Belt tearing detection method and belt tearing detection system of belt conveyor

Abstract

The invention provides a belt tearing detection method and a belt tearing detection system for a belt conveyor. The method comprises the steps that a line laser image is obtained, and the line laser image is a line laser image on a belt of the belt conveyor at the current moment; a laser line segment in the line laser image is extracted, and the laser line segment is a line segment formed by the line laser on the belt; when the number of the laser line segments is larger than or equal to a preset number threshold value, it is determined that the belt is torn, and when the number of the laser line segments is smaller than the preset number threshold value, it is determined that the belt is not torn; and under the condition of determining that the belt is torn, controlling the alarm equipment to give an alarm. According to the method, the problem of low accuracy of longitudinal tear detection of the mining belt conveyor in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of non-destructive testing technology for conveyor belts, and more specifically, to a method for detecting belt tearing in a belt conveyor, a device for detecting belt tearing in a belt conveyor, a computer program product, and a belt tearing detection system. Background Technology

[0002] Belt conveyors are the primary coal conveying equipment in coal mines, operating under high-intensity working conditions for extended periods. The belt is a key component of the belt conveyor, mainly responsible for transporting coal.

[0003] During coal transportation, hard foreign objects such as flaky rocks, waste stones, and metal anchors are often mixed in. These sharp and hard impurities pose a potential threat to the conveyor belt and may cause serious axial damage and longitudinal tearing.

[0004] Once a longitudinal tear occurs, it will continue to expand along the direction of the conveyor belt, leading to a series of problems such as material spillage and equipment damage, severely reducing production efficiency, causing significant economic losses, and threatening the safety of operators.

[0005] However, the accuracy of longitudinal tear detection for mining belt conveyors is currently low, making it impossible to effectively detect longitudinal tears in mining belt conveyors. Summary of the Invention

[0006] The main objective of this application is to provide a method for detecting belt tearing in a belt conveyor, a device for detecting belt tearing in a belt conveyor, a computer program product, and a belt tearing detection system, so as to at least solve the problem of low accuracy in longitudinal tear detection of mining belt conveyors in the prior art.

[0007] To achieve the above objectives, according to one aspect of this application, a method for detecting belt tearing in a belt conveyor is provided, comprising: acquiring a line laser image, wherein the line laser image is a line laser image on the belt of the belt conveyor at a current moment; extracting laser line segments from the line laser image, wherein the laser line segments are line segments formed by the line laser on the belt; determining that the belt is torn if the number of laser line segments is greater than or equal to a preset number threshold, and determining that the belt is not torn if the number of laser line segments is less than the preset number threshold; and controlling an alarm device to sound an alarm if the belt is determined to be torn.

[0008] Optionally, extracting laser line segments from the line laser image includes: preprocessing the line laser image to obtain a target line laser image, wherein the preprocessing method includes cropping and / or grayscale processing; and extracting the laser line segments from the target line laser image.

[0009] Optionally, preprocessing the line laser image to obtain a target line laser image includes one of the following: cropping or performing grayscale processing on the line laser image to obtain the target line laser image; cropping the line laser image and performing grayscale processing on the cropped line laser image to obtain the target line laser image; or performing grayscale processing on the line laser image and cropping the grayscale processed line laser image to obtain the target line laser image.

[0010] Optionally, cropping the line laser image to obtain the target line laser image includes: extracting a portion of the line laser image to obtain a Region of Interest (ROI), wherein the ROI contains at least the complete area of ​​the line laser on the belt, and the area of ​​the ROI is smaller than the area of ​​the line laser image; cutting off the area in the line laser image other than the ROI, retaining the ROI in the line laser image, to obtain the target line laser image.

[0011] Optionally, performing grayscale processing on the line laser image to obtain the target line laser image includes: extracting color channel features from the line laser image to obtain R channel features, G channel features, and B channel features; performing grayscale processing on at least one of the color channel features to obtain an intermediate line laser image; when performing grayscale processing on one or two of the color channel features, setting the color channel features of the intermediate line laser image that have not undergone grayscale processing to zero to obtain the target line laser image; when performing grayscale processing on three of the color channel features, determining the intermediate line laser image as the target line laser image.

[0012] Optionally, extracting the laser line segment from the target line laser image includes: obtaining a training image recognition model, wherein the training image recognition model is one of GMM model, HMRF model, U-Net model, and isolated forest model; forming a training image set from the pixels in the target line laser image, and training the training image set and corresponding category labels to obtain a category recognition model, wherein the category label is the category of the pixel in the training image set, and the category is laser, belt, or background; sequentially inputting all the pixels in the target line laser image into the category recognition model to obtain the category recognition result corresponding to each pixel; determining two adjacent pixels of the same category as laser as adjacent laser pixel pairs; connecting the two pixels in each adjacent laser pixel pair to form at least one initial laser line segment; and performing image optimization on at least one initial laser line segment to obtain at least one laser line segment, wherein the image optimization method includes one or more of image gradient optimization, image dilation optimization, and image erosion optimization.

[0013] Optionally, after determining that the belt is torn, the method further includes: extracting torn pixels from the line laser image, wherein the torn pixels are pixels between two laser line segments, and the laser line segments and the torn pixels are located on the same straight line; counting the number of torn pixels; obtaining a mapping relationship, wherein the mapping relationship is the relationship between the number of torn pixels and the tear length; determining the tear length corresponding to the number of torn pixels as the target tear length according to the mapping relationship; and controlling the belt conveyor to stop if the target tear length is greater than or equal to a preset length threshold.

[0014] According to another aspect of this application, a belt tear detection device for a belt conveyor is provided, comprising: a first acquisition unit for acquiring a line laser image, wherein the line laser image is a line laser image on the belt of the belt conveyor at a current moment; a first extraction unit for extracting laser line segments from the line laser image, wherein the laser line segments are line segments formed by the line laser on the belt; a first determination unit for determining that the belt is torn when the number of laser line segments is greater than or equal to a preset number threshold, and determining that the belt is not torn when the number of laser line segments is less than the preset number threshold; and a first control unit for controlling an alarm device to sound an alarm when the belt is determined to be torn.

[0015] According to another aspect of this application, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of any of the belt tear detection methods for the belt conveyor.

[0016] According to another aspect of this application, a belt tear detection system is provided, comprising: a belt conveyor including a belt; a linear laser for emitting a linear laser beam onto the belt of the belt conveyor; a camera for acquiring an image of the linear laser beam reflected by the belt; an alarm device for issuing an alarm; and a controller communicatively connected to the belt conveyor, the linear laser, the camera, and the alarm device, wherein the controller is used to execute any of the belt tear detection methods of the belt conveyor described above.

[0017] Using the technical solution of this application, a linear laser emitter can generate a continuous linear laser beam. When this linear laser beam irradiates a flat surface, ideally, it will form a straight line with uniform brightness on the surface. However, if the belt is torn, it will form an opening or crack, and the laser beam will break at these torn points, failing to form a continuous laser line segment on the belt surface. This will appear as an interruption in the laser line segment on the image. Therefore, by extracting the number of laser line segments in the belt image, it can be determined whether the belt is torn. For example, when the belt is torn, two laser line segments will be formed on the belt, while when the belt is not torn, one complete laser line segment will be formed on the belt. Compared with traditional pressure detection and temperature detection methods, this solution can perform detection without contacting the belt, reducing interference from equipment operation and making the tear detection accuracy of this solution higher. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0019] Figure 1 A hardware structure block diagram of a mobile terminal for performing a belt tear detection method for a belt conveyor according to an embodiment of this application is shown.

[0020] Figure 2 A schematic flowchart of a method for detecting belt tearing in a conveyor belt according to an embodiment of this application is shown.

