A method and system for detecting the roundness and sphericity of proppants based on image analysis

Abstract

The application provides a proppant roundness and sphericity detection method and system based on image analysis, and relates to the technical field of oil and gas field fracturing proppant detection. The method comprises the following steps: collecting original images containing a plurality of independent proppant particles and calibration reference objects on a proppant particle conveying line, wherein the actual physical coordinates of the calibration reference objects are pre-calibrated; sequentially performing grayscale, binarization, morphological erosion and expansion processing on the original images to obtain preprocessed binarized images; performing edge detection on the preprocessed binarized images to extract the contour data of each particle; based on the contour data of each particle, calculating the minimum circumscribed circle of each contour, and obtaining the contour area, contour perimeter and minimum circumscribed circle radius of each particle. The application realizes automatic detection and production closed-loop management of proppant roundness and sphericity, and provides accurate data support.

Description

Technical Field

[0001] This invention relates to the field of proppant detection technology in oil and gas field fracturing, and in particular to a method and system for detecting the roundness and sphericity of proppant based on image analysis. Background Technology

[0002] Proppant is a key material in hydraulic fracturing operations in oil and gas fields, and its roundness and sphericity directly affect the conductivity of the formation fractures after fracturing. According to the industry standard SY / T5108-2014, the roundness and sphericity of ceramsite proppants must both be greater than 0.80, and the proportion of unqualified particles must not exceed 5%.

[0003] Currently, the roundness and sphericity of proppant are mainly detected by manual microscopy, which may have problems such as low detection efficiency and high subjectivity. Some researchers have tried to introduce image processing technology for automatic detection, but the following technical shortcomings still exist in data processing: Existing methods do not consider camera lens distortion, resulting in a non-linear deviation between the pixel coordinates of the particle outline and its actual physical coordinates. The particle projection in the image edge region is stretched and deformed, which may lead to qualified particles being misclassified as unqualified, with a misclassification rate as high as 8% to 12%. The modules of image acquisition, outline extraction, parameter calculation, and quality judgment only transmit the final results and do not establish an intermediate data association mechanism. When the proportion of unqualified particles exceeds the standard, it is impossible to trace which specific image or particle caused the unqualified particle, and the detection results may lack traceability. The existing system adopts an offline calibration method, and the calibration data is only used once during installation. Subsequent camera parameter drift caused by equipment vibration and temperature changes cannot be corrected in real time, and long-term operational stability cannot be guaranteed. When calibrating the camera, a conventional corner detection algorithm is used to extract feature points of the calibration board. Affected by dust and lighting on site, the positioning error is large, which may lead to a decrease in calibration accuracy and thus affect the accuracy of all particle detection data. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for detecting the roundness and sphericity of proppant based on image analysis, so as to realize automated detection of proppant roundness and sphericity and closed-loop control of production, and provide accurate data support.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: In a first aspect, a method for detecting the roundness and sphericity of proppant based on image analysis is provided, the method comprising: On the proppant particle delivery line, original images containing multiple independent proppant particles and calibration references are acquired, wherein the actual physical coordinates of the calibration references are pre-calibrated. The original image is sequentially subjected to grayscale conversion, binarization, morphological erosion, and dilation to obtain a preprocessed binarized image; Edge detection is performed on the preprocessed binarized image to extract the contour data of each particle; based on the contour data of each particle, the minimum circumcircle of each contour is calculated, and the contour area, contour perimeter and minimum circumcircle radius of each particle are obtained. Based on the outline area, outline perimeter, and minimum circumscribed circle radius of each particle, calculate the roundness and sphericity values ​​of each particle. The image coordinates of feature points on the calibration reference are extracted from the original image; the spatial transformation parameters are obtained based on the correspondence between the image coordinates of the feature points and the actual physical coordinates; the roundness and sphericity values ​​of each particle are corrected using the spatial transformation parameters to obtain the calibrated roundness and sphericity values. Based on the roundness and sphericity values ​​of all calibrated particles, the average roundness value, average sphericity value, and percentage of non-conforming particles are statistically analyzed. The percentage of non-conforming particles is then compared with the threshold specified in the industry standard to obtain the trigger judgment signal. Based on the trigger judgment signal, the alarm device and control valve are triggered to perform operations, and a quality inspection report containing test data and conclusions is synchronized to the production management platform.

