A computer vision-based belt conveyor longitudinal tear detection system and method

Abstract

A kind of longitudinal tear detection method of belt conveyor based on computer vision, reflective strip is set at equal intervals on the non-load surface of the conveying belt of belt conveyor, light source, ccd camera and photodiode are set on the rack.The reflection of reflective strip is detected by photodiode to trigger the camera to shoot the non-load surface of the conveying belt, the digital mark set on the conveying belt is used to distinguish the area that cannot be distinguished, the photos of each area are compared with the initial photo to determine whether there is change, if there is abnormality, the state of the changed area is identified by artificial intelligence using depth vision technology.The traditional picture comparison technology is combined with computer vision technology, and the detection efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of detecting longitudinal tear of belt conveyor, and particularly relates to a longitudinal tear detection method of belt conveyor based on computer vision and a detection device thereof. BACKGROUND

[0002] In the mining, port and metallurgical industries, the belt conveyor is an important equipment for material conveying. The conveying belt of the belt conveyor may be longitudinally torn due to material impact, equipment wear and other reasons in the long-term operation process. The traditional detection method mainly relies on manual inspection, which is not only low in efficiency, but also prone to missed detection. Therefore, it is of great practical significance to develop an efficient and automatic detection method. SUMMARY

[0003] The present application aims to provide a longitudinal tear detection method of belt conveyor based on computer vision, which improves the detection efficiency and accuracy of longitudinal tear of belt conveyor through computer vision technology. In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0004] A longitudinal tear detection method of belt conveyor based on computer vision, comprising the following steps:

[0005] Step 1: setting reflective strips 1 at equal intervals on the non-load bearing surface of the conveying belt of the belt conveyor, and dividing the non-load bearing surface into multiple areas;

[0006] Step 2: setting a light source 2, a CCD camera 3 and a photodiode 4, and a roller brush 7 on the rack of the belt conveyor, the field of view of the CCD camera 3 can completely cover the non-load bearing surface between the adjacent two reflective strips 1 to collect a complete image of an area of the non-load bearing surface, the light emitted by the light source 2 can be received by the photodiode 4 after being reflected by the reflective strips 1, and the non-load bearing surface is first cleaned by the roller brush 7 and then illuminated by the planar light spot projected by the light source 2 during movement;

[0007] Step 3: using the photodiode 4 to detect the reflection of the reflective strips 1 to trigger the CCD camera to shoot the non-load bearing surface of the conveying belt, and the image of the non-load bearing surface of the conveying belt shot by the CCD camera 3 is transmitted to a field computer 5 for processing;

[0008] Step 4: initial image acquisition and storage, when the system is first operated, the initial image of each area of the non-load bearing surface of the conveying belt is acquired as a reference for subsequent comparison;

[0009] Step 5: During the subsequent operation of the belt conveyor, image acquisition and real-time image comparison are performed. The on-site computer 5 uses an image comparison algorithm to compare the real-time acquired area image with the initial image to detect whether there is a significant change. If there is a significant change, step 6 is entered, otherwise step 5 is continued;

[0010] Step 6: Abnormal identification is performed on the area image with significant changes. The on-site computer 5 transmits the area image with significant changes to the server 6. The server 6 identifies the area image with significant changes to determine whether it is a longitudinal tearing fault. If there is a longitudinal tearing fault, step 7 is entered, otherwise step 5 is entered;

[0011] Step 7: Longitudinal tearing alarm. When the server 6 identifies that the belt conveyor has a longitudinal tearing fault, it sends an alarm signal and records the fault area and time for maintenance personnel to handle in a timely manner.

[0012] Further, the computer vision-based longitudinal tearing detection method of the belt conveyor according to the present application preferably comprises the following sub-steps in step 5:

[0013] Sub-step 5.1: Before image comparison, the non-load-bearing surface image of the conveying belt is pre-processed to improve the accuracy and efficiency of the comparison.

[0014] Sub-step 5.2: Compare the real-time acquired image with the stored initial image to detect whether there is a change.

[0015] Further, the computer vision-based longitudinal tearing detection method of the belt conveyor according to the present application preferably comprises the following operations in sub-step 5.1:

[0016] Grayscale: Convert the color image output by the CCD camera to a grayscale image to simplify the subsequent processing steps.

[0017] Further, the computer vision-based longitudinal tearing detection method of the belt conveyor according to the present application preferably comprises the following operations in sub-step 5.2:

[0018] Image alignment: Calculate the transformation parameters according to the reflection strips 1 in the image, and transform the images according to the calculated transformation parameters to align the two images in space position for comparison.

[0019] Calculate the difference between the two images: Calculate the difference between the two aligned images. Each pixel value in the difference image reflects the degree of change of the corresponding pixels in the two images, which can directly show the difference between the two images.

[0020] Thresholding: thresholding the difference image to extract significant change regions, by setting pixel values less than a change threshold to 0 and pixel values greater than the change threshold to 1;

[0021] Statistical different pixels: counting the number of pixels with a value of 1 in the thresholded image, which is the number of different pixels, if this number exceeds a pre-set threshold of different pixels, it is determined that the conveying belt is in an abnormal state.

[0022] Further, the longitudinal tearing detection method of the belt conveyor based on computer vision, preferably, the step 6 comprises the following operations: a sub-step 6.1 of pre-processing the region picture with obvious changes, and a sub-step 6.2 of detecting longitudinal cracks in the pre-processed picture.