[0021] Figure 3 A schematic diagram of image grayscale processing is shown;

[0022] Figure 4 A schematic diagram of image enhancement is shown;

[0023] Figure 5 A schematic diagram of laser line segment extraction is shown;

[0024] Figure 6 A schematic diagram showing the number of laser line segments is provided.

[0025] Figure 7 A flowchart illustrating another method for detecting belt tearing in a belt conveyor is shown.

[0026] Figure 8 A structural block diagram of a belt tear detection device for a belt conveyor provided according to an embodiment of this application is shown;

[0027] Figure 9 A schematic diagram of the belt tear detection system is shown.

[0028] The above figures include the following reference numerals:

[0029] 102. Processor; 104. Memory; 106. Transmission equipment; 108. Input / output equipment; 10. Belt conveyor; 11. Belt; 12. Linear laser; 13. Camera; 14. Alarm equipment; 15. Mounting frame. Detailed Implementation

[0030] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0031] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] As described in the background section, the accuracy of longitudinal tear detection for mining belt conveyors in the prior art is relatively low. To solve the above problems, embodiments of this application provide a method for detecting belt tearing of a belt conveyor, a device for detecting belt tearing of a belt conveyor, a computer program product, and a belt tear detection system.

[0034] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0035] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a method of detecting belt tearing in a conveyor belt according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0036] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the device information display method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0037] This embodiment provides a method for detecting belt tearing in a belt conveyor that runs on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0038] Figure 2 This is a schematic flowchart illustrating a method for detecting belt tearing in a conveyor belt according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0039] Step S201: Obtain a line laser image, wherein the line laser image is a line laser image on the belt of the belt conveyor at the current moment;

[0040] Specifically, during the operation of the belt conveyor, a linear laser is placed directly below the belt, emitting a linear laser beam onto the belt surface. This laser works closely with a camera, which is calibrated to capture real-time images of the belt under laser illumination, resulting in a linear laser image.

[0041] Step S202: Extract the laser line segment from the above-mentioned line laser image, wherein the laser line segment is the line segment formed by the above-mentioned line laser on the above-mentioned belt;

[0042] Specifically, laser lines are continuous or broken bright lines that appear in an image after the belt surface is irradiated by a laser. Therefore, laser lines can be extracted from the acquired laser image.

[0043] Step S203: If the number of laser segments is greater than or equal to a preset number threshold, the belt is determined to be torn; if the number of laser segments is less than the preset number threshold, the belt is determined not to be torn.

[0044] Specifically, a normally functioning conveyor belt forms a continuous and stable laser line under laser irradiation, theoretically appearing as a single, complete line. However, once a tear occurs, this continuity is broken, resulting in multiple independent short segments. By setting a threshold of 2 or another value, the tear condition can be quickly determined through a simple and intuitive quantitative comparison.

[0045] Step S204: If the belt is confirmed to be torn, the alarm device is activated.

[0046] Specifically, when a belt tear is detected, an alarm mechanism is automatically triggered. The alarm device can be an audible and visual alarm, capable of immediately sending alarm information to operators and maintenance teams, indicating that the belt has torn.

[0047] In this embodiment, a linear laser emitter can generate a continuous linear laser beam. When this linear laser beam illuminates a flat surface, ideally, it will form a straight line with uniform brightness on the surface. However, if the belt is torn, it will form an opening or crack, and the laser beam will break at these torn points, failing to form a continuous laser line segment on the belt surface. This will appear as an interruption in the laser line segment on the image. Therefore, by extracting the number of laser line segments in the belt image, it is possible to determine whether the belt is torn. For example, when the belt is torn, two laser line segments will be formed on the belt, while when the belt is not torn, one complete laser line segment will be formed on the belt. Compared with traditional pressure detection and temperature detection methods, this solution can perform detection without contacting the belt, reducing interference from equipment operation and making the tear detection accuracy of this solution higher.

[0048] With technological advancements and innovation, traditional detection methods that involve direct contact with objects, such as electromagnetic induction, pressure sensing, and ultrasonic scanning, are gradually giving way to more advanced solutions due to their sensitivity to external interference. Among numerous emerging technologies, machine vision is gaining prominence due to its unique advantages, particularly in the crucial field of longitudinal tear detection in conveyor belts. Researchers are exploring how to utilize machine vision to accurately identify tears by analyzing visible light images. However, the application of this technology to coal mine belt conveyors faces challenges, as these devices often operate in extreme environments characterized by dust, low light, and large temperature fluctuations. While infrared image detection technology can mitigate the impact of harsh environments on detection reliability, its ability to distinguish subtle differences in conveyor belt friction heat is insufficient, especially in identifying longitudinal tears, where it is difficult to accurately locate the problem area in infrared images. The accuracy and noise suppression capabilities of existing algorithms still need improvement. Overall, these detection methods each have their shortcomings, including insufficient reliability, high operational complexity, low efficiency, and unsatisfactory accuracy. Whether these methods will inevitably encounter problems in practical applications and fail to achieve the expected detection results remains to be seen.

[0049] The solution proposed in this application effectively reduces the interference of environmental and lighting factors by introducing line laser-assisted detection technology, resulting in a higher accuracy rate for tear detection.

[0050] Specifically, the aforementioned hardware equipment includes a line laser emitter and a CCD camera. The line laser emitter can be a single-line laser, and the CCD camera captures the laser stripe image on the lower surface of the upper belt to obtain a line laser image.

[0051] Specifically, a line laser emitter is used instead of a point laser source. A line laser emitter can generate a continuous line of laser light, rather than a single spot. When this line of laser light shines on a flat surface, ideally, it will form a straight line of uniform brightness on the surface. However, if the surface has cracks, depressions, or longitudinal tears, the reflection characteristics of the laser line at these defects will change, causing the laser line to deform or break in these areas, thus forming unique optical features. Longitudinal tears create breaks, and when the laser line encounters such a break, part of the laser may not be able to find a reflective surface at the tear, thus failing to continue forming a continuous laser line, which appears as an abrupt interruption of the laser line in the image.

[0052] In the specific implementation process, the laser line segments in the above-mentioned line laser image can be extracted through the following steps: preprocessing the above-mentioned line laser image to obtain the target line laser image, wherein the above-mentioned preprocessing methods include cropping and / or grayscale processing; extracting the above-mentioned laser line segments in the above-mentioned target line laser image.

[0053] In this scheme, after the cropping operation, the image size is reduced, which can reduce the computational complexity. Grayscale conversion further highlights the brightness information of the laser lines, making the laser line area in the image more prominent, which allows for more accurate extraction of the laser line segments.

[0054] Specifically, in the internal environment of a mine, due to factors such as poor lighting conditions, high dust concentration, and temperature fluctuations, traditional visual inspection technologies struggle to accurately identify laser line features on a conveyor belt. This embodiment employs pre-defined cropping and grayscale processing, cropping the acquired image size from 1024x1024 pixels to 256x1024 pixels (focusing on the laser line area) and converting the image to grayscale mode. The resulting target laser line image not only displays the laser lines clearly but also boasts fast processing speed, making it highly suitable for use in complex environments such as mines.

[0055] In some embodiments, preprocessing the line laser image to obtain a target line laser image includes one of the following: cropping or performing grayscale processing on the line laser image to obtain the target line laser image; cropping the line laser image and performing grayscale processing on the cropped line laser image to obtain the target line laser image; or performing grayscale processing on the line laser image and cropping the grayscale processed line laser image to obtain the target line laser image.

[0056] In this scheme, cropping can remove image parts that are not related to laser detection, reducing noise interference. Grayscale processing further enhances the brightness characteristics of the laser lines, making the lines stand out more in the image and easier to identify laser line segments.

[0057] Specifically, the above solution includes the following three implementation methods:

[0058] The first method involves directly cropping the ROI of the original line laser image, focusing on the laser line area, removing irrelevant background, and obtaining the cropped target line laser image.