[0006] Secondly, a proppant roundness and sphericity detection system based on image analysis includes: The acquisition module is used to acquire original images containing multiple independent proppant particles and calibration references on the proppant particle conveying line, wherein the actual physical coordinates of the calibration references are pre-calibrated. The preprocessing module is used to perform grayscale conversion, binarization, morphological erosion and dilation on the original image in sequence to obtain a preprocessed binarized image; The extraction module is used to perform edge detection on the preprocessed binarized image and extract the contour data of each particle; based on the contour data of each particle, the minimum circumcircle of each contour is calculated, and the contour area, contour perimeter and minimum circumcircle radius of each particle are obtained. The calculation module is used to calculate the roundness and sphericity values ​​of each particle based on its outline area, outline perimeter, and minimum circumscribed circle radius. The correction module is used to extract the image coordinates of feature points on the calibration reference in the original image; obtain spatial transformation parameters based on the correspondence between the image coordinates of the feature points and the actual physical coordinates; and use the spatial transformation parameters to correct the roundness and sphericity values ​​of each particle to obtain calibrated roundness and sphericity values. The judgment module is used to calculate the average roundness value, average sphericity value and the percentage of unqualified particles based on the roundness and sphericity values ​​of all calibrated particles, and compare the percentage of unqualified particles with the threshold specified in the industry standard to obtain the trigger judgment signal; The control module is used to trigger the alarm device and control valve to perform operations based on the trigger judgment signal, and to synchronize the quality inspection report containing the detection data and conclusions to the production management platform.

[0007] Thirdly, a computing device, comprising: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0008] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0009] The above-described solution of the present invention has at least the following beneficial effects: Raw images containing multiple independent proppant particles and calibration references are acquired on the conveyor line. The actual physical coordinates of the references are pre-calibrated to obtain the original data source and calibration benchmark, providing a complete and reliable foundation for subsequent data processing. The raw images are sequentially processed by grayscale conversion, binarization, morphological erosion, and dilation to remove noise, dust, and particle adhesion interference, resulting in clear pre-processed images. This simplifies subsequent operations, ensures accurate contour extraction, and provides a high-quality foundation for geometric parameter calculation. Edge detection is performed on the pre-processed images, contour data is extracted, and the minimum circumcircle and related geometric parameters are calculated. This captures particle geometric features and quantifies core parameters, providing support for roundness and sphericity value calculations and standardizing the parameter extraction process. Based on the particle geometric parameters, roundness and sphericity values ​​are calculated, and the geometric parameters are... The data is converted into quantifiable quality indicators to determine the morphological characteristics of individual particles, providing standardized data for subsequent quality judgment, data calibration, and batch statistics. The coordinates of feature points of the calibration reference are extracted, spatial transformation parameters are obtained, and particle morphological indicators are corrected to eliminate detection errors, ensuring that the morphological data reflects the actual state of the particles and improving the accuracy of single-particle detection. Based on the calibrated morphological indicators, relevant data are statistically analyzed and compared with industry standard thresholds to obtain trigger signals, enabling systematic statistical analysis and standardized judgment of batch quality. This provides a basis for decision-making in subsequent operations and standardizes the judgment process. Based on the trigger signals, equipment operation is triggered, quality inspection reports are obtained, and synchronized to the production management platform, enabling real-time response to quality anomalies and production control. The integrated testing data forms standard documentation, connecting testing and production control, and improving closed-loop management. Attached Figure Description

[0010] Figure 1 This is a schematic flowchart of a proppant roundness and sphericity detection method based on image analysis provided by an embodiment of the present invention.