[0023] Further, the longitudinal tearing detection method of the belt conveyor based on computer vision, preferably, the sub-step 6.1 comprises the following operations:

[0024] Grayscale: converting the image output by the CCD camera 3 into a grayscale image;

[0025] Noise removal: using Gaussian filtering and median filtering techniques to remove noise in the image, enhancing the clarity of the image, and providing a clearer image for subsequent edge detection and recognition operations;

[0026] Edge detection: using edge detection algorithms to extract edge information in the image, including noise suppression, gradient calculation, non-maximum suppression and double threshold detection;

[0027] Binaryzation: converting the image into a binary image through threshold segmentation.

[0028] Further, the longitudinal tearing detection method of the belt conveyor based on computer vision, preferably, the sub-step 6.2 comprises the following operations:

[0029] Grid division: dividing the binary image into N* N grids, and the size of each grid cell is k* k pixels;

[0030] Node initialization: calculating the gray mean value of each grid cell, if the mean value is greater than a pre-set threshold T, marking it as a crack node with a value of 1, otherwise marking it as a background node with a value of 0;

[0031] Node connectivity calculation: using 8-connectivity rules to calculate the connectivity of the nodes, for each crack node, checking its connectivity with adjacent nodes, if connected, marking it as a crack region;

[0032] Crack morphology analysis: calculate the length of the crack in the crack region and calculate the width of the crack;

[0033] Crack detection: calculate the ratio of the length of the crack to the width of the crack, when the ratio of the length of the crack to the width of the crack is greater than a threshold U and the length of the crack is greater than a threshold V, it is determined that the belt conveyor appears longitudinal tearing.

[0034] Further, the computer vision-based longitudinal tearing detection method of the belt conveyor, preferably the node connectivity calculation includes the following operations:

[0035] Initialize a label matrix to record the crack region to which each node belongs.

[0036] Iterate through all nodes, and for unmarked crack nodes, assign a new label.

[0037] Recursively or iteratively check its adjacent nodes, and if the adjacent nodes are crack nodes and are not marked, assign the same label.

[0038] Further, the computer vision-based longitudinal tearing detection method of the belt conveyor, preferably the crack morphology analysis includes the following detailed steps:

[0039] From top to bottom, scan the connected domain, record the position of the first white pixel point as (x1, y1),

[0040] From bottom to top, scan the connected domain, record the position of the first white pixel point as (x2, y2),

[0041] From left to right, scan the connected domain, record the position of the first white pixel point as (x3, y3),

[0042] From right to left, scan the connected domain, record the position of the first white pixel point as (x4, y4);

[0043] Xleft =min{x1, x2, x3, x4} ,

[0044] Xright =max{x1, x2, x, x} ,

[0045] Yup = min{y1, y 2, y 3, y 4} ,

[0046] Ydown =max{y1, y2, y3, y4};

[0047] The length of the crack is Ydown-Yup, and the width of the crack is Xright-Xleft.

[0048] Further, the computer vision-based longitudinal tear detection method of the belt conveyor according to the present application preferably comprises the following steps:

[0049] adjusting the image to a fixed size for input into the deep learning model for processing;

[0050] normalizing the image to map the pixel value range to the interval [0, 1], improving the training effect and convergence speed of the model;

[0051] enhancing the image, such as random rotation, flipping, cropping, adding noise, etc., to increase the diversity of the data and prevent overfitting of the model.

[0052] Further, the computer vision-based longitudinal tear detection method of the belt conveyor according to the present application preferably comprises the following steps:

[0053] selecting a convolutional neural network (CNN) as the basic architecture to build a deep learning model, and setting two output nodes in the output layer for predicting whether it is a longitudinal tear and the severity of the tear;

[0054] dividing the pre-collected conveyor belt images into training and test sets according to a certain proportion for model training and evaluation;

[0055] inputting the training set into the deep learning model, calculating the output of the model through forward propagation, then calculating the loss function value according to the difference between the output and the true label, then calculating the gradient of the loss function with respect to the model parameters through the backpropagation algorithm, and updating the model parameters using the optimization algorithm, repeating this process until the accuracy and loss value of the model on the training set meet the preset requirements or the training round ends;

[0056] after processing the images that need to be detected for longitudinal tear according to the same preprocessing method as the training set, inputting them into the trained deep learning model, and outputting the prediction results of whether it is a longitudinal tear and the prediction results of the severity of the tear.

[0057] Further, the computer vision-based longitudinal tear detection method of the belt conveyor according to the present application preferably comprises the following steps:

[0058] Meanwhile, the present application also provides a detection device for the above-mentioned longitudinal tear detection method of the belt conveyor, which comprises a reflective strip 1, a light source 2, a CCD camera 3, a photodiode 4, a field computer 5, a server 6, and a roller brush 7.

[0059] A light reflection strip 1 is arranged on the non-load bearing surface of the conveying belt of the belt conveyor at a distance L, and the distance L is the same as the width of the conveying belt;

[0060] The light source 2 is a high-brightness LED lamp, which can project a planar light spot on the non-load bearing surface of the conveying belt of the belt conveyor, and the planar light spot can uniformly illuminate the non-load bearing surface between the adjacent two light reflection strips 1;

[0061] The CCD camera 3 is an industrial camera with high resolution, and the installation position of the CCD camera 3 is adjusted so that the field of view of the CCD camera 3 can completely cover the non-load bearing surface between the adjacent two light reflection strips 1;

[0062] The light source 2 and the photodiode 4 are respectively arranged on the left and right sides of the CCD camera 3, so that the light emitted by the light source 2 can be received by the photodiode 4 after being reflected by the light reflection strip 1;

[0063] The rolling brush 7 is arranged on one side of the light source 2, so that the non-load bearing surface of the conveying belt is first cleaned by the rolling brush 7 and then illuminated by the planar light spot projected by the light source 2 during movement, so as to avoid dust and the like adhering to the non-load bearing surface of the conveying belt and interfering with subsequent longitudinal tearing detection;

[0064] The field computer 5 is electrically connected with the light source 2, the CCD camera 3 and the photodiode 4, and the field computer 5 can control the opening and closing of the light source 2 through a software program, trigger the CCD camera 3 to shoot and receive the image output by the CCD camera 3;

[0065] The field computer 5 is connected with the server 6 through a network.