[0059] The second method involves first cropping the Region of Interest (ROI) to obtain an image of the laser line area. Then, the cropped image is converted to grayscale to obtain a grayscale image of the target laser line.

[0060] The third method involves first converting the original RGB line laser image into a grayscale image to improve the contrast of the laser lines. Then, the grayscale image is cropped by region of interest (ROI) to ensure that only the laser lines are retained, resulting in the final target line laser image.

[0061] If the image contains a large amount of background noise and non-laser line areas, more data points may need to be processed, which not only increases the computational load but may also introduce additional recognition errors. By cropping, the processed image contains only the laser lines, enabling faster and more accurate location of the laser lines. Grayscale processing further enhances the contrast of the laser lines, helping the algorithm to more accurately identify line boundaries, thereby improving the accuracy of laser line segment extraction.

[0062] In the specific implementation process, the above-mentioned cropping of the line laser image to obtain the target line laser image can be achieved through the following steps: extracting a portion of the above-mentioned line laser image to obtain a Region of Interest (ROI), wherein the ROI region at least contains the complete area of ​​the above-mentioned line laser on the belt, and the area of ​​the ROI region is smaller than the area of ​​the above-mentioned line laser image; cutting off the areas of the above-mentioned line laser image other than the ROI region, retaining the ROI region in the above-mentioned line laser image, to obtain the target line laser image.

[0063] In this scheme, the laser lines occupy a much smaller proportion of the laser image than the rest of the image, with most of the area being background. By cropping the Region of Interest (ROI), only the area containing the laser lines is retained, significantly reducing the image size. The ROI cropping operation isolates the laser line region from the entire image, narrowing the scope of image processing, reducing unnecessary computations, and accelerating subsequent image analysis.

[0064] Specifically, the Region of Interest (ROI) is defined as the area of ​​interest. The ROI is located and extracted from the acquired linear laser images. First, image analysis algorithms are used to identify the laser lines on the belt and determine their approximate location in the image. Then, based on the location information of the laser lines, a minimal ROI containing the laser lines is defined. The boundary of this region should closely surround the laser lines to ensure that the laser lines are completely included within the ROI, while excluding non-laser line areas as much as possible, thereby reducing the proportion of the ROI relative to the original image.

[0065] Next, the original line laser image is cropped, retaining only the defined ROI region and removing the rest. This process ensures the processed image (target line laser image) is completely focused on the laser line, reducing computational burden in subsequent processing and effectively avoiding interference from background noise or other irrelevant information.

[0066] In the detection scenario of mining belt conveyors, the complex environment may contain a lot of dust, reflections, and shadows in the image, which can greatly interfere with the recognition of laser lines. However, by cropping the Region of Interest (ROI), the focus can be directly on the laser lines, minimizing the influence of these external factors.

[0067] In a specific application, during belt conveyor inspection in a mine, the original image size is relatively large, for example, 1024x1024 pixels, due to the belt width and laser illumination range. By cropping the Region of Interest (ROI), only the area containing the laser lines is retained; assuming the ROI size is adjusted to 256x1024 pixels, this method not only reduces the computational load on subsequent laser line segment recognition algorithms but also eliminates significant background noise interference, improving the accuracy and efficiency of the recognition.

[0068] In another specific application, under certain special conditions, such as in mines with smooth belt surfaces or strong ambient light, the brightness and contrast of the laser lines are not prominent. By precisely trimming the Region of Interest (ROI), the brightness and contrast of the laser lines are maximized within the ROI, further enhancing the recognition effect of the laser line segments. Even in complex lighting environments, the laser lines remain clearly distinguishable, improving the accuracy of tear detection.

[0069] Specifically, the video of the belt conveyor running can be read by a device that acquires line laser images of the conveyor belt. The captured line laser images can be cropped by ROI to eliminate unnecessary elements in the images and retain only the images containing the line laser feature areas.

[0070] Specifically, the red-striped laser region is extracted from the video image using ROI extraction technology. In line laser images, the core feature of the laser stripes is that their grayscale values ​​are significantly higher than other areas of the image, and they are typically distributed parallel to the image's u-axis. Based on the search results, the ROI can be located by analyzing the image's grayscale distribution: calculating the sum of the grayscale values ​​of pixels in each row, since the laser stripe region has a higher grayscale value, the sum of the grayscale values ​​in its row will be significantly higher than in other rows. This feature determines the row range where the laser stripes are concentrated; combined with the horizontal distribution range of the laser stripes, the column range is further determined. The finally delineated rectangular or irregular area is the region of interest containing the line laser feature.

[0071] After determining the ROI range, cropping can be performed using image processing tools (such as OpenCV). Define the region using `Rect`: by specifying the top-left corner coordinates (x, y), width, and height of the ROI, the region can be directly extracted from the original image. For example, in OpenCV, this can be achieved using `Mat imageROI = image(Rect(x, y, width, height))`.

[0072] The process of cropping the Region of Interest (ROI) essentially involves preserving the target area while discarding other areas, thereby eliminating irrelevant elements such as background and noise in the image. Since subsequent processing is performed only on the ROI, it reduces computational load and avoids interference from irrelevant elements on analysis accuracy. In summary, ROI cropping of line laser images requires first locating the laser stripe region based on grayscale features, then extracting this region using tools, and finally eliminating unnecessary elements.

[0073] Specifically, the original line laser image needs to be preprocessed, including but not limited to noise reduction and contrast enhancement, to ensure that the features of the laser lines are clear enough for subsequent processing.

[0074] Thresholding segmentation is applied, setting a threshold based on the brightness or color characteristics of the laser line. Laser lines typically appear as brighter or more vividly colored lines than the background in an image, thus distinguishing them from other parts. The segmented binary image will display the outline of the laser line, while the rest will be set to the background (black or white).

[0075] Using contour detection algorithms in computer vision, such as the findContours function in the OpenCV library, the contour boundaries of the laser lines obtained by threshold segmentation are detected.

[0076] For each detected laser line profile, its minimum bounding rectangle or polygon is calculated. This helps determine the boundary of the laser feature region. The coordinates of the minimum bounding rectangle or polygon can be used for subsequent image cropping.

[0077] Use image processing software to crop out the area containing the laser lines. The cropped image should contain the complete laser line features while removing as much background as possible. If part of the laser line is found to be accidentally cut off or there is too much background, it can be corrected by adjusting the threshold or contour detection parameters.

[0078] In some embodiments, the grayscale processing of the aforementioned line laser image to obtain the target line laser image can be achieved through the following steps: extracting color channel features from the aforementioned line laser image to obtain R channel features, G channel features, and B channel features; performing grayscale processing on at least one of the aforementioned color channel features to obtain an intermediate line laser image; when performing grayscale processing on one or two of the aforementioned color channel features, setting the color channel features of the intermediate line laser image that have not undergone grayscale processing to zero to obtain the target line laser image; when performing grayscale processing on three of the aforementioned color channel features, determining the intermediate line laser image as the target line laser image.

[0079] In this scheme, since laser lines may have higher brightness and contrast in certain color channels, grayscale processing of these channels can specifically enhance the features of the laser lines. For example, if the laser light source is biased towards red, prioritizing grayscale processing of the R channel can significantly improve the recognition effect of the laser lines. The zeroing operation removes unnecessary color information, allowing the image to focus on the grayscale-processed features, thereby reducing noise during the recognition process and improving recognition accuracy.

[0080] Specifically, the image containing the extracted line laser feature region can be processed by extracting the R channel based on the R, G, and B color channel features of the image and then converting it to grayscale. The grayscale processed image is as follows: Figure 3 As shown. When extracting line laser features from an image, the R channel is selected, which is directly related to the color characteristics of the line laser. Since the line laser uses red, its grayscale value in the R channel will be significantly higher than that in the G and B channels. The G and B channels can be deleted or set to zero.

