A conveyor belt abnormality monitoring method and system based on image processing

Abstract

The present invention relates to the field of image data processing, and more specifically, to a conveyor belt abnormality monitoring method and system based on image processing, the method comprising: obtaining the grayscale image and edge of the surface image of the sampling frame when the conveyor belt is conveyed according to a preset interval, taking any sampling frame as the target frame, and using the gradient direction difference level and the mark point similarity calculation to detect the regularity of the target edge in the target frame; using the regularity of the target edge as an adjustment factor to calculate the similarity of the position relationship of the mark points between the target edge and the edge to be detected, obtaining the most matching edge, and combining the time series weight to evaluate the significant matching of the target edge. The ratio of regularity to significant matching is used as a potential tear score to determine the conveyor belt abnormality; when there is potential tearing, a detailed analysis is performed using structured light technology. The present invention improves overall efficiency by calling the more expensive structured light technology for detailed analysis when it is determined that there is potential tearing.

Description

Technical Field

[0001] The present invention relates to the field of image data processing, and more specifically, to a conveyor belt abnormality monitoring method and system based on image processing. Background Art

[0002] In the coal mining industry, belt conveyors are key transportation equipment, and monitoring their operating status is crucial to ensuring the safety and efficiency of coal mining production. During the operation of the conveyor belt, due to the objects it carries and the wear of the conveyor belt surface caused by the operation of the machine, conveyor belt tearing failures often occur. If these failures are not detected and handled in time, they may cause serious production accidents and economic losses.

[0003] The existing Chinese patent application document with publication number CN102621168A discloses an X-ray transmission detection system for underground conveyor belts in coal mines, which includes an acquisition and storage subsystem and a data analysis and display subsystem. The data acquisition and storage subsystem is used for acquiring and storing data of underground conveyor belts; the data analysis and display subsystem can read the image data of the storage unit in the acquisition box, perform intelligent analysis on the image data, locate the joints and abnormal areas of steel wires in the belt, and intuitively present the internal structure information of the belt in the form of images; the method is: connecting the data acquisition box with the control box and starting the data acquisition subsystem; storing the acquired transmission data in real time in the internal storage unit of the data acquisition box; after storage, connecting the data acquisition box with the data analysis and display subsystem to transmit the transmission data; reading the transmission data, analyzing the transmission anomaly, and displaying the transmission image.

[0004] In the above scheme, transmission data is read, transmission anomalies are analyzed, and conveyor belt anomaly monitoring is obtained. Currently, the camera is used to capture the deformation of the light pattern on the surface of the object to calculate the shape and position of the object. When monitoring the conveyor belt for anomalies, it has high accuracy and can provide detailed image information. However, the data processing and analysis in structured light technology requires strong computing power and consumes a lot of computing resources. The processing of large amounts of data will affect the ability of real-time monitoring. Summary of the invention

[0005] In order to solve the problem that data processing and analysis in structured light technology requires strong computing power, thereby affecting the ability of real-time monitoring, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a conveyor belt abnormality monitoring method based on image processing includes: obtaining a surface image of a sampling frame when the conveyor belt is conveyed according to a preset interval and performing grayscale processing to obtain a grayscale image, and performing edge detection, taking any sampling frame as a target frame, obtaining a target edge of the target frame and an edge to be detected of other frame images, grading the gradient direction difference in the grayscale image corresponding to the sampling frame, and obtaining the frequency of each grading to obtain the regularity of the target edge; respectively presetting marking points for the target edge and the edge to be detected, using the regularity of the target edge as an adjustment factor to calculate the similarity of the positional relationship of the marking points between the target edge and the edge to be detected, and obtaining the most matching edge; using the gradient direction difference, grayscale difference and grayscale gradient difference of each pixel point between the target edge and the most matching edge to calculate the matching degree between the target edge and the most matching edge; assigning weights to the matching degree of the most matching edge according to the image frame sequence to obtain the significant matching degree of the target edge in multiple frames of images. , , where Indicates the total number of sample frame images, Indicates the sequence number of the frame to which the best matching edge belongs, Indicates the target edge and the The matching degree of the most matching edge in the sampling frames is determined; the ratio of the regularity of the target edge to the significant matching is used as the score of the potential tear of the target edge, and whether there is an abnormality in the conveyor belt is determined based on the score.