[0011] Figure 2This is a schematic diagram of a proppant roundness and sphericity detection system based on image analysis, provided by an embodiment of the present invention. Detailed Implementation

[0012] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0013] like Figure 1 As shown, embodiments of the present invention propose a method for detecting the roundness and sphericity of proppant based on image analysis. The method includes the following steps: Step 100: On the proppant particle conveying line, acquire original images containing multiple independent proppant particles and calibration references, wherein the actual physical coordinates of the calibration references are pre-calibrated; Step 200: Perform grayscale conversion, binarization, morphological erosion and dilation on the original image in sequence to obtain a preprocessed binarized image; Step 300: Perform edge detection on the preprocessed binarized image and extract the contour data of each particle; based on the contour data of each particle, calculate the minimum circumcircle of each contour and obtain the contour area, contour perimeter and minimum circumcircle radius of each particle. Step 400: Calculate the roundness and sphericity values ​​of each particle based on its outline area, outline perimeter, and minimum circumscribed circle radius. Step 500: Extract the image coordinates of feature points on the calibration reference in the original image; obtain the spatial transformation parameters based on the correspondence between the image coordinates of the feature points and the actual physical coordinates; use the spatial transformation parameters to correct the roundness and sphericity values ​​of each particle to obtain the calibrated roundness and sphericity values. Step 600: Based on the roundness and sphericity values ​​of all calibrated particles, calculate the average roundness value, average sphericity value, and percentage of unqualified particles, and compare the percentage of unqualified particles with the threshold specified in the industry standard to obtain the trigger judgment signal; Step 700: Based on the trigger judgment signal, the alarm device and control valve are triggered to perform operations, and a quality inspection report containing test data and conclusions is synchronized to the production management platform.

[0014] In this embodiment of the invention, on the proppant particle conveying line, original images containing multiple independent proppant particles and calibration references with pre-calibrated actual physical coordinates are acquired to achieve standardized acquisition of original detection data. This provides basic image data for subsequent data processing and parameter calibration, ensuring the reliability and traceability of the detection data source. The original images are sequentially processed by grayscale conversion, binarization, morphological erosion, and dilation to effectively filter noise, dust, and surface stains from the original images, clearly separating the proppant particles from the background area, resulting in a pre-processed binarized image with clear contours and minimal interference, providing a high-quality image foundation for subsequent contour extraction. Edge detection is performed on the pre-processed binarized image, and contour data of each particle is extracted. Simultaneously, the minimum circumcircle, contour area, contour perimeter, and minimum circumcircle radius of each contour are calculated to capture the contour features of the proppant particles and obtain the core parameters required for roundness and sphericity calculations, ensuring the completeness and accuracy of parameter extraction. Roundness and sphericity values ​​are calculated based on the particle contour area, contour perimeter, and minimum circumcircle radius. This process establishes standardized indicator calculation logic to achieve precise quantification of roundness and sphericity indicators, ensuring the standardization and consistency of indicator calculations for each particle. It extracts the image coordinates of feature points from the calibration reference object, and obtains spatial transformation parameters by combining the correspondence between the feature point image coordinates and actual physical coordinates. These parameters are used to correct the roundness and sphericity values ​​of each particle, eliminating detection errors caused by angle and distance deviations during image acquisition and improving the accuracy of roundness and sphericity detection data. Based on the calibrated roundness and sphericity values, it statistically analyzes the average roundness value, average sphericity value, and the percentage of non-conforming particles. The percentage of non-conforming particles is compared with industry standard thresholds to obtain trigger judgment signals, achieving systematic statistical analysis and standardized judgment of detection data, providing a basis for subsequent operations. Based on the trigger judgment signals, alarm devices and control valves are activated, and a quality inspection report containing detection data and conclusions is synchronized to the production management platform. This enables rapid response and data synchronization of detection results, ensuring timely control of non-conforming products and providing real-time data support for production management and process adjustments.

[0015] In a preferred embodiment of the present invention, step 100 above involves acquiring an original image containing multiple independent proppant particles and a calibration reference on the proppant particle conveying line, wherein the actual physical coordinates of the calibration reference are pre-calibrated, including: Step 101: Obtain a sample of the proppant particles to be tested from the finished proppant discharge end. Specifically, this includes obtaining a sample of the proppant particles to be tested from the finished proppant discharge end. The sample to be tested is a finished particle that has just completed production and processing and has not undergone any subsequent processing. Its quality status is consistent with the product leaving the factory, which can effectively avoid the distortion of test results caused by improper sample selection and provide a basic sample that conforms to the actual production scenario for subsequent testing and data processing.