[0081] Of course, it is not limited to the above situations; other color channel features can also be grayscaled.

[0082] Specifically, the obtained grayscale image can be processed through image enhancement processing, and the enhanced image is as follows: Figure 4 As shown, grayscale adjustment is performed using an image grayscale conversion matrix to obtain a grayscale adjusted image. The first grayscale adjustment is used to expand the overall grayscale dynamic range and solve the problem of insufficient image contrast; the second adjustment is used to refine the processing of specific grayscale ranges to further highlight the details of the region of interest.

[0083] For example, the extracted line laser region image is converted to grayscale by extracting the R channel based on the R, G, and B channel features of the image.

[0084] The expression for grayscale transformation is as follows: g' = g·k + c, where,

[0085]

[0086] g is the gray value of the pixel before grayscale transformation, g' is the gray value after adjustment, and GMax and GMin are the maximum and minimum values ​​in the grayscale matrix, respectively.

[0087] A "grayscale matrix" refers to a two-dimensional array, or matrix, formed by organizing the grayscale values ​​of each pixel in a color image after it has been converted to a grayscale image. The rows and columns of this matrix correspond to the height and width of the image, respectively, and each element represents the grayscale value of the pixel at that location.

[0088] A grayscale image is different from an RGB image. It has only one brightness level, usually ranging from 0 (black) to 255 (white). Therefore, a grayscale matrix is used to represent the brightness information of each pixel in a grayscale image. When performing image analysis and processing, the grayscale matrix is a very important data structure because it simplifies the processing complexity of the original RGB image while retaining the basic structure and feature information of the image, facilitating subsequent image enhancement, feature extraction, image segmentation, etc.

[0089] In specific image processing tasks, the grayscale matrix will be used for further transformations and operations, such as grayscale level adjustment, morphological operations (such as dilation, erosion), detection of feature points, etc. By adjusting the element values in the grayscale matrix, specific features in the image can be enhanced. For example, in laser-assisted tear detection, the grayscale matrix can help highlight the line laser stripe, thus more accurately identifying and locating the tear position on the belt.

[0090] In operations such as image brightness adjustment or signal amplitude adjustment, g can represent the original pixel value or signal amplitude, k is used to control the adjustment ratio (for example, k > 1 means amplification, 0 < k < 1 means reduction), and c can be used for overall brightness or amplitude offset.

[0091] Specifically, if the line laser emitter emits red laser, then the signal of the red channel will be the strongest, while the responses of the green (G) and blue (B) channels will be relatively weak. Therefore, analyzing only the red channel can more effectively extract the features of the laser line and improve the signal-to-noise ratio.

[0092] In the specific implementation process, the above-mentioned laser line segments in the above-mentioned target line laser image can be extracted through the following steps: Obtain a to-be-trained image recognition model, where the above-mentioned to-be-trained image recognition model is one of the GMM model, HMRF model, U-Net model, and isolation forest model; form a training image set with the pixel points in the above-mentioned target line laser image, and use the above-mentioned training image set and the corresponding class labels to train the above-mentioned to-be-trained image recognition model to obtain a class recognition model, where the above-mentioned class label is the class of the above-mentioned pixel points in the above-mentioned training image set, and the above-mentioned class is laser, belt, or background; input all the above-mentioned pixel points in the above-mentioned target line laser image into the above-mentioned class recognition model in sequence to obtain the class recognition result corresponding to each of the above-mentioned pixel points; determine adjacent two above-mentioned pixel points with the class of the above-mentioned laser as adjacent laser pixel point pairs; connect the two above-mentioned pixel points in each of the above-mentioned adjacent laser pixel point pairs to form at least one initial laser line segment; perform image optimization on at least one of the above-mentioned initial laser line segments to obtain at least one of the above-mentioned laser line segments, where the above-mentioned image optimization method includes one or more of image gradient optimization, image dilation optimization, and image erosion optimization.

[0093] In this scheme, the characteristics of laser lines can be learned from a large number of labeled laser images, and then the category of pixels can be accurately identified. By accurately classifying laser lines and background, the recognition error caused by changes in light or background interference can be reduced. Image gradient optimization can identify and repair tiny breaks in laser lines through gradient calculation, while dilation optimization and erosion optimization remove small noise and fill small gaps on the lines by expanding the lines and shrinking the line edges, respectively. This ensures that the accuracy of the obtained laser line segments is high.

[0094] Specifically, selecting an appropriate image recognition model is crucial for pre-training preparation. For example, the U-Net model can be chosen as the image recognition model to be trained. U-Net is a convolutional neural network model widely used in image segmentation, favored for its compact structure and good segmentation performance.

[0095] A training image set and corresponding category labels are constructed to train the U-Net model described above. The training image set consists of pixels in the target laser line image, with each pixel accompanied by a category label indicating whether the pixel belongs to the laser, belt, or background. Through a large amount of labeled image data, a category recognition model capable of accurately distinguishing between laser lines and the background is trained.

[0096] Using a trained category recognition model, each pixel in the target line laser image is classified to obtain the category recognition result. This process labels each pixel with its most likely category, laying the foundation for subsequent laser line segment recognition. From the classification results, pixels labeled as lasers are identified, and adjacent laser pixel pairs are determined. These adjacent laser pixel pairs are the constituent units of the laser line segment.

[0097] Adjacent laser pixels are connected to form at least one initial laser line segment. This is accomplished using the positional information of the pixels, ensuring the continuity and integrity of the laser line.

[0098] Image optimization is performed on all initial laser lines to eliminate noise and breaks, resulting in clearer and more accurate laser lines. Image optimization can be achieved through image gradient optimization, which repairs interrupted lines by calculating the gradient direction and intensity of the laser lines; or through image dilation and erosion optimization, where the former fills in small gaps in the laser lines, and the latter removes protrusions or noise from the lines, ensuring smoothness and continuity of the laser lines.

[0099] In a specific application, the detection of laser lines on conveyor belts in a mining environment presents significant challenges. The complex lighting conditions inside the mine, the irregular features on the belt surface such as wear, cracks, and stains, and the potential impact of dust and humidity on laser emission all contribute to the difficulty of identifying laser lines. In this case, the U-Net model is used for initial classification of the laser lines, accurately distinguishing them from the belt and background, providing clear guidance for subsequent laser line segment extraction. Subsequent image optimization operations, such as image gradient optimization, automatically repair line interruptions caused by belt wear or unstable laser emission; image dilation optimization fills in minor gaps in the lines, ensuring their integrity and continuity; and image erosion optimization removes protrusions or noise, making the lines smoother and improving the accuracy of tear detection.

[0100] For example, setting the image gradient optimization threshold to 0.5 means that if the gradient value of a detected pixel is lower than 0.5, it will be considered background noise or line breaks. Through filling or removal operations, clearer and more continuous laser lines can be obtained. In this way, even in the dim light inside the mine and where laser emission is not ideal, high-precision detection of belt tears can be achieved, ensuring the safe operation of the belt conveyor.

[0101] Specifically, the laser centerline can be extracted from the grayscale-transformed image using an improved grayscale centroid method (IGGM). The laser centerline is converted to a world coordinate system via spatial mapping, generating a corresponding image of the laser centerline in that world coordinate system. Linear elements in the image are identified, and interconnected regions are extracted, recording the total number N of these connected regions. The presence of longitudinal tearing in the image is assessed based on the number of connected regions N.