[0007] By comparing the edge features in different frame images, anomalies on the conveyor belt can be identified more accurately. By calculating the significant matching of the target edge in multiple frame images, the severity of the anomaly can be quantified. The ratio of the regularity of the target edge to the significant matching is used as the score of potential tearing, which can more accurately determine whether there is an abnormality on the conveyor belt.

[0008] In one embodiment, the step of grading the gradient direction difference in the grayscale image corresponding to the sampling frame includes:

[0009] Get the gradient direction of each pixel in the grayscale image corresponding to the sampling frame, where the value range of the gradient direction is , divide the direction value range into several intervals, and each interval is used as a level of gradient direction.

[0010] By dividing the value range of the gradient direction into different intervals, the edge features in the image can be analyzed more carefully and the recognition ability of the surface details of the conveyor belt can be enhanced. The hierarchical processing helps to extract the key features of the conveyor belt surface, such as texture, cracks, wear, etc., which can reduce the impact of noise on edge feature extraction and improve the reliability of detection.

[0011] In one embodiment, obtaining the regularity of the target edge includes:

[0012] Calculate the mean of the gradient direction differences of all adjacent pixels on the target edge and the edge to be detected, and perform normalization processing;

[0013] The level frequency of each gradient direction difference is counted, the entropy of each level frequency is calculated, and the product of the entropy and the normalized mean is used as the regularity of the target edge.

[0014] Edges with higher regularity usually indicate that the conveyor belt surface is in normal condition, while edges with lower regularity may indicate abnormalities, such as tears or damage. The size of the entropy value can distinguish the characteristics of different edges. A high entropy value may indicate that the edge is more complex or irregular, while a low entropy value may indicate that the edge is simpler or regular.

[0015] In one embodiment, the step of respectively presetting marking points for the target edge and the edge to be detected includes:

[0016] In response to the target edge being an unclosed edge, marking two breakpoints of the unclosed edge as point 1 and point 2, marking the midpoint of the edge as point 3, and marking the midpoint between two adjacent points as point 4 and point 5;

[0017] In response to the target edge being a closed edge, the centroid of the closed edge is marked as point 3, the pixel point farthest from point 3 is marked as point 1, a straight line is determined based on point 1 and point 3, and the straight line is extended in the opposite direction of point 1 to the point where the closed edge intersects is marked as point 2, and the midpoints between the two adjacent points are marked as points 4 and 5.

[0018] By presetting specific marking points on the edge, the key features of the edge can be located more accurately. The setting of marking points helps to extract the geometric features of the edge and compare the marking points on different edges. At the same time, presetting a limited number of marking points on the edge can reduce the amount of calculation and improve the execution efficiency of the algorithm.

[0019] In one embodiment, the best matching edge includes:

[0020] Calculate the average value of the cumulative sum of the distance between each pair of marked points on the target edge and the distance difference between each pair of marked points on the edge to be detected, and obtain the shape similarity of the target edge and the edge to be detected; calculate the average value of the distance between the position of the marked point on the target edge in the grayscale image and the position of the marked point on the edge to be detected in the grayscale image, and obtain the position similarity of the target edge and the edge to be detected;

[0021] The product of the shape similarity and the position similarity is normalized, and the reciprocal of the regularity is used as the normalized exponential power to obtain the approximation between the target edge and the edge to be detected, and the edge to be detected with the greatest approximation is used as the edge that best matches the target edge.

[0022] Combining shape similarity and position similarity can comprehensively evaluate the similarity of two edges. Taking the inverse of regularity as the normalized exponential power can take into account the impact of edge regularity on matching and improve the robustness of matching. At the same time, it is applicable to different types of edge features.

[0023] In one embodiment, the matching degree satisfies the following relationship:

[0024] ;

[0025] In the formula, Indicates the matching degree between the target edge and the best matching edge, Represents the total number of pixels on the edge of the target. Represents the first The pixel point corresponds to the first The difference in the gray gradient direction of pixels, represents the cosine function, Represents the first The pixel point corresponds to the first The difference in the gray value of pixels is Represents the first The pixel point corresponds to the first The difference in the grayscale gradient value of each pixel is Represents the normalization function.

[0026] In one embodiment, judging whether there is an abnormality in the conveyor belt based on the score includes:

[0027] In response to the score being greater than or equal to a preset threshold, there is an abnormality in the conveyor belt, the target edge is potentially torn, and the line structured light technology is activated for detection; conversely, when the score is less than the preset threshold, the conveyor belt is normal.