[0066] Further, the detection device of the application preferably causes the light emitted by the light source 2 to be diffusely reflected on the non-load bearing surface of the conveying belt and to be specularly reflected on the light reflection strip 1, so that the electric signal output by the photodiode 4 increases when the light reflection strip 1 passes near the photodiode 4, and the field computer 5 can obtain a trigger signal by detecting the rising edge of the received electric signal output by the photodiode 4, and control the CCD camera 3 to shoot a picture according to the trigger signal.

[0067] The application has the following beneficial effects:

[0068] 1. In this invention, the light emitted by the light source 2 undergoes diffuse reflection on the non-load-bearing surface of the conveyor belt and specular reflection on the reflective strip 1. This increases the amount of light that can be received by the photodiode 4 when the reflective strip 1 passes near the photodiode 4, thereby increasing the electrical signal output by the photodiode 4. The on-site computer 5 detects the rising edge of the received electrical signal output by the photodiode 4 to obtain a trigger signal, and controls the CCD camera 3 to capture an image based on the trigger signal.

[0069] 2. This invention can quickly identify whether changes have occurred on the surface of the conveyor belt by performing real-time image comparison on the on-site computer 5, thereby detecting possible tears or other abnormalities. The server 6 can further identify the changed areas and use complex computer vision algorithms for offline analysis using the abundant computing resources on the server 6 to determine whether it is a longitudinal tear fault. Through the above methods, efficient and automated detection of longitudinal tears in conveyor belts can be achieved, which not only improves the accuracy and efficiency of detection, but also allows one server 6 to host multiple on-site computers 5, enabling simultaneous detection of multiple conveyor belts.

[0070] 3. This invention sets identifiers on the conveyor belt to distinguish different areas. Each identifier corresponds to a specific area. Automatic identification of identifiers in the image can determine the specific area corresponding to the current image. By setting identifiers in each area, the system can perform image comparison and anomaly detection more efficiently, which not only improves the accuracy and speed of detection, but also provides convenience for fault location and maintenance. Attached Figure Description

[0071] Appendix Figure 1 This is a schematic diagram of a computer vision-based longitudinal tear detection device for belt conveyors.

[0072] Appendix Figure 2 This is a flowchart of a computer vision-based method for detecting longitudinal tears in belt conveyors. Detailed Implementation

[0073] The present invention will now be described in further detail with reference to the accompanying drawings and preferred embodiments. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention and therefore showing only the components relevant to the invention. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0074] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set" and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or a mutual cooperation. They can be mechanical or electrical connections. They can be direct connections or indirect connections through an intermediate medium, and can refer to the internal communication of two components. Furthermore, in this document, "coupled" between two devices means that a laser beam emitted from one device is incident on the other device. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. In addition, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0075] Example 1:

[0076] As attached Figure 1 As shown, a computer vision-based longitudinal tear detection device for belt conveyors includes a reflective strip 1, a light source 2, a CCD camera 3, a photodiode 4, a field computer 5, a server 6, and a roller brush 7, all installed and debugged according to the method of this invention. After installation, the system is started for testing. When the photodiode detects the reflection from the reflective strip, it triggers the CCD camera to take a picture. The system compares the captured image with the initial image. If obvious areas of change are found in the image, further analysis is performed on these areas.

[0077] Select reflective strips 1 of appropriate width and length, and install one reflective strip 1 at intervals L on the non-load-bearing surface of the conveyor belt of the belt conveyor. The interval L is the same as the width of the conveyor belt. The length-to-width ratio of the reflective strip 1 is greater than 10, preferably greater than 20.

[0078] The light source 2 uses a high-brightness LED lamp, which can project a planar light spot and illuminate the non-load-bearing surface of the conveyor belt of the belt conveyor. The planar light spot can evenly illuminate the non-load-bearing surface between two adjacent reflective strips 1.

[0079] The CCD camera 3 is an industrial camera with high resolution. The installation position of the CCD camera 3 is adjusted so that the field of view of the CCD camera 3 can completely cover the non-load-bearing surface between two adjacent reflective strips 1.

[0080] The light source 2 and the photodiode 4 are respectively installed on the left and right sides of the CCD camera 3, so that the light emitted by the light source 2 can be received by the photodiode 4 after being reflected by the reflective strip 1.

[0081] The roller brush 7 is positioned on one side of the light source 2, so that the non-load-bearing surface of the conveyor belt is first cleaned by the roller brush 7 during the movement and then illuminated by the planar light spot projected by the light source 2, so as to avoid dust and other particles adhering to the non-load-bearing surface of the conveyor belt and interfering with the subsequent longitudinal tear detection.

[0082] The on-site computer 5 is electrically connected to the light source 2, CCD camera 3, and photodiode 4. The on-site computer 5, through software settings, can control the on / off state of the light source 2, trigger the CCD camera 3 to capture images, and receive the images output by the CCD camera 3. The on-site computer 5 is connected to the server 6 via a network.