[0102] Specifically, the Gaussian distribution of the grayscale distribution curve is fitted using pixel grayscale value samples, and the criterion function is given as follows:

[0103]

[0104] Where A is the magnitude of the grayscale pixel value, i.e., the position with the largest grayscale value in the image, μ is the mean of the grayscale values ​​in the image, σ is the standard deviation of the Gaussian distribution parameter, and x is the mean of the grayscale values ​​in the image. i y i Let J be the coordinates of the sample, J be the objective function, N be the number of samples, and x be the coordinates of the sample. i Let y represent the grayscale value sample of the i-th pixel. iThis formula represents a certain observed or target value associated with a pixel, where N represents the total number of samples (i.e., the number of pixels involved in the calculation), A represents the parameters of the Gaussian distribution, which are related to the amplitude of the distribution, and μ represents the parameters of the Gaussian distribution, which is the mean. This formula measures the features of an image by calculating how well a sample fits a hypothetical normal distribution.

[0105] The above formula is constructed based on the logarithmic form of the probability density function of the Gaussian distribution. Taking the logarithm of the probability density function of the Gaussian distribution yields the above formula. The formula sums over all sample points and then squares the absolute value of the inner expression. This form is similar to the objective function of the least squares method. By minimizing this function, the parameters of the Gaussian distribution are optimized, making the fitted Gaussian distribution as close as possible to the actual gray-value sample distribution. The expression on the right is equivalent to the expression on the left, indicating that some algebraic transformations have been performed on the inner expression, facilitating the use of different optimization methods or parameter estimation strategies in actual calculations.

[0106] Specifically, sample data is extracted from the grayscale values ​​of pixels in the laser line region, and then the grayscale distribution of these sample data is modeled by fitting a Gaussian distribution. This is to better understand the brightness characteristics of the laser lines and their distribution patterns in the image. The next step involves optimizing the grayscale distribution, which involves adjusting the grayscale values ​​to improve the contrast of the laser lines.

[0107] By setting a reasonable threshold and employing lateral gradient optimization, the criterion function also helps reduce the impact of noise on the grayscale centroid position. Due to their randomness, noise points often do not conform to the grayscale distribution pattern of the laser line. Optimizing the criterion function can filter out these noise points that do not conform to the Gaussian distribution model, making the centerline of the laser line more accurate.

[0108] A lateral gradient optimization calculation is introduced, and the position of the gray-level centroid is further optimized based on the lateral gradient trend of the laser line. This means that after initially locating the gray-level centroid, fine adjustments can be made based on the gray-level change trend of the line in the lateral direction, further improving the positioning accuracy of the center line.

[0109] At this time, for y i Noise near 0, ln y i This will amplify the noise, so we discard sample data close to 0 and only retain sample points in the peak region, converting it to a univariate quadratic minimum fit. Assume the fitting result is: Where p1, p2, and p3 are the correlation constant coefficients of the fitted quadratic equation, and the corresponding parameter estimates are:

[0110] D participates in the calculation as an intermediate expression, helping to simplify or organize the entire calculation process;

[0111]

[0112] Solve the above system of equations and obtain the values ​​of p1, p2, and p3 through matrix operations and other methods.

[0113] Taking a confidence interval [μ-t, μ+t] at a certain confidence level as the threshold T = G(μ-t), the original gray-level matrix A1 of the image is optimized into a new gray-level matrix A2:

[0114]

[0115] Where g'(i,j) is the optimized gray value, g(i,j) is the original gray value, and G is a function related to the threshold setting. It plays a role in adjusting the original gray value g(i,j) based on the threshold T during the entire gray-scale optimization process. In some simple threshold processing scenarios, G is a piecewise function similar to the above, or a more complex function related to the local features of the image.

[0116] Since G is a function related to threshold setting, its range depends on its specific function form. If it's a simple piecewise function like the example above, then the range of G can be {0, 1}. However, if it's a more complex function, such as one involving a linear or non-linear transformation of the original grayscale value g(i, j), then its range can be determined based on the specific transformation rules and the grayscale range of the image. Generally, if the grayscale range of the original image is [0, L-1] (where L is the number of grayscale levels, e.g., L = 256 for an 8-bit grayscale image), then the range of G can be within [0, L-1] or within the normalized interval [0, 1], depending on its function definition.

[0117] `t` is a parameter used to determine a confidence interval. In image processing, confidence intervals are typically related to assessing the reliability of the processing results or determining a range based on certain statistical characteristics. For example, in statistically based threshold selection (such as the Otsu algorithm) or in judging the attributes of a region in some region-based processing, `t` is used to define a reasonable range of values ​​or as a key parameter in the calculation of certain statistics. Specifically, in algorithms that optimize grayscale values ​​based on the statistical characteristics of local regions, `t` is related to a combination or multiple relationship of statistical measures such as the standard deviation and mean of the pixel grayscale values ​​within that local region, thereby determining a suitable confidence interval to decide which pixel grayscale values ​​need to be adjusted and how.

[0118] The range of values ​​for 't', used to determine the confidence interval, is typically related to the specific application scenario and the algorithm employed. If it's based on statistical characteristics, such as using the mean and standard deviation to determine the confidence interval, 't' might take the form of k / σ (where k is a constant, such as k = 1, 2, 3, etc.), and its range could be within [0, +∞). However, in practical applications, a suitable finite interval, such as [0, 3σ], is selected based on the statistical characteristics of the image and processing requirements. In other non-statistical scenarios, such as determining the confidence interval based on experience or fixed rules, the range of values ​​for 't' must be determined based on the range of variables involved in that rule.

[0119] When a confidence interval is chosen at a certain confidence level, it does not mean that data outside the interval is absolutely outlier. In this image processing context, it is merely a classification standard. Elements in the original grayscale matrix are classified based on a comparison with the threshold. Elements greater than or equal to the threshold retain their original grayscale value g(i,j), while those less than the threshold are set to 000. This process is used to optimize and obtain a new grayscale matrix.

[0120] The tangent vector of the grayscale centroid is calculated, and a new center point is determined using lateral gradient optimization. Then, the new center points are connected to form the center line. Lateral gradient optimization rules:

[0121] A′(x p1 +Δx)+B′(y p1 +Δy)+C′=0;

[0122] This equation is used to determine the gray-level centroid C of an image. p The tangent line, whose gray-level centroid C is obtained by calculating its adjacent columns. p+1 It also follows a consistent displacement to move to the new grayscale centroid C' p+2 :

[0123]

[0124] x p1 and y p1 The image's grayscale centroid C represents... p The coordinates are the positions of the gray-scale centroid currently under consideration. C' represents a constant term. In the calculation of tangent vector correlation, Δx and Δy usually represent small increments or displacements in the x and y directions. Here, they are used to determine the tangent vector and other correlation quantities by calculating adjacent positions.

[0125] The main purpose of displacement is to determine new center points through certain optimization rules. These new center points, when connected, can form center lines that better conform to certain characteristics. By continuously displacing the grayscale centroid based on certain rules, specific structures or boundaries in an image can be gradually tracked or outlined.

[0126] By performing morphological closing operations on the image, first dilating and then eroding, the image of the center line of the laser stripe, i.e., the image of the laser line segment, is obtained.

[0127] The core of the dilation operation is to expand the coverage area of ​​the target region. Specifically, dilation merges background pixels in contact with the laser stripes into the stripes, causing the stripe boundaries to expand outward. For linear laser stripe images, any breaks (small cracks) or internal voids (small lakes) that may exist will be filled during the dilation process, connecting broken stripes into continuous areas, while the stripes as a whole become thicker.

[0128] After expansion, the stripes may become burr-like at the edges or thicken overall due to overexpansion. An erosion operation is then performed to restore the original thickness of the stripes by shrinking their boundaries. Erosion removes excess pixels introduced during expansion (such as edge burrs), retaining only the core areas that perfectly match the structural elements. Since expansion has filled in the breaks and voids, erosion does not disrupt these repaired continuous structures. The resulting stripes retain their original main shape while eliminating internal defects and edge noise.

