain of the j-th connected domain are the two connected domains farthest from the j-th connected domain.

4. A coal washing plant belt deviation monitoring and protection system according to claim 1, characterized in that: The method for obtaining the connected domain and the main direction of the connected domain is specifically as follows: Perform threshold segmentation on the local area according to the initial segmentation threshold H to obtain a binary image, perform connected domain analysis on all pixels marked as 1, and obtain several connected domains in each local area; The geometric center of each connected domain is obtained, and the angle between the line connecting each highlighted pixel and the geometric center and the horizontal direction is taken as the direction of each highlighted pixel, and the average value of the directions of all highlighted pixels is taken as the main direction of the connected domain.

5. A coal washing plant belt deviation monitoring and protection system according to claim 1, characterized in that: The step of obtaining a binary image of the belt transport image according to the optimal segmentation threshold of each local area includes the following specific steps: Threshold segmentation is performed on each local area according to the optimal segmentation threshold of each local area to obtain binary images of all local areas. The binary images of all local areas constitute the binary image of the belt conveyor image.

6. A coal washing plant belt deviation monitoring and protection system according to claim 1, characterized in that: The specific steps of obtaining the coal distribution edge baseline according to the binary image are as follows: Hough line detection is performed on the binary image of the belt transport image. Among all the lines with angles in the range of [d-5, d+5], the line with the largest voting rate is used as the coal distribution edge baseline, and d represents the empirical value of the angle.

7. A coal washing plant belt deviation monitoring and protection system according to claim 1, characterized in that: The specific steps of judging whether the distribution of coal in the belt transport image is uniform according to the coal distribution edge baseline are as follows: Connected domain analysis is performed on the binary image of the belt transport image to obtain all connected domains; the pixel points that belong to the same connected domain as the pixel points on the coal distribution edge baseline are taken as coal edge pixels, and the mean distance from all coal edge pixels to the coal distribution edge baseline is calculated. If it is greater than the preset distance threshold X, the distribution of coal in the belt transport image is uneven.

Images