[0083] Because the light emitted by the light source 2 undergoes diffuse reflection on the non-load-bearing surface of the conveyor belt and specular reflection on the reflective strip 1, the amount of light that can be received by the photodiode 4 when the reflective strip 1 passes near the photodiode 4 increases, thereby increasing the electrical signal output by the photodiode 4. The on-site computer 5 detects the rising edge of the received electrical signal output by the photodiode 4 to obtain a trigger signal, and controls the CCD camera 3 to take pictures based on the trigger signal.

[0084] As attached Figure 2 As shown, the computer vision-based method for detecting longitudinal tears in belt conveyors includes the following steps:

[0085] Step 1: Set reflective strips 1 at equal intervals on the non-load-bearing surface of the conveyor belt to divide the non-load-bearing surface into multiple areas.

[0086] Preferably, a high-reflectivity material is used for the reflective strip 1, such as reflective tape or reflective paint. The material should be wear-resistant and weather-resistant to adapt to the working environment of the conveyor belt. The reflective strips 1 are installed at equal intervals on the non-load-bearing surface of the conveyor belt. One reflective strip 1 is placed at regular intervals to ensure that each area can be accurately detected. The reflective strip 1 is securely attached to the non-load-bearing surface of the conveyor belt using strong adhesive or other fixing devices to ensure that it does not fall off during conveyor belt operation.

[0087] Step 2: Set up the light source 2, CCD camera 3, photodiode 4, and roller brush 7 on the frame.

[0088] Light source 2 preferably uses a high-brightness LED light as its source due to its advantages of high brightness, low energy consumption, and long lifespan. Mounted above the conveyor belt frame, the light source projects a planar light spot that illuminates the non-load-bearing surface of the conveyor belt. This planar light spot evenly illuminates the non-load-bearing surface between two adjacent reflective strips 1. To ensure uniform illumination of the reflective strips 1, the angle of the light source should be adjusted according to site conditions to achieve the best reflective effect. The brightness and on / off control of the light source should be configured to adapt to different ambient lighting conditions, ensuring that the reflective strips are clearly identifiable under various lighting conditions.

[0089] The CCD camera 3 is preferably a high-resolution industrial camera to ensure the capture of clear image details. The CCD camera 3 is mounted on a rack, positioned at a certain distance from the reflective strip 1. The mounting position of the CCD camera 3 is adjusted, and the mounting height and angle are determined based on the width of the conveyor belt and the required inspection accuracy, so that the field of view of the CCD camera 3 completely covers the non-load-bearing surface between two adjacent reflective strips 1, ensuring that a complete image of a region of the non-load-bearing surface can be acquired.

[0090] The photodiode 4 is preferably a photodiode with fast response speed and high sensitivity to ensure accurate detection of the reflection from the reflective strip. The light source 2 and the photodiode 4 are respectively mounted on the left and right sides of the CCD camera 3, so that the light emitted by the light source 2 can be received by the photodiode 4 after being reflected by the reflective strip 1.

[0091] The roller brush 7 is positioned on one side of the light source 2, so that the non-load-bearing surface of the conveyor belt is first cleaned by the roller brush 7 during the movement and then illuminated by the planar light spot projected by the light source 2, so as to avoid dust and other particles adhering to the non-load-bearing surface of the conveyor belt and interfering with the subsequent longitudinal tear detection.

[0092] Step 3: Use photodiode 4 to detect the reflection of reflective strip 1 to trigger the CCD camera to photograph the non-load-bearing surface of the conveyor belt.

[0093] When reflective strip 1 passes near photodiode 4, the amount of light received by photodiode 4 increases, causing the output electrical signal of photodiode 4 to increase. The on-site computer 5 detects the rising edge of the received output electrical signal from photodiode 4 to obtain a trigger signal, and uses this trigger signal to control CCD camera 3 to capture an image. The CCD camera 3 captures an image of the non-load-bearing surface of the conveyor belt and transmits it to the on-site computer 5 for processing. The image captured by CCD camera 3 under the trigger of the photodiode's output signal should cover the complete area between two adjacent reflective strips 1.

[0094] Step 4: Initial Image Acquisition and Storage

[0095] When the system is first run, initial images of each region on the non-load-bearing surface of the conveyor belt are acquired as a reference for subsequent comparisons.

[0096] Step 5: During the operation of the belt conveyor, image acquisition and real-time image comparison are performed.

[0097] When the reflective strip 1 passes near the photodiode 4, the electrical signal output by the photodiode 4 increases. The on-site computer 5 detects the rising edge of the received electrical signal output by the photodiode 4 to obtain a trigger signal, and controls the CCD camera 3 to take pictures based on the trigger signal.

[0098] The on-site computer 5 uses an image comparison algorithm to compare the real-time acquired image with the initial image to detect any significant changes, such as tearing or damage. If significant changes are found, proceed to step 6; otherwise, continue with step 5.

[0099] Step 6: Perform anomaly identification on images showing obvious changes.

[0100] If an abnormal change is detected, the on-site computer 5 transmits an image of the area with obvious changes to the server 6. The server 6 then uses this image to further identify the changed area and determine whether it is a longitudinal tear fault. If a longitudinal tear fault exists, the process proceeds to step 7; otherwise, it proceeds to step 5.

[0101] Step 7: Longitudinal tear alarm.

[0102] Once a longitudinal tear fault is detected in the belt conveyor, server 6 issues an alarm signal and records the fault area and time so that maintenance personnel can handle it in a timely manner.