[0016] Step 102: Vibrate and sieve the sample of proppant particles to be tested to obtain particles that meet the preset particle size requirements; evenly spread the particles that meet the preset particle size requirements on the surface of a transparent conveyor belt. Specifically, this includes: considering that proppant testing must meet the requirements specified in industry standard SY / T5108-2014, and to eliminate the interference of particles with inconsistent particle sizes on subsequent contour extraction and parameter calculation, the particles with the preset particle size requirements refer to proppant particles that meet the actual needs of oil and gas field fracturing operations and industry testing specifications. The preset particle size range is determined by combining the actual application scenario of the proppant, the relevant provisions of industry standard SY / T5108-2014, and the technical requirements for subsequent image acquisition and contour extraction, and pre-setting the particle size range in the testing system, typically selecting 20 to... The 40-mesh size range, commonly used in deep well production in oil and gas fields, is taken as the preset particle size range. Based on this preset particle size requirement, a vibrating sieving device is used to vibrate and sieve the sample of proppant particles to be tested. During the sieving process, particles exceeding the preset particle size range are removed by setting a sieve that matches the preset particle size requirement, retaining only the proppant particles that meet the preset particle size requirement. After sieving, the particles that meet the preset particle size requirement are evenly spread on the surface of a transparent conveyor belt. During the spreading process, the thickness of the particles is controlled to ensure that each proppant particle is independently distributed without overlapping or stacking. This ensures that the complete outline of each particle can be clearly captured during subsequent image acquisition, providing a clear and standardized image basis for subsequent outline extraction and parameter calculation.

[0017] Step 103: Set a calibration reference on the surface of the transparent conveyor belt with evenly distributed proppant particles, and pre-calibrate the actual physical coordinates of the calibration reference. Specifically, this includes: setting a calibration reference on the surface of the transparent conveyor belt with evenly distributed proppant particles. The calibration reference is a standard part with clear feature points on its surface and dimensional accuracy that meets the testing requirements. Its material does not have significant reflective interference with the transparent conveyor belt, which can effectively avoid the influence of on-site dust and light on feature point identification. After setting, pre-calibrate the actual physical coordinates of the calibration reference. The specific pre-calibration method is to fix the calibration reference... The calibration reference is positioned within a pre-defined calibration area on the transparent conveyor belt to ensure that it does not shift or shake during the conveyor belt's operation. A high-precision measuring tool is then used to sequentially align with each feature point on the calibration reference, measuring the three-dimensional physical position of each feature point in actual space. The actual physical coordinates of each feature point are then determined. After measurement, the actual physical coordinates of all feature points are stored one-to-one in the detection system's database, completing the pre-calibration process. This pre-calibration process establishes a benchmark between image coordinates and actual physical coordinates, providing a calibration basis for subsequent lens distortion elimination, parameter drift correction, and improved detection accuracy.

[0018] Step 104: When the transparent conveyor belt carrying proppant particles and calibration references is synchronously transported to the image acquisition area, the image acquisition device acquires original images containing multiple independent proppant particles and calibration references. Specifically, this includes: defining a fixed acquisition area on the transport path of the transparent conveyor belt. This area must completely cover the camera's field of view to ensure that the proppant particles and calibration references can be completely captured when they enter the field of view simultaneously, while avoiding areas at the edges of the conveyor belt that are prone to image distortion. The preset image acquisition area is specifically defined in the middle of the transparent conveyor belt, with a length set to 10 to 15 cm and a width equal to the conveyor belt width. The width of the conveyor belt is consistent; the preset image acquisition area is preset by combining the installation position of the image acquisition device, the shooting field of view of the high-definition industrial camera, the width of the transparent conveyor belt, and the spreading range of the proppant particles; the preset speed is set based on the spreading density of the proppant particles, the image acquisition clarity requirements, and the efficiency of subsequent data processing. Combined with actual production and testing needs, the preset speed is specifically set to 0.5m / min. This speed matches the acquisition frequency of the image acquisition device to avoid problems such as blurred images, missed particle captures, or overlapping captures caused by the conveyor belt running too fast or too slow.