[0028] In a second aspect, a conveyor belt abnormality monitoring system based on image processing includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned conveyor belt abnormality monitoring method based on image processing is implemented.

[0029] The present invention has the following effects:

[0030] 1. The present invention can evaluate the regularity of the edge by grading and frequency analysis of the gradient direction difference, thereby identifying irregular edges and obtaining potential tearing or damage, setting marking points for the target edge and the edge to be detected respectively, and quickly obtaining the matching degree between the target edge and the edge to be detected according to the marking points, thereby improving the detection efficiency while determining the monitoring accuracy.

[0031] 2. The present invention uses an intelligent algorithm to determine the presence of potential tearing and calls on the more expensive structured light technology for detailed analysis, avoiding unnecessary full processing of all areas, thereby improving overall efficiency and cost-effectiveness and effectively reducing computing resource consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0033] Figure 1 It is a method flow chart of steps S1 to S4 in a conveyor belt abnormality monitoring method based on image processing in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0035] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0036] Reference Figure 1 A conveyor belt abnormality monitoring method based on image processing includes steps S1 to S4, which are as follows:

[0037] S1: Acquire the surface image of the sampling frame when the conveyor belt is conveying according to the preset interval and perform grayscale processing to obtain the grayscale image, and perform edge detection, take any sampling frame as the target frame, obtain the target edge of the target frame and the edge to be detected of other frame images, classify the gradient direction difference in the grayscale image corresponding to the sampling frame, and obtain the frequency of each classification, which is used to obtain the regularity of the target edge.

[0038] It should be noted that ordinary cameras (marked as camera 1) are installed on both sides of the conveyor belt, and the camera required for structured light technology (marked as camera 2) is installed at a distance behind the ordinary cameras. The images of the conveyor belt surface are captured in real time by ordinary cameras installed on both sides of the conveyor belt; the preset interval is to sample 1 frame every 3 frames, and a total of 20 frames of images after the current sampling frame are sampled, which can be adjusted according to actual conditions; the Canny edge detection algorithm is used to detect the edges in the grayscale image, and the edges in a frame of sampled image are called edges to be detected.

[0039] Get the gradient direction of each pixel in the grayscale image corresponding to the sampling frame, where the value range of the gradient direction is , divide the direction value range into several intervals, and each interval is used as a level of gradient direction.

[0040] Exemplarily, in this embodiment, the Sobel operator is used to obtain the gradient direction of each pixel in the grayscale image, which is divided into four levels: Level 1: ; Level 2: ; Level 3: ; Level 4: ;in, Indicates the first Pixels and The gradient direction difference of each pixel.

[0041] Calculate the mean of the gradient direction differences of all adjacent pixels on the target edge and the edge to be detected, and perform normalization processing;

[0042] The level frequency of each gradient direction difference is counted, the entropy of each level frequency is calculated, and the product of the entropy and the normalized mean is used as the regularity of the target edge.

[0043] Specifically, the regularity satisfies the following relationship:

[0044] ;

[0045] In the formula, represents the regularity of the target edge, Indicates The level frequency of gradient direction differences, Indicates the first Pixels and The gradient direction difference of each pixel is represents the maximum value function, Represents the total number of pixels on the edge of the target. Represents a logarithmic function.

[0046] It needs to be explained that It is for Perform normalization processing, The entropy of the gradient direction difference reflects the degree of disorder of the gradient direction difference, that is, the overall regularity of the edge. The larger the value, the more chaotic the gradient change on the target edge. The larger it is, the greater the gradient change between the two adjacent pixels, and the smaller the regularity of the target edge.

[0047] Further explanation: regularity is an important property to distinguish dirt from tears. Dirt is generally less regular than tears. On conveyor belts, dirt usually refers to dust, mud, oil stains or other contaminants attached to the surface of the conveyor belt. The shapes and sizes of these dirt are often irregular because they may be composed of many different substances and may be dispersed or reorganized due to wind, friction or other external forces during the conveying process. In contrast, tears are the destruction of the structure of the conveyor belt itself. This destruction usually appears as linear or quasi-linear gaps because the tears occur along the fiber or layered structure of the conveyor belt. Due to the linear nature of tears, they have a high degree of regularity visually. Even under different lighting and angles, the edges of the tears are still relatively clear and linear. In contrast, the irregularity of dirt means that their edges are blurred, the shapes are varied, and they lack consistent geometric features.