[0103] By comparing images in real time on the on-site computer 5, it is possible to quickly identify whether changes have occurred on the surface of the conveyor belt, thereby detecting possible tears or other abnormalities.

[0104] By using server 6 to further identify the changed areas, the abundant computing resources on server 6 can be used to perform offline analysis using complex computer vision algorithms to determine whether it is a longitudinal tear fault.

[0105] The above methods enable efficient and automated detection of longitudinal tears in conveyor belts, improving both accuracy and efficiency. Furthermore, a single server can host multiple field computers, allowing simultaneous detection of multiple conveyor belts.

[0106] Example 2:

[0107] Based on the scheme of Example 1, step 5 further includes the following detailed sub-steps:

[0108] Sub-step 5.1: Before performing image comparison, preprocess the acquired images of the non-load-bearing surface of the conveyor belt to improve the accuracy and efficiency of the comparison. Preprocessing preferably includes the following operations:

[0109] Grayscale conversion: Converting the color image output by the CCD camera into a grayscale image simplifies subsequent processing steps. This reduces computation and highlights structural features in the image. A common weighted averaging method can be used, for example, multiplying the red, green, and blue components by different weights (such as 0.299, 0.587, and 0.114) according to the human eye's sensitivity to different colors, and then summing them to obtain the grayscale value.

[0110] Geometric correction: In actual belt conveyor applications, factors such as vibration during belt conveyor operation may cause geometric distortion in the acquired images. Geometric correction is achieved through image registration technology, ensuring that images acquired at different times maintain consistent geometric shape.

[0111] Sub-step 5.2: Compare the real-time acquired images with the stored initial images to detect any changes. Preferably, this includes the following operations:

[0112] Image alignment: In practice, the acquired real-time image and the initial image may differ in position, angle, etc. Image alignment can adjust them to the same coordinate system. This process requires first determining the geometric transformation model, such as affine transformation or perspective transformation, then calculating the transformation parameters based on the reflection strip 1 in the image, and finally transforming the real-time image according to the calculated transformation parameters so that the two images being compared are aligned in spatial position.

[0113] Image difference calculation: After image alignment, the difference between the two aligned images is calculated. Each pixel value in the difference image reflects the degree of change between the pixels in the two images at the corresponding location. For example, if a pixel value in the difference image is 0, it means that the pixel value of the real-time image and the initial image is the same at that location; if the pixel value is not 0, it means that there is a change. By calculating the image difference, the differences between the two images can be directly displayed.

[0114] Thresholding: To ignore minor changes, the differential image is thresholded to extract regions of significant change. In actual conveyor belt inspection, slight environmental fluctuations or minor errors during image acquisition may cause minute changes in the differential image. These minor changes may not represent an actual abnormal condition of the conveyor belt. By setting an appropriate threshold, pixel values ​​below the threshold are set to 0, and pixel values ​​above the threshold are set to 1, resulting in a binary image. The regions with values ​​of 1 in this binary image are the regions of significant change. The choice of threshold depends on the specific application and noise level and needs to be adjusted according to the actual situation.

[0115] Counting distinct pixels: This involves counting the number of pixels with a value of 1 in the thresholded binary image; this represents the number of distinct pixels. By counting these distinct pixels, the degree of difference between the real-time image and the initial image can be quantified. If this number exceeds a preset threshold, it can be determined that the conveyor belt is in an abnormal state, such as tearing, wear, or foreign object adhesion.

[0116] As a preferred approach, a threshold of 5% of the total number of pixels is selected. That is, if the number of different pixels exceeds 10% of the total number of pixels, the conveyor belt is judged to be in an abnormal state.

[0117] Furthermore, the preferred threshold is 10% or 15% of the total number of pixels.

[0118] Example 3:

[0119] Based on the scheme of the aforementioned embodiment, step 6 further includes the following detailed sub-steps: server 6 preprocesses the image of the area with obvious changes and performs longitudinal crack detection to improve the accuracy and efficiency of longitudinal tear detection.

[0120] Sub-step 6.1 of server 6 preprocessing images of regions with significant changes preferably includes the following operations:

[0121] Grayscale conversion: Converting a color image output by a CCD camera into a grayscale image.

[0122] Noise Removal: During image acquisition, noise often exists in images due to environmental factors such as fine dust. This noise can interfere with accurate image analysis. Filtering techniques are used to remove noise from images, including Gaussian filtering and median filtering. Gaussian filtering smooths the image by weighted averaging of each pixel and its neighboring pixels, reducing the impact of noise. Median filtering sorts the pixel values ​​in a pixel's neighborhood and takes the median value as the new value for that pixel. Noise removal enhances image clarity, providing a clearer image foundation for subsequent edge detection and recognition operations.

[0123] Edge detection: Edge detection algorithms are used to extract edge information in an image. The Sobel algorithm detects edges by calculating the gradients of the image in the horizontal and vertical directions. It performs convolution operations on the horizontal and vertical directions of the image respectively, and then determines the edge pixels based on a certain threshold. The Canny algorithm is a more complex and accurate edge detection algorithm, which includes multiple steps such as noise suppression, gradient calculation, non-maximum suppression, and double threshold detection.

[0124] Binarization: The image is converted into a binary image (the cracked area is white and the background is black) by thresholding.

[0125] Sub-step 6.2 of server 6 performing longitudinal crack detection on the preprocessed image preferably includes the following operations:

[0126] Mesh generation: Divide the binary image into an N*N grid, with each grid cell being k*k pixels in size. The choice of grid size (k*k) should be based on the typical width of the crack.