[0019] A transparent conveyor belt carrying proppant particles and calibration references is controlled to run at a preset speed at a uniform speed. When the transparent conveyor belt synchronously transports the proppant particles and calibration references to the preset image acquisition area, the image acquisition device is activated. The image acquisition device uses a high-definition industrial camera and is equipped with a ring-shaped supplementary lighting device. The supplementary lighting device can effectively eliminate uneven lighting and reflections on site, ensuring that the acquired original image is clear and has moderate contrast. The preset frequency of the image acquisition device is set based on the preset speed of the transparent conveyor belt, the length of the preset image acquisition area, and the number of proppant particles to be included in a single image. Combined with actual testing requirements, the preset frequency is specifically set to take one picture every 0.3 seconds. The image acquisition device takes original images containing multiple independent proppant particles and calibration references according to this preset frequency. After acquisition, the original images are transmitted to the background data processing module in real time, providing complete and synchronous image data for subsequent image preprocessing, contour extraction, parameter calculation, and calibration work based on the calibration references.

[0020] In this embodiment of the invention, the sample of the proppant to be tested is obtained from the finished product discharge end, ensuring that the sample comes directly from the finished product, guaranteeing the representativeness and authenticity of the sample, and laying the foundation for the validity of all subsequent test data; the sample of the proppant to be tested is subjected to vibration sieving to screen out particles that meet the preset particle size requirements, and at the same time, the particles that meet the requirements are evenly spread on the surface of a transparent conveyor belt to remove interfering particles that do not meet the size requirements, avoiding the influence of particle size differences on subsequent contour extraction and data calculation, while ensuring that the particles are dispersed without overlap, ensuring that the contour of each particle in the acquired image is independent and clear, providing a standardized image sample for subsequent data processing; a calibration reference is set on the surface of the transparent conveyor belt, and a calibration reference is set on the surface of the transparent conveyor belt. The actual physical coordinates are pre-calibrated to establish a reference between the image coordinates and the actual physical coordinates. This provides a reference for the subsequent acquisition of spatial transformation parameters and calibration of detection data, ensuring the accuracy of subsequent data correction. When the transparent conveyor belt carrying proppant particles and calibration references is synchronously transported to the image acquisition area, the image acquisition device acquires original images containing multiple independent proppant particles and calibration references. This achieves synchronous image acquisition of proppant particles and calibration references, ensuring that the original image simultaneously contains the detection object and the calibration reference, avoiding coordinate deviations caused by separate acquisitions. This provides complete and synchronous image data support for subsequent parameter calibration and data processing based on the calibration references.

[0021] In a preferred embodiment of the present invention, step 200 above involves sequentially performing grayscale conversion, binarization, morphological erosion, and dilation processing on the original image to obtain a preprocessed binarized image, including: Step 201 involves converting the original image to grayscale to obtain a grayscale image. Specifically, this includes converting the original color image containing proppant particles and a calibration reference object, acquired through an image acquisition device, to grayscale. During grayscale conversion, a weighted average method is used to comprehensively calculate the red, green, and blue color components of each pixel in the original image, converting the color information of each pixel into a single grayscale level value, thus completing the conversion from a color image to a grayscale image. The grayscale level values ​​of the aforementioned grayscale image range from 0 to 255, where 0 corresponds to pure black and 255 corresponds to pure white, with different values ​​corresponding to different shades of gray. This processing method effectively preserves the brightness differences and contour edge information between the proppant particles, the calibration reference object, and the background area, while removing redundant color information unrelated to morphological recognition from the color image. This provides a simple and clearly defined image foundation for subsequent binarization processing, ensuring the efficiency and stability of the subsequent data processing flow.

[0022] Step 202 involves binarizing the grayscale image to obtain an initial binarized image. Specifically, this includes: based on the brightness distribution characteristics of the grayscale image, to further enhance the target contour features of the proppant particles and the calibration reference, and to achieve separation between the target area and the background area, the grayscale image is binarized. During the binarization process, a specific and implementable binarization threshold algorithm is used. The binarization threshold algorithm refers to determining a unique critical grayscale value from the pixel grayscale level numerical distribution of the grayscale image through specific calculation logic, which serves as the dividing standard for distinguishing between the target area and the background area. This algorithm can adaptively adjust the critical grayscale value according to the actual brightness distribution of the grayscale image to ensure the accuracy of target and background separation and adapt to image brightness fluctuations caused by changes in dust and lighting conditions on site.