[0129] By fitting a Gaussian distribution, the grayscale distribution pattern of the laser lines can be identified, and a threshold can be set to remove most of the background noise. Because noise typically deviates from the grayscale mean of the laser lines, statistical analysis can distinguish the laser line features from random noise, thereby improving the clarity and detectability of the laser lines. Gaussian fitting helps to highlight the grayscale characteristics of the laser lines, making them more prominent in the image and facilitating subsequent accurate extraction.

[0130] The gray-scale centroid method is used to calculate the center position of the laser line. By calculating the tangent vector of the gray-scale centroid, the principal direction of the laser line can be obtained, providing an accurate positioning reference for subsequent centerline extraction. The improved gray-scale centroid method (IGGM) introduces lateral gradient optimization, which performs a secondary adjustment of the center point position based on the lateral gray-scale variation trend of the laser line. This optimization can track the laser line more accurately, especially when the laser line width changes or there is slight deformation, ensuring the accuracy of centerline extraction. Lateral gradient optimization can enhance the algorithm's robustness to background changes or lighting conditions, maintaining high detection accuracy even in complex or harsh environments.

[0131] By employing a morphological closing operation that combines dilation and erosion, small gaps or protrusions along the center line of the laser line can be removed, smoothing the line contour and making the center line more continuous and smooth. This morphological closing operation removes noise while preserving the main features and continuity of the laser line, which is crucial for identifying subtle tearing characteristics. The smoothed laser line center line provides more reliable data for subsequent spatial mapping and tear detection, ensuring precise location of the tear and accurate assessment of its state.

[0132] Specifically, world coordinates [u w ,v w [1] is obtained by transforming the coordinates [u,v,1] on the laser centerline using the transformation matrix H. The direct linear transformation algorithm is employed, and the solution is obtained using the least squares method. According to equation w... w =Au w +Bv w +C, using [u w ,v w ,1] to get w w Where A, B, and C are the parameters of the equation, based on the multiplication operation of the transformation matrix, the coordinates in the world coordinate system are obtained by multiplying with the transformation matrix. Other information is needed; connecting the coordinate points in the world coordinate system, the centerline image is obtained, i.e., the image of the laser line segment, such as... Figure 5 As shown.

[0133] Specifically, based on linear length and rectangularity thresholds, feature thresholding is performed on the image to remove interfering regions, resulting in connected components with linear characteristics and their number N. If N < 2, the conveyor belt is not malfunctioning and is in normal operating condition. Figure 6 As shown, if N≥2, the conveyor belt is considered to have longitudinally torn, which is a fault condition.

[0134] In normal operation, the conveyor belt surface is free of tears, and the laser stripes, after being projected onto its surface, should form continuous and complete lines. At this point, the centerline image of the laser stripes is a single, continuous region, corresponding to a connected region number N = 1 (or, due to slight noise, N < 2, but close to 1). Since no tears have occurred, the laser stripes are not disrupted, and therefore, multiple independent line segments will not appear.

[0135] When the conveyor belt tears longitudinally, the crack causes severe deformation of the laser stripe at the tear point (such as breakage, displacement, or bifurcation). At this time, the originally continuous laser center line will be divided into multiple discontinuous parts by the crack at the tear point, forming multiple independent connected regions (N≥2).

[0136] In summary, N<2 indicates that the laser stripes are continuous and the conveyor belt is normal; N≥2 indicates that the laser stripes are broken and the conveyor belt has suffered longitudinal tearing.

[0137] In some embodiments, after determining that the belt is torn, the method further includes the following steps: extracting torn pixels from the line laser image, wherein the torn pixels are pixels between two laser line segments, and the laser line segments and the torn pixels are located on the same straight line; counting the number of torn pixels; obtaining a mapping relationship, wherein the mapping relationship is the relationship between the number of torn pixels and the tear length; determining the tear length corresponding to the number of torn pixels as the target tear length according to the mapping relationship; and controlling the belt conveyor to stop when the target tear length is greater than or equal to a preset length threshold.

[0138] In this solution, by setting a threshold, it can automatically determine whether the tear has reached the level of shutdown and maintenance. When the tear damage reaches the preset threshold, the conveyor will stop running immediately to prevent the damage from expanding and causing equipment failure or even safety accidents. Early intervention can not only avoid more extensive equipment damage caused by severe tearing, but also minimize the production impact caused by unplanned shutdowns.

[0139] Specifically, after confirming the belt tear, the torn pixels are extracted. Using the identified laser lines, the torn area is located, and pixels located between two laser lines, especially those fracture areas projected by the laser lines, are marked as torn pixels. These pixels are located in the anomalous areas of the laser stripes, directly reflecting the presence and extent of the tear.

[0140] Count the number of torn pixels. This step quantifies the degree of tearing; a higher number indicates a longer or wider tear. Establish a mapping relationship between the number of torn pixels and the actual tear length.

[0141] Based on the established mapping relationship, the number of torn pixels is converted into a specific tear length, thereby determining the target tear length. A preset length threshold, such as 10 centimeters, is set as a standard for determining whether an emergency shutdown is required. If the calculated target tear length is greater than or equal to this threshold, a shutdown procedure will be automatically triggered to prevent the tear from expanding further and causing greater damage.

[0142] In a specific application, within a mine conveyor belt tear detection system, once the laser-assisted detection algorithm identifies an abnormal area between two laser lines, it immediately enters tear length quantification mode. Assuming the established mapping relationship is that every 50 tear pixels correspond to a tear length of 1 centimeter, if the count of tear pixels is 550, the converted tear length is 11 centimeters. When this length reaches or exceeds a preset 10-centimeter threshold, a stop signal is immediately sent to the conveyor control system to prevent the tear damage from spreading and causing more serious consequences.

[0143] In practical applications, the preset threshold is determined based on a combination of factors, including the belt material's durability, the conveyor's operating speed, load, and past tearing incidents, aiming to balance production efficiency and equipment safety. For example, under high-load operating conditions, the preset length threshold for the conveyor may be set to a more stringent 8 centimeters to ensure that tear damage can be detected and addressed promptly even under high stress, preventing major accidents.

[0144] Specifically, the solution of this application introduces line laser assistance and realizes the transformation of line laser images from the image coordinate system to the world coordinate system through laser imaging acquisition, effectively suppressing the interference of lens distortion, thereby capturing more accurate laser stripe features.

[0145] Specifically, the solution presented in this application employs an innovative line laser fringe center extraction technique, significantly enhancing the extraction accuracy of line laser fringe centers. This achieves sub-pixel-level high-precision extraction of the laser fringe center line, mapping it onto a center line image in a world coordinate system. This advancement aims to more effectively extract the linear characteristics of the line laser fringe center line, thereby overcoming the problem of insufficient extraction accuracy in traditional line laser fringe center line extraction methods. The introduction of this technology not only helps improve the accuracy of feature recognition but also demonstrates broad prospects for practical applications.

[0146] Specifically, the innovative technology of this application significantly improves the accuracy of longitudinal tear signal identification in conveyor belts by employing advanced image segmentation strategies and precise linear feature thresholding, while simplifying the feature analysis process. Its advantage lies in effectively reducing the risk of severe longitudinal tearing events, and it is particularly suitable for obtaining unclear line laser scan images under extreme conditions such as mines. This technology is highly practical and of great significance for ensuring safe production.

[0147] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the belt tear detection method for belt conveyors of this application will be described in detail below with reference to specific embodiments.

[0148] This embodiment relates to a specific method for detecting belt tearing in a belt conveyor, such as... Figure 7 As shown, it includes the following steps:

[0149] Step S1: The video of the belt conveyor in operation is used to read the line laser image of the conveyor belt using a line laser image acquisition system;

[0150] Step S2: Crop the ROI of the captured line laser image to eliminate unnecessary elements in the image and retain only the image containing the line laser feature region;

[0151] Step S3: Extract the R channel from the image containing the line laser feature region and perform grayscale processing based on the R, G, and B color channel features of the image;

[0152] Step S4: The obtained grayscale image is processed by image enhancement processing, and grayscale adjustment is performed using an image grayscale conversion matrix to obtain a grayscale adjusted image;

[0153] Step S5: Perform laser centerline extraction on the grayscale transformed image. The laser line centerline is calculated and extracted using the improved grayscale centroid method (IGGM).