[0127] Node initialization: Calculate the average gray value of each grid cell. If the average value is greater than a preset threshold (T), it is marked as a crack node (value 1); otherwise, it is marked as a background node (value 0). Set the seepage threshold (T) according to the connectivity of the crack, which is usually 0.5-0.8.

[0128] Node connectivity calculation: The connectivity of nodes is calculated using the 8-connectivity rule to simulate the crack propagation path. Starting from the top of the image, the crack propagation process is simulated. For each crack node, its connectivity with neighboring nodes is checked; if they are connected, it is marked as a crack region.

[0129] Crack morphology analysis: Calculate the length and width of the crack.

[0130] Crack detection: Calculate the ratio of crack length to crack width. When the ratio of crack length to width is greater than a threshold (U) and the crack length is greater than the threshold (V), it is determined that there is a longitudinal tear in the belt conveyor.

[0131] Example 4:

[0132] Based on the scheme of the aforementioned embodiments, the node connectivity calculation in step 6 further includes the following detailed sub-steps:

[0133] Iterate through each node and check its connectivity with its neighboring nodes using the 8-connectivity rule. Neighboring nodes include the four directions: up, down, left, and right, and the four diagonal directions: top left, top right, bottom left, and bottom right.

[0134] For the current node ((i, j)), check its 8 neighboring nodes ((i-1, j), (i+1, j), (i, j-1), (i, j+1), (i-1, j-1), (i-1, j+1), (i+1, j-1), (i+1, j+1)). If a neighboring node is a crack node (value 1), then the current node is considered to be connected to its neighboring nodes.

[0135] Use either the Region Growing algorithm or the Connected Component Labeling algorithm to label nodes belonging to the same crack region. The specific steps are as follows:

[0136] Initialize a label matrix to record the crack region to which each node belongs.

[0137] Iterate through all nodes, and for unlabeled crack nodes, assign a new label.

[0138] Recursively or iteratively check its neighboring nodes. If a neighboring node is a crack node and is not labeled, assign it the same label.

[0139] Example 5:

[0140] Based on the solution of the aforementioned embodiment, the crack morphology analysis in step 6 further includes the following detailed steps:

[0141] Perform a global scan on each of the obtained connected components.

[0142] Scan the connected components from top to bottom and record the position of the first white pixel as (x1, y1).

[0143] Scan the connected components from bottom to top, and record the position of the first white pixel as (x2, y2).

[0144] Scan the connected components from left to right and record the position of the first white pixel as (x3, y3).

[0145] Scan the connected components from right to left and record the position of the first white pixel as (x4, y4);

[0146] Xleft =min{x1, x2, x3, x4},

[0147] Xright =max{x1, x2, x, x},

[0148] Yup = min{y1, y 2, y 3, y 4},

[0149] Ydown =max{y1, y2, y3, y4};

[0150] Finally, the crack length = Ydown - Yup, and the crack width = Xright - Xleft.

[0151] Example 6:

[0152] Based on the scheme of the aforementioned embodiment, step 6 further includes the following detailed sub-steps: before performing image comparison and alignment, server 6 preprocesses the images of regions with obvious changes and uses a deep learning model to perform longitudinal tear detection, so as to improve the accuracy and efficiency of longitudinal tear detection.

[0153] Sub-step 6.1 of server 6 preprocessing images of regions with significant changes preferably includes the following operations:

[0154] The collected images are subjected to size standardization processing, preferably all images are adjusted to a fixed size to facilitate input into the deep learning model for processing;

[0155] Normalizing the image maps the pixel value range to the [0, 1] interval, improving the training effect and convergence speed of the model;

[0156] Image enhancements, such as random rotation, flipping, cropping, and adding noise, can increase data diversity and prevent model overfitting.

[0157] Sub-step 6.2 of server 6 performing longitudinal crack detection on the preprocessed image preferably includes the following operations:

[0158] When building a deep learning model, a Convolutional Neural Network (CNN) is chosen as the basic architecture because CNNs perform exceptionally well in image classification and feature extraction. The network structure of a Convolutional Neural Network (CNN) includes the following model architecture:

[0159] Input layer: Receives preprocessed image data.

[0160] Convolutional layers: These layers extract features by sliding multiple convolutional kernels across the image, capturing local texture and pattern information. For example, multiple convolutional layers can be set up, each using a different sized kernel, to extract features at different scales.

[0161] Pooling layers: These downsample the output of convolutional layers, reducing the spatial dimensionality of the data while preserving important feature information, thus reducing computational cost and the number of parameters. Common pooling operations include max pooling and average pooling.

[0162] Fully connected layer: Flatten the output of the pooling layer into a one-dimensional vector and perform a linear transformation to further extract global feature information.

[0163] Output Layer: Based on task requirements, two output nodes are set up to predict crack type (vertical) and crack severity (mild or severe), respectively. For crack type prediction, a binary cross-entropy loss function can be used; for crack severity prediction, a binary cross-entropy loss function or other suitable loss functions can be used (such as mean squared error loss function, if the severity is quantified as a continuous value).

[0164] The pre-collected conveyor belt images are divided into training and testing sets according to a certain ratio (e.g., 8:2) for model training and evaluation.

[0165] The training set is fed into the deep learning model, and the model's output is calculated through forward propagation. Then, the loss function value is calculated based on the difference between the output and the true label. Next, the gradient of the loss function with respect to the model parameters is calculated using the backpropagation algorithm, and the model parameters are updated using an optimization algorithm. This process is repeated until the model's accuracy and loss value on the training set reach the preset requirements or the training epochs are completed.