[0023] Specifically, the Otsu method is used as an example to implement the binarization thresholding algorithm. This method does not require manual intervention to set the critical gray value and can automatically determine the optimal critical gray value based on the pixel gray-level distribution characteristics of the gray-level image. The specific implementation process is as follows: statistically analyze the gray-level value distribution of all pixels in the gray-level image, with the gray-level value range from 0 to 255, and count the number of pixels corresponding to each gray-level. Among them, the pixels corresponding to gray-levels 0 to 80 are mainly target area pixels such as proppant particles and calibration reference objects; the pixels corresponding to gray-levels 120 to 255 are mainly background area pixels; and the pixels corresponding to gray-levels 81 to 119 are gray-level pixels. Gradient midtone pixels; calculating the inter-class variance between the target area and the background area when using different gray levels as critical gray values, and selecting the gray level with the largest inter-class variance, 100, as the critical gray value for distinguishing the target from the background. The determination of this critical gray value is based on the actual gray difference between the proppant particles, the calibration reference, and the background area. Combined with the lighting conditions, dust interference, and acquisition parameters of the high-definition industrial camera during on-site testing, after multiple experiments, it has been verified that when gray level 100 is used as the critical threshold, the best separation effect between the target area and the background area can be achieved, the inter-class variance reaches the maximum value, and the midtone interference can be effectively avoided.

[0024] Using the critical grayscale value of 100 determined by the Otsu method as the dividing criterion, the grayscale value of each pixel in the grayscale image is compared one by one with the critical grayscale value of 100. If the grayscale value of a pixel is greater than 100, the grayscale value of that pixel is set to 255, corresponding to a pure white state; if the grayscale value of a pixel is less than or equal to 100, the grayscale value of that pixel is set to 0, corresponding to a pure black state. In this way, the grayscale image is converted into an initial binary image containing only two grayscale values, 0 and 255. The target area of ​​the proppant particles and the calibration reference in the image is presented as the pure white state corresponding to 255, forming a clear and distinct boundary with the pure black background area corresponding to 0.

[0025] The aforementioned processing method can eliminate the interference of intermediate tones caused by grayscale gradients, further highlight the edge features of the target contour, delineate unambiguous target areas for subsequent morphological operations, improve the accuracy and processing efficiency of subsequent image morphology analysis, and implement a binarization threshold algorithm through the specific Otsu method to determine the critical grayscale value of 100, the grayscale value of pure white state of 255, and the grayscale value of pure black state of 0, adapting to the complex detection environment on site and ensuring the stability and consistency of binarization processing.

[0026] Step 203 involves performing morphological erosion and dilation on the initial binarized image to obtain a preprocessed binarized image. Specifically, this includes sequentially performing morphological erosion and dilation on the initial binarized image. The preset size of the structural element used in the morphological erosion and dilation is determined after multiple experiments, taking into account the actual size of small noise points and contour burrs in the initial binarized image, the preset particle size of the proppant particles, and the accuracy requirements for subsequent contour extraction. The preset size is specifically set to 3×3 pixels. This size effectively removes small noise and edge burrs caused by dust and reflection without excessively shrinking the target contour and destroying its core morphological features. In the morphological erosion process, the 3×3 pixel structural element is used to scan the initial binarized image, determining the pixel state of each structural element's coverage area. Only when all pixels within the coverage area are target area pixels is the central pixel state retained; otherwise, it is converted to background area pixels. This effectively removes isolated small noise points, appropriately shrinks the target contour, and eliminates edge burrs.

[0027] Morphological dilation processing uses the same 3×3 pixel structuring element as erosion processing. It scans the eroded image traversally, and as long as there is at least one target region pixel within the area covered by the structuring element, the center pixel is converted into a target region pixel. This compensates for the contour shrinkage caused by erosion, repairs contour edge damage and breaks, and fills in tiny holes inside the target region. After completing erosion and dilation processing in sequence, a preprocessed binarized image with clear and complete contours, a clean and interference-free background, and a shape highly consistent with the actual target is finally obtained. This provides high-quality data support for subsequent edge detection, contour extraction, and feature parameter calculation, ensuring the accuracy and consistency of subsequent data processing.