[0154] By setting a reasonable threshold, the grayscale matrix is ​​reset to reduce the impact of noise on the grayscale centroid position. Lateral gradient optimization calculations are introduced, and the position of the grayscale centroid is further optimized based on the lateral gradient trend of the laser lines. Morphological methods are then used to smooth and optimize the centerline.

[0155] Step S6: Convert the line laser centerline to the world coordinate system through spatial mapping, and generate the corresponding line laser centerline image in the world coordinate system;

[0156] Step S7: Identify linear elements in the image and extract the interconnected regions, recording the total number N of these connected regions. Assess the presence of vertical tearing in the image based on the number of connected regions N. If vertical tearing occurs, trigger an alarm.

[0157] This application also provides a belt tear detection device for a belt conveyor. It should be noted that the belt tear detection device for a belt conveyor in this application can be used to execute the belt tear detection method for a belt conveyor provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0158] The following describes the belt tear detection device for a belt conveyor provided in the embodiments of this application.

[0159] Figure 8 This is a structural block diagram of a belt tear detection device for a belt conveyor according to an embodiment of this application. Figure 8 As shown, the device includes:

[0160] The first acquisition unit 100 is used to acquire a line laser image, wherein the line laser image is a line laser image on the belt of the belt conveyor at the current moment;

[0161] The first extraction unit 200 is used to extract laser line segments from the above-mentioned line laser image, wherein the laser line segments are the line segments formed by the above-mentioned line laser on the above-mentioned belt;

[0162] The first determining unit 300 is used to determine that the belt is torn when the number of the laser segments is greater than or equal to a preset number threshold, and to determine that the belt is not torn when the number of the laser segments is less than the preset number threshold.

[0163] The first control unit 400 is used to control the alarm device to sound an alarm when it is determined that the belt has torn.

[0164] In this embodiment, a linear laser emitter can generate a continuous linear laser beam. When this linear laser beam illuminates a flat surface, ideally, it will form a straight line with uniform brightness on the surface. However, if the belt is torn, it will form an opening or crack, and the laser beam will break at these torn points, failing to form a continuous laser line segment on the belt surface. This will appear as an interruption in the laser line segment on the image. Therefore, by extracting the number of laser line segments in the belt image, it is possible to determine whether the belt is torn. For example, when the belt is torn, two laser line segments will be formed on the belt, while when the belt is not torn, one complete laser line segment will be formed on the belt. Compared with traditional pressure detection and temperature detection methods, this solution can perform detection without contacting the belt, reducing interference from equipment operation and making the tear detection accuracy of this solution higher.

[0165] In the specific implementation process, the first extraction unit includes a preprocessing module and an extraction module. The preprocessing module is used to preprocess the above-mentioned line laser image to obtain the target line laser image. The preprocessing method includes cropping and / or grayscale processing. The extraction module is used to extract the laser line segments in the above-mentioned target line laser image.

[0166] In this scheme, after the cropping operation, the image size is reduced, which can reduce the computational complexity. Grayscale conversion further highlights the brightness information of the laser lines, making the laser line area in the image more prominent, which allows for more accurate extraction of the laser line segments.

[0167] In some embodiments, the preprocessing module includes a first preprocessing submodule, a second preprocessing submodule, and a third preprocessing submodule. The first preprocessing submodule is used to perform the aforementioned cropping or grayscale processing on the line laser image to obtain the target line laser image. The second preprocessing submodule is used to perform the aforementioned cropping on the line laser image and perform the aforementioned grayscale processing on the cropped line laser image to obtain the target line laser image. The third preprocessing submodule is used to perform the aforementioned grayscale processing on the line laser image and perform the aforementioned cropping on the grayscale processed line laser image to obtain the target line laser image.

[0168] In this scheme, cropping can remove image parts that are not related to laser detection, reducing noise interference. Grayscale processing further enhances the brightness characteristics of the laser lines, making the lines stand out more in the image and easier to identify laser line segments.

[0169] In the specific implementation process, the preprocessing module is used to extract a portion of the above-mentioned line laser image to obtain the ROI region, wherein the above-mentioned ROI region at least contains the complete area of ​​the above-mentioned line laser on the belt, and the area of ​​the above-mentioned ROI region is smaller than the area of ​​the above-mentioned line laser image; the preprocessing module is used to cut off the areas in the above-mentioned line laser image other than the above-mentioned ROI region, retain the above-mentioned ROI region in the above-mentioned line laser image, and obtain the above-mentioned target line laser image.

[0170] In this scheme, the laser lines occupy a much smaller proportion of the laser image than the rest of the image, with most of the area being background. By cropping the Region of Interest (ROI), only the area containing the laser lines is retained, significantly reducing the image size. The ROI cropping operation isolates the laser line region from the entire image, narrowing the scope of image processing, reducing unnecessary computations, and accelerating subsequent image analysis.

[0171] In some embodiments, the preprocessing module is used to extract color channel features from the aforementioned line laser image to obtain R channel features, G channel features, and B channel features; the preprocessing module is used to perform grayscale processing on at least one of the aforementioned color channel features to obtain an intermediate line laser image; the preprocessing module is used to, when performing grayscale processing on one or two of the aforementioned color channel features, set the color channel features of the intermediate line laser image that have not undergone grayscale processing to zero to obtain the aforementioned target line laser image; the preprocessing module is used to, when performing grayscale processing on three of the aforementioned color channel features, determine the aforementioned intermediate line laser image as the aforementioned target line laser image.

[0172] In this scheme, since laser lines may have higher brightness and contrast in certain color channels, grayscale processing of these channels can specifically enhance the features of the laser lines. For example, if the laser light source is biased towards red, prioritizing grayscale processing of the R channel can significantly improve the recognition effect of the laser lines. The zeroing operation removes unnecessary color information, allowing the image to focus on the grayscale-processed features, thereby reducing noise during the recognition process and improving recognition accuracy.

[0173] In the specific implementation process, the extraction module includes an acquisition submodule, a training submodule, a processing submodule, a determination submodule, a connection submodule, and an optimization submodule. The acquisition submodule is used to acquire the image recognition model to be trained, wherein the image recognition model to be trained is one of the GMM model, HMRF model, U-Net model, and isolated forest model. The training submodule is used to form a training image set from the pixels in the target line laser image, and to train the image recognition model to be trained using the training image set and the corresponding category labels to obtain the category recognition model, wherein the category label is the category of the pixel in the training image set, and the category is laser, belt, or background. The processing submodule is used to sequentially input all the above-mentioned pixels in the target line laser image into the above-mentioned category recognition model to obtain the category recognition result corresponding to each of the above-mentioned pixels; the determination submodule is used to determine two adjacent above-mentioned pixels that are both of the above-mentioned laser as adjacent laser pixel pairs; the connection submodule is used to connect two above-mentioned pixels in each of the above-mentioned adjacent laser pixel pairs to form at least one initial laser line segment; the optimization submodule is used to perform image optimization on at least one above-mentioned initial laser line segment to obtain at least one above-mentioned laser line segment, wherein the above-mentioned image optimization method includes one or more of image gradient optimization, image dilation optimization, and image erosion optimization.

[0174] In this scheme, the characteristics of laser lines can be learned from a large number of labeled laser images, and then the category of pixels can be accurately identified. By accurately classifying laser lines and background, the recognition error caused by changes in light or background interference can be reduced. Image gradient optimization can identify and repair tiny breaks in laser lines through gradient calculation, while dilation optimization and erosion optimization remove small noise and fill small gaps on the lines by expanding the lines and shrinking the line edges, respectively. This ensures that the accuracy of the obtained laser line segments is high.