[0048] Further analysis shows that for the detected edges, in addition to measuring regularity, it is also necessary to measure the matching of the edges in multiple frames. Since the texture of the conveyor belt surface has strong regularity and continuity, and the dirt and tearing on the surface generally only exist in parts of the conveyor belt, the shape and distribution are relatively random and irregular. Therefore, in continuous multiple frames of images, the dirty and torn parts will quickly leave the image screen as the conveyor belt moves after entering the image screen, resulting in that they cannot always match well in multiple frames of images; while the texture of the conveyor belt surface has strong continuity and regularity, it will always exist in the image screen and can be well matched in multiple frames of images.

[0049] S2: Preset marking points for the target edge and the edge to be detected respectively, and use the regularity of the target edge as an adjustment factor to calculate the similarity of the positional relationship of the marking points between the target edge and the edge to be detected to obtain the most matching edge.

[0050] Specifically, the steps for presetting the marking points are as follows:

[0051] In response to the target edge being an unclosed edge, marking two breakpoints of the unclosed edge as point 1 and point 2, marking the midpoint of the edge as point 3, and marking the midpoint between two adjacent points as point 4 and point 5;

[0052] In response to the target edge being a closed edge, the centroid of the closed edge is marked as point 3, the pixel point farthest from point 3 is marked as point 1, a straight line is determined based on point 1 and point 3, and the straight line is extended in the opposite direction of point 1 to the point where the closed edge intersects is marked as point 2, and the midpoints between the two adjacent points are marked as points 4 and 5.

[0053] To further illustrate, multiple frames of images after the target frame are extracted at preset intervals, the target edge is obtained based on the target frame, and multiple edges to be detected are obtained based on the multiple frames of images after the target frame. Step S2 matches the position relationship of the marking points between the target edge and the edge to be detected for the target frame and the multiple frames of images after the target frame, and then obtains the most matching edges of the target frame and the multiple frames of images after the target frame, respectively.

[0054] Calculate the average value of the cumulative sum of the distance between each pair of marked points on the target edge and the distance difference between each pair of marked points on the edge to be detected, and obtain the shape similarity of the target edge and the edge to be detected; calculate the average value of the distance between the position of the marked point on the target edge in the grayscale image and the position of the marked point on the edge to be detected in the grayscale image, and obtain the position similarity of the target edge and the edge to be detected;

[0055] The product of shape similarity and position similarity is normalized, and the inverse of regularity is used as the normalized exponential power to obtain the approximation between the target edge and the edge to be detected, and the edge to be detected with the greatest approximation is used as the edge that best matches the target edge.

[0056] Specifically, the approximation between the target edge and the edge to be detected satisfies the following relationship:

[0057] ;

[0058] In the formula, Indicates the target edge and the The approximation of the edge to be detected, represents the first The marking point and The distance between the marking points, Indicates The first The marking point and The distance between the marking points, represents the first The position of the marker in the image is The first The distance between the positions of the marker points in the image, represents the normalization function, Indicates the number of marking points. Indicates the total number of sample frame images, Indicates the regularity of the target edge.

[0059] That is to say, It reflects the proximity of the positions of the two edges in the image. The larger it is, the weaker the proximity of positions; The smaller it is, the closer the positions are; Indicates the distance between each key point on the target edge and the The difference between the distances of the key points on the edge to be detected reflects the similarity of the overall shapes of the two edges. The smaller it is, the more similar the overall shapes of the two edges are; The larger the value, the greater the difference in the overall shape of the two edges. Since the texture of the conveyor belt is continuous and regularly distributed, and the dirt or tear on the conveyor belt is generally less continuous and randomly distributed, it will make and The values ​​are too large.

[0060] Exemplary: Responding to regularity When it is large (high edge regularity), The smaller the value, the weaker the effect of the exponential power, the smaller the correction to the approximation, and the more original approximation score is retained; in response to regularity When it is small (low edge regularity), If is larger, the effect of the exponential power will be stronger, making a larger correction to the similarity, reducing the impact of the original similarity score.

[0061] S3: The gradient direction difference, grayscale difference and grayscale gradient difference of each pixel point between the target edge and the best matching edge are used to calculate the matching degree between the target edge and the best matching edge.

[0062] It should be explained that, since the number of pixels on the two edges may be inconsistent, the edge with fewer pixels is interpolated using bilinear interpolation so that the number of pixels on the two edges is the same, and each pixel is sorted.