[0166] Images requiring longitudinal tear detection are preprocessed using the same methods as the training set before being input into the trained deep learning model. The model extracts and analyzes features from the input images, outputting a prediction of whether a tear is longitudinal and the severity of the tear (mild or severe).

[0167] Example 7:

[0168] Based on the solutions described in the foregoing embodiments, as a preferred solution, identifiers are placed on the non-load-bearing surface of the belt near each reflective strip. These identifiers can be numbers, letters, QR codes, or other easily identifiable symbols. When selecting identifiers, readability and durability should be considered. Identifiers should be made of wear-resistant and weather-resistant materials, such as abrasion-resistant coatings, durable stickers, or engraved markings. The identifiers should be large enough to be clearly visible in the image without affecting the normal use of the belt. The specific size can be adjusted according to the camera resolution and installation distance. The identifiers should be kept at a certain distance from the reflective strips to avoid glare interference. The spacing of the identifiers should match the spacing of the reflective strips to ensure that the system can identify all identifiers in each detection cycle.

[0169] Different areas are distinguished using numerical markers on the conveyor belt, with each marker corresponding to a specific region. Image processing algorithms (such as OCR technology) automatically identify these markers in the image, thereby determining the specific region corresponding to the current image. Based on the identified markers, the conveyor belt is divided into several independent detection areas. By setting markers within each region, the system can perform image comparison and anomaly detection more efficiently. This method not only improves the accuracy and speed of detection but also facilitates fault location and maintenance.

[0170] The above description is merely a specific embodiment of the present invention. Various examples and illustrations do not constitute a limitation on the substantive content of the present invention. Those skilled in the art can modify or transform the specific embodiments described above after reading the specification without departing from the essence and scope of the present invention.

Claims

1. A computer vision-based method for detecting longitudinal tears in belt conveyors, characterized in that, Includes the following steps: Step 1: Reflective strips are evenly spaced on the non-load-bearing surface of the conveyor belt, dividing the non-load-bearing surface into sections. Multiple regions; Step 2: Install a light source, CCD camera, photodiode, and roller brush on the conveyor frame; the field of view of the CCD camera... It can completely cover the non-load-bearing surface between two adjacent reflective strips to capture an area of ​​the non-load-bearing surface. The complete image shows that the light emitted by the light source is reflected by the reflective strip and can be received by the photodiode. The non-load-bearing surface... During the movement, it is first cleaned by the roller brush and then illuminated by the planar light spot projected by the light source; Step 3: Use a photodiode to detect the reflection from the reflective strip to trigger a CCD camera to capture images of the non-load-bearing conveyor belt. On the other hand, a CCD camera captures images of the non-load-bearing surface of the conveyor belt and transmits them to a field computer for processing. Step 4: Initial image acquisition and storage. During the first run of the system, images are acquired for each area on the non-load-bearing surface of the conveyor belt. The initial image of the domain serves as the benchmark for subsequent comparisons; Step 5: During the subsequent operation of the belt conveyor, image acquisition and real-time image comparison are performed. The on-site computer uses the image... The comparison algorithm compares the real-time acquired region image with the initial image to detect whether there are significant changes. If so, it will... If there is a significant change, proceed to step 6; otherwise, continue with step 5. Step 6: Perform anomaly identification on areas of the image showing significant changes. The on-site computer will identify areas with significant changes. The image is transmitted to the server, which identifies areas of significant change and determines whether they are longitudinal tears. If a longitudinal tear fault exists, proceed to step 7; otherwise, proceed to step 5. Step 7: Longitudinal Tear Alarm. Upon detecting a longitudinal tear fault in the conveyor belt, the server issues an alarm signal, and... Record the fault location and time so that maintenance personnel can handle it promptly.

2. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 1, characterized in that, Place Step 5 includes the following sub-steps: Sub-step 5.1: Before performing image comparison, preprocess the acquired images of the non-load-bearing surface of the conveyor belt. To improve the accuracy and efficiency of comparison; Sub-step 5.2: Compare the real-time acquired images with the stored initial images to detect any changes.

3. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 2, characterized in that, Place Step 5.2 of the descriptor includes the following operations: Image alignment: Calculate transformation parameters based on the reflection stripes in the image, and then align the image according to the calculated transformation parameters. Perform a transformation to align the two images being compared in spatial position; Image difference calculation: Calculate the difference between two aligned images. Each pixel value in the difference image reflects the corresponding... The degree of change in pixels between two images at different positions can directly show the differences between the two images; Thresholding: Thresholding is performed on the difference image to extract regions of significant change. This is done by setting a threshold value for the changes. In the difference image, pixel values ​​less than the change threshold are set to 0, and pixel values ​​greater than the change threshold are set to 1. Counting distinct pixels: This involves counting the number of pixels with a value of 1 in the thresholded image; these are considered distinct pixels. If the quantity exceeds a preset threshold for different pixels, the conveyor belt is judged to be in an abnormal state.

4. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 1, characterized in that, Place Step 6 includes the following operations: sub-step 6.1, which preprocesses the image of the region with obvious changes, and the preprocessed image... Sub-step 6.2 involves detecting longitudinal cracks in the processed images; Sub-step 6.1 includes the following operations: Grayscale conversion: Converting the image output by a CCD camera into a grayscale image; Noise Removal: Gaussian filtering and median filtering techniques are used to remove noise from the image, enhancing image clarity. Subsequent edge detection and recognition operations provide a clearer image; Edge detection: Edge detection algorithms are used to extract edge information from images, including noise suppression, gradient calculation, and non-polar edge detection. Large value suppression and dual threshold detection; Binarization: Converting an image into a binary image by thresholding.

5. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 4, characterized in that, Place Sub-step 6.2 includes the following operations: Grid partitioning: The binary image is divided into an N*N grid, with each grid cell being k*k pixels in size; Node initialization: Calculate the average gray value of each grid cell. If the average value is greater than a preset threshold T, mark it as having a value of [value missing]. Crack nodes with a value of 1 are marked as 1, otherwise they are marked as background nodes with a value of 0. Node connectivity calculation: The connectivity of nodes is calculated using connectivity rules. For each crack node, its connectivity is checked. Connectivity with adjacent nodes; if connected, mark it as a crack region. Crack morphology analysis: Calculate the length and width of the crack in the crack region; Crack detection: Calculate the ratio of crack length to crack width. When the ratio of crack length to width is greater than one... When a threshold U is met and the length of the crack is greater than the threshold V, it is determined that a longitudinal tear has occurred in the belt conveyor. The node connectivity calculation includes the following operations: Initialize a label matrix to record the crack region to which each node belongs; Iterate through all nodes, and for unlabeled crack nodes, assign a new label; Recursively or iteratively check its neighboring nodes. If a neighboring node is a crack node and is not labeled, assign it the same label.

6. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 5, characterized in that, Place The crack morphology analysis includes the following detailed steps: Scan the connected components from top to bottom and record the position of the first white pixel as (x1, y1). Scan the connected components from bottom to top, and record the position of the first white pixel as (x2, y2). Scan the connected components from left to right and record the position of the first white pixel as (x3, y3). Scan the connected components from right to left and record the position of the first white pixel as (x4, y4); Xleft =min{x1, x2, x3, x4}, Xright =max{x1, x2, x, x}, Yup = min{y1, y 2, y 3, y 4}, Ydown =max{y1, y2, y3, y4}; Crack length = Ydown - Yup, crack width = Xright - Xleft.

7. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 4, characterized in that, Place Predicate step 6.1 includes: The image is resized to a fixed size to facilitate input into the deep learning model for processing; Normalizing the image maps pixel values ​​to the [0, 1] interval, improving the training performance and acquisition efficiency of the model. Convergence speed; Image enhancement includes random rotation, flipping, cropping, and adding noise to increase data diversity and prevent... To prevent model overfitting.

8. The computer vision-based longitudinal tear detection method for belt conveyors according to claim 4, characterized in that, Place Predicate step 6.2 includes: A convolutional neural network (CNN) is chosen as the basic architecture to build a deep learning model, with its output layer having two output sections. The points are used to predict whether it is a longitudinal crack and the severity of the crack, respectively. The pre-collected conveyor belt images are divided into training and testing sets according to a certain ratio for model training. and assessment; The training set is input into the deep learning model, the model's output is calculated through forward propagation, and then the output is compared with the true value. The difference between real labels is used to calculate the loss function value, and then the gradient of the loss function with respect to the model parameters is calculated using the backpropagation algorithm. The model parameters are updated using an optimization algorithm, and this process is repeated until the model achieves the desired accuracy and loss value on the training set. Once the preset requirements are met or the training rounds are completed; The images requiring longitudinal tear detection are preprocessed using the same method as the training set, and then input into... The trained deep learning model outputs a prediction of whether a longitudinal tear exists and a prediction of the tear's severity. fruit.

9. A belt conveyor longitudinal tear detection device for use in the method of any one of claims 1-8, comprising: Reflective strips Belt, light source, CCD camera, photodiode, field computer, server, roller brush; A reflective strip is installed at regular intervals L on the non-load-bearing surface of the conveyor belt. The interval L is related to the conveyor belt's position. The width of the conveyor belts is the same; The light source uses high-brightness LEDs, which can project a planar light spot and illuminate the non-load-bearing surface of the conveyor belt. Above, the planar light spot can evenly illuminate the non-load-bearing surface between two adjacent reflective strips; The CCD camera uses a high-resolution industrial camera. Adjusting the mounting position of the CCD camera allows for a wider field of view. It can completely cover the non-load-bearing surface between two adjacent reflective strips; The light source and photodiode are mounted on the left and right sides of the CCD camera, respectively, so that the light emitted by the light source is reflected by the reflective strips. The reflected light can be received by the photodiode. The roller brush is positioned on one side of the light source, ensuring that the non-load-bearing surface of the conveyor belt is cleaned by the roller brush first during operation. Illuminated by a planar light spot projected by a light source to prevent dust from adhering to the non-load-bearing surface of the conveyor belt and affecting subsequent longitudinal... Interferes with tear detection; The field computer is electrically connected to the light source, CCD camera, and photodiode. The field computer can be configured via software. Control the light source to turn on and off, trigger the CCD camera to capture images, and receive the images output by the CCD camera; The on-site computer is connected to the server via a network.

10. The belt conveyor longitudinal tear detection device according to claim 9, characterized in that, Light emitted by the light source Diffuse reflection occurs on the non-load-bearing surface of the conveyor belt, while specular reflection occurs on the reflective strip. The reflective strip then passes through a photoelectric sensor. When the photodiode is near a diode, the electrical signal output by the photodiode increases. The field computer then processes the received electrical signal from the photodiode output. A trigger signal can be obtained by detecting the rising edge of the signal, and the CCD camera can be controlled to take pictures based on the trigger signal.

Images