[0175] In some embodiments, the above-described apparatus further includes a second extraction unit, a statistics unit, a second acquisition unit, a second determination unit, and a second control unit. The second extraction unit is used to extract tear pixels from the line laser image after determining that the belt is torn, wherein the tear pixels are pixels between two laser line segments, and the laser line segments and the tear pixels are located on the same straight line. The statistics unit is used to count the number of tear pixels. The second acquisition unit is used to acquire a mapping relationship, wherein the mapping relationship is the relationship between the number of tear pixels and the tear length. The second determination unit is used to determine the tear length corresponding to the number of tear pixels as the target tear length according to the mapping relationship. The second control unit is used to control the belt conveyor to stop when the target tear length is greater than or equal to a preset length threshold.

[0176] In this solution, by setting a threshold, it can automatically determine whether the tear has reached the level of shutdown and maintenance. When the tear damage reaches the preset threshold, the conveyor will stop running immediately to prevent the damage from expanding and causing equipment failure or even safety accidents. Early intervention can not only avoid more extensive equipment damage caused by severe tearing, but also minimize the production impact caused by unplanned shutdowns.

[0177] The belt tear detection device for the aforementioned conveyor belt includes a processor and a memory. The first acquisition unit, first extraction unit, first determination unit, and first control unit are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve their respective functions. All of the above modules reside in the same processor; alternatively, the modules may be located in different processors in any combination.

[0178] The processor contains a kernel, which retrieves the corresponding program unit from memory. One or more kernels can be configured, and adjusting kernel parameters can address the low accuracy of longitudinal tear detection in existing mining belt conveyors.

[0179] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0180] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the belt tear detection method of the belt conveyor.

[0181] This invention provides a processor for running a program, wherein the program executes the belt tear detection method of the belt conveyor.

[0182] This application also provides a belt tear detection system, such as Figure 9 As shown, the system includes a belt conveyor 10, a linear laser, a camera, an alarm device, and a controller. The belt conveyor 10 includes a belt 11; the linear laser 12 is used to emit a linear laser beam onto the belt of the belt conveyor; the camera 13 is used to capture an image of the linear laser beam reflected from the belt; the alarm device 14 is used to issue an alarm; and the controller (…) Figure 9 (Not shown) is communicatively connected to the aforementioned belt conveyor, the aforementioned linear laser, the aforementioned camera, and the aforementioned alarm device, respectively. The aforementioned controller is used to execute any one of the aforementioned belt tear detection methods for the belt conveyor.

[0183] Specifically, the linear laser, camera, and controller can be located in a mounting bracket 15, which is used to hold items. This embodiment of the invention provides a device including a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements at least the steps of a method for detecting belt tearing in a conveyor belt. The device described herein can be a server, PC, PAD, mobile phone, etc.

[0184] A computer program product includes a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the belt tear detection method for the belt conveyor described in various embodiments of this application.

[0185] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0186] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0187] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0188] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0189] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0190] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0191] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0192] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0193] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0194] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for detecting belt tearing in a belt conveyor, characterized in that, include: Acquire a line laser image, wherein the line laser image is a line laser image on the belt of the belt conveyor at the current moment; Extract laser line segments from the line laser image, wherein the laser line segments are the line segments formed by the line laser on the belt; If the number of laser segments is greater than or equal to a preset threshold, the belt is determined to be torn; if the number of laser segments is less than the preset threshold, the belt is determined not to be torn. If the belt is determined to be torn, the alarm device will sound an alarm.

2. The method according to claim 1, characterized in that, Extracting laser line segments from the line laser image includes: The line laser image is preprocessed to obtain the target line laser image, wherein the preprocessing method includes cropping and / or grayscale processing; Extract the laser line segment from the target line laser image.

3. The method according to claim 2, characterized in that, The line laser image is preprocessed to obtain a target line laser image, including one of the following: The target line laser image is obtained by cropping or performing grayscale processing on the line laser image. The line laser image is cropped, and the cropped line laser image is then subjected to grayscale processing to obtain the target line laser image. The line laser image is subjected to grayscale processing, and the grayscale processed line laser image is cropped to obtain the target line laser image.

4. The method according to claim 3, characterized in that, The cropping of the line laser image to obtain the target line laser image includes: A portion of the line laser image is extracted to obtain a Region of Interest (ROI), wherein the ROI contains at least the complete area of ​​the line laser on the belt, and the area of ​​the ROI is smaller than the area of ​​the line laser image; The area other than the ROI region in the line laser image is removed, while the ROI region in the line laser image is retained, to obtain the target line laser image.

5. The method according to claim 3, characterized in that, The grayscale processing of the line laser image to obtain the target line laser image includes: The color channel features in the line laser image are extracted to obtain the R channel features, G channel features, and B channel features; The grayscale processing is performed on at least one of the color channel features to obtain the midline laser image; When performing grayscale processing on one or two of the color channel features, the color channel features of the intermediate line laser image that have not undergone grayscale processing are set to zero to obtain the target line laser image; When the grayscale processing is performed on the three color channel features, the intermediate line laser image is determined as the target line laser image.

6. The method according to claim 2, characterized in that, Extracting the laser line segment from the target line laser image includes: Obtain the image recognition model to be trained, wherein the image recognition model to be trained is one of the following: GMM model, HMRF model, U-Net model, and isolated forest model; The pixels in the target line laser image are used to form a training image set. The training image set and the corresponding category labels are used to train the image recognition model to obtain a category recognition model. The category label is the category of the pixel in the training image set, and the category is laser, belt or background. All the pixels in the target line laser image are sequentially input into the category recognition model to obtain the category recognition result corresponding to each pixel; Two adjacent pixels that are both of the laser category are identified as adjacent laser pixel pairs; Connect the two pixels in each pair of adjacent laser pixels to form at least one initial laser line segment; Image optimization is performed on at least one of the initial laser line segments to obtain at least one laser line segment, wherein the image optimization method includes one or more of image gradient optimization, image dilation optimization, and image erosion optimization.

7. The method according to claim 1, characterized in that, After determining that the belt is torn, the method further includes: Extract the tearing pixels from the line laser image, wherein the tearing pixels are pixels between two laser line segments, and the laser line segments and the tearing pixels are located in the same straight line; Count the number of torn pixels; Obtain the mapping relationship, wherein the mapping relationship is the relationship between the number of torn pixels and the tear length; Based on the mapping relationship, the tear length corresponding to the number of tear pixels is determined as the target tear length; If the target tear length is greater than or equal to a preset length threshold, the belt conveyor is controlled to stop.

8. A device for detecting belt tearing in a belt conveyor, characterized in that, include: The first acquisition unit is used to acquire a line laser image, wherein the line laser image is a line laser image on the belt of the belt conveyor at the current moment; The first extraction unit is used to extract laser line segments from the line laser image, wherein the laser line segments are the line segments formed by the line laser on the belt; The first determining unit is configured to determine that the belt is torn when the number of laser segments is greater than or equal to a preset number threshold, and to determine that the belt is not torn when the number of laser segments is less than the preset number threshold. The first control unit is used to control the alarm device to sound an alarm when it is determined that the belt is torn.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the belt tear detection method for the belt conveyor according to any one of claims 1 to 7.

10. A belt tear detection system, characterized in that, include: Belt conveyors, including belts; A line laser for emitting a linear laser beam onto the belt of the conveyor. A camera is used to acquire images of the linear laser beams reflected from the belt. Alarm devices are used to issue alarms. The controller is communicatively connected to the belt conveyor, the linear laser, the camera, and the alarm device, respectively, and is used to execute the belt tear detection method of the belt conveyor as described in any one of claims 1 to 7.

Images