[0063] Specifically, the matching degree between the target edge and the best matching edge satisfies the following relationship:

[0064] ;

[0065] In the formula, Indicates the matching degree between the target edge and the best matching edge, Represents the total number of pixels on the edge of the target. represents the first The pixel point corresponds to the first The difference in the gray gradient direction of pixels, represents the cosine function, Represents the first The pixel point corresponds to the first The difference in the gray value of pixels is Represents the first The pixel point corresponds to the first The difference in the grayscale gradient value of each pixel is Represents the normalization function.

[0066] That is to say, the stronger the match between two pixels, the smaller the difference in gray value. , the difference in grayscale gradient values The difference between the gradient direction should be smaller. When the matching degree of most pixels on the edge is strong, it will lead to the matching degree of the two edges. The larger the value, the better the match of the edge of the texture belonging to the conveyor belt. Larger values ​​indicate a dirty or torn edge match. The value is smaller.

[0067] S4: Assign weights to the matching degree of the most matching edge according to the image frame time sequence to obtain the significant matching of the target edge in multiple frames of images, and use the ratio of the regularity of the target edge to the significant matching as the score of the potential tearing of the target edge. Based on the score, determine whether there is an abnormality in the conveyor belt.

[0068] Specifically, the significant matching of the target edge in multiple frames of images satisfies the following relationship:

[0069] ;

[0070] In the formula, Indicates the significant matching of target edges in multiple frames. Indicates the total number of sample frame images, Indicates the sequence number of the frame to which the best matching edge belongs, Indicates the target edge and the The matching degree of the most matching edge in the sample frames.

[0071] Further explanation: 20 most matching edges of the target edge are obtained from the 20 frames sampled after the target frame where the target edge is located. Different weights are assigned to these 20 most matching edges according to the time sequence of the image frames to which they belong, and finally the matching of the target edge in multiple frames is obtained. The weight of the matching degree of the most matching edge of the image frame is set according to the time sequence, because the interval time between each frame image is short, the movement speed of the conveyor belt is limited, and the adjacent frame images themselves are highly similar. Even for a small area of ​​tearing, the adjacent frame images may contain the feature, but after a large number of frame images, the feature will disappear. For the texture of the conveyor belt, even if there are many frame images, its feature matching is still very good.

[0072] Specifically, the score of potential tearing of the target edge satisfies the following relationship:

[0073] ;

[0074] In the formula, The score indicating the target edge is a potential tear. represents the regularity of the target edge, Indicates the significant matching of target edges in multiple frame images.

[0075] That is, in terms of texture, dirt and tear characteristics, the matching The performance on the surface should be that the texture matches well, the dirt and tear matches poorly; The performance should be that the texture is more regular, the tearing is more regular, and the dirt is less regular.

[0076] In response to the score being greater than or equal to a preset threshold, there is an abnormality in the conveyor belt, the target edge is potentially torn, and the line structured light technology is activated for detection; conversely, when the score is less than the preset threshold, the conveyor belt is normal.

[0077] Further explanation: the preset threshold is 1.5, which can be adjusted by the implementer according to the specific situation. Among them, line structured light is a 3D scanning technology that obtains the 3D information of the object by projecting one or more laser lines onto the surface of the object and using the camera to capture the deformation of these lines. When the laser line shines on the surface of the object, the line will be distorted or broken in the camera image due to the uneven surface. By analyzing these distortion or fracture patterns, the 3D coordinates of each point on the surface of the object can be calculated.

[0078] Specifically, the steps for starting the line structured light technology for detection are as follows:

[0079] Calculate the distance between camera 1 and camera 2, and ensure that the line connecting camera 1 and camera 2 is parallel to the straight line where the conveyor belt is located, as well as the running speed of the conveyor belt at this time. When the edge detected in the image captured by camera 1 is a potential tear, the frame image with the most complete edge displayed in the image frame captured by camera 1 is recorded as the image to be confirmed for the potential tear. Calculate how long it will take for the area to appear in the image captured by camera 2 after camera 1 captures the frame image. At this moment, start the line structured light and camera 2 to detect the area where the potential tear is located. The line structured light technology is a prior art and will not be described in detail here.

[0080] The system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the conveyor belt abnormality monitoring method based on image processing according to the first aspect of the present invention is implemented.