[0028] In this embodiment of the invention, the original image is converted to grayscale, simplifying the image data dimension, reducing redundant data from the color channels, and lowering the computational load of subsequent data processing. Simultaneously, it preserves the core contour differences between the proppant particles, the background, and the calibration reference, providing concise and efficient basic data for subsequent binarization. The grayscale image is then binarized, converting it into an initial binary image containing only black and white pixels. This clearly distinguishes the proppant particles, the calibration reference, and the image background, enhancing the contour features of the particles and reference, eliminating interference from grayscale gradients, and defining clear target areas for subsequent contour extraction and morphological processing. The initial binary image is then subjected to morphological erosion and dilation. Erosion removes minor noise, dust residue, and edge burrs, while dilation repairs minor damage and breaks in the particle contours, restoring the true contour morphology of the proppant particles and the calibration reference. This results in a clear, interference-free, and undamaged pre-processed binary image, providing high-quality image data support for subsequent edge detection, contour extraction, and parameter calculation, ensuring the accuracy and stability of subsequent data processing.

[0029] In a preferred embodiment of the present invention, step 300, which involves edge detection of the preprocessed binary image and extraction of contour data for each particle, and calculation of the minimum circumcircle of each contour based on the contour data, and obtaining the contour area, contour perimeter, and minimum circumcircle radius of each particle, includes: Step 301 involves edge detection on the preprocessed binarized image to extract the contours of each proppant particle, obtaining contour data for each particle. Specifically, this includes: based on the preprocessed binarized image, which has undergone noise removal and contour restoration with clear boundaries between the target and background regions, to extract the proppant particle contours and obtain reliable contour data, an edge detection algorithm combined with a line segment and rectangle intersection algorithm is used to process the preprocessed binarized image, achieving accurate extraction of the proppant particle contours. The line segment and rectangle intersection algorithm determines whether there is an intersection point between any line segment in the image and a preset rectangular region, thereby filtering out particles located within the rectangular region. The algorithm, which selects valid line segments and removes invalid line segments that exceed the rectangular area, assists the edge detection algorithm in edge filtering, further improving the accuracy of contour extraction and effectively avoiding interference from false and incomplete edges on contour data. Specifically, based on the contour range of the proppant particles, a rectangular filtering area is set, and the algorithm of the intersection of line segments and rectangles is integrated into the edge filtering stage of Canny edge detection. As a concrete and feasible example, the preset rectangular filtering area is consistent with the effective detection range of the image acquisition area, with a length of 10 to 15 cm and a width consistent with the width of the transparent conveyor belt, corresponding to an image pixel range of 800×400 pixels. The algorithm is used to filter the identified boundary line segments.

[0030] During edge detection, the Canny edge detection algorithm is used to perform a full-image scan of the preprocessed binarized image, smoothing the image to further suppress any residual minor noise and avoid false or missed edge detections due to noise interference. By calculating the gradient intensity and direction of each pixel in the image, regions where pixel grayscale values ​​change drastically are identified. These regions are the boundary areas between the proppant particles and the background, resulting in several boundary line segments. A line segment and rectangle intersection algorithm is then used to compare each identified boundary line segment with a preset 800×400 pixel rectangular screening area. It is determined whether each boundary line segment intersects with the four boundaries of the rectangular screening area. If the boundary line segment is completely within the rectangular screening area without any intersection, it is considered a valid edge line segment. If a boundary line segment partially extends beyond the rectangular screening area and intersects with the rectangular boundary, the portion of the line segment within the rectangular area is truncated, and invalid line segments extending beyond the area are discarded. If a boundary line segment completely extends beyond the rectangular screening area, the entire line segment is discarded.

[0031] By filtering through the line segment and rectangle intersection algorithm described above, false and incomplete edges are further eliminated, and complete, continuous proppant particle boundaries within the effective detection range are retained. Finally, the