[0081] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0082] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0083] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0084] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A conveyor belt abnormality monitoring method based on image processing, characterized in that: include: The surface image of the sampling frame during conveying of the conveyor belt is obtained according to the preset interval and grayscale processing is performed to obtain the grayscale image, and edge detection is performed. Any sampling frame is used as the target frame to obtain the target edge of the target frame and the edges to be detected of other frame images. The gradient direction differences in the grayscale images corresponding to all sampling frames are graded to obtain the frequency of each grade, which is used to obtain the regularity of the target edge, including: Calculate the mean of the gradient direction differences of all adjacent pixels on the target edge and the edge to be detected, and perform normalization processing; Counting the frequency of each classification of gradient direction difference, calculating the entropy of each classification frequency, and taking the product of the entropy and the normalized mean as the regularity of the target edge; Presetting marking points for the target edge and the edge to be detected respectively, taking the regularity of the target edge as an adjustment factor to calculate the similarity of the positional relationship of the marking points between the target edge and the edge to be detected, and obtaining the most matching edge; The gradient direction difference, grayscale difference and grayscale gradient difference of each pixel point between the target edge and the best matching edge are used to calculate the matching degree between the target edge and the best matching edge; Assign weights to the matching degree of the best matching edge according to the image frame time sequence to obtain the significant matching of the target edge in multiple frames of images. , , where Indicates the total number of sample frame images, Indicates the sequence number of the frame to which the best matching edge belongs, Indicates the target edge and the The matching degree of the most matching edge in the sampling frames is determined; the ratio of the regularity of the target edge to the significant matching is used as the score of the potential tear of the target edge, and whether there is an abnormality in the conveyor belt is determined based on the score.

2. The conveyor belt abnormality monitoring method based on image processing according to claim 1 is characterized in that: The step of grading the gradient direction differences in the grayscale images corresponding to all the sampling frames comprises: Get the gradient direction of each pixel in the grayscale image corresponding to the sampling frame, where the value range of the gradient direction is , divide the direction value range into several intervals, and each interval is used as a level of gradient direction.

3. The conveyor belt abnormality monitoring method based on image processing according to claim 1 is characterized in that: The step of respectively presetting marking points for the target edge and the edge to be detected comprises: In response to the target edge being an unclosed edge, marking two breakpoints of the unclosed edge as point 1 and point 2, marking the midpoint of the edge as point 3, and marking the midpoint between two adjacent points as point 4 and point 5; In response to the target edge being a closed edge, the centroid of the closed edge is marked as point 3, the pixel point farthest from point 3 is marked as point 1, a straight line is determined based on point 1 and point 3, and the straight line is extended in the opposite direction of point 1 to the point where the closed edge intersects is marked as point 2, and the midpoints between the two adjacent points are marked as points 4 and 5.

4. The conveyor belt abnormality monitoring method based on image processing according to claim 1 is characterized in that: The best matching edge includes: Calculate the average value of the cumulative sum of the distance between each pair of marked points on the target edge and the distance difference between each pair of marked points on the edge to be detected, and obtain the shape similarity of the target edge and the edge to be detected; calculate the average value of the distance between the position of the marked point on the target edge in the grayscale image and the position of the marked point on the edge to be detected in the grayscale image, and obtain the position similarity of the target edge and the edge to be detected; The product of the shape similarity and the position similarity is normalized, and the reciprocal of the regularity is used as the normalized exponential power to obtain the approximation between the target edge and the edge to be detected, and the edge to be detected with the greatest approximation is used as the edge that best matches the target edge.

5. The conveyor belt abnormality monitoring method based on image processing according to claim 1 is characterized in that: The matching degree satisfies the following relationship: ； In the formula, Indicates the matching degree between the target edge and the best matching edge, Represents the total number of pixels on the edge of the target. represents the first The pixel point corresponds to the first The difference in the gray gradient direction of pixels, represents the cosine function, represents the first The pixel point corresponds to the first The difference in the gray value of pixels is represents the first The pixel point corresponds to the first The difference in the grayscale gradient value of each pixel is Represents the normalization function.

6. The conveyor belt abnormality monitoring method based on image processing according to claim 1 is characterized in that: Determine whether there is any abnormality in the conveyor belt based on the score, including: In response to the score being greater than or equal to a preset threshold, there is an abnormality in the conveyor belt, the target edge is potentially torn, and the line structured light technology is activated for detection; conversely, when the score is less than the preset threshold, the conveyor belt is normal.

7. A conveyor belt abnormality monitoring system based on image processing, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the conveyor belt abnormality monitoring method based on image processing according to any one of claims 1 to 6 is implemented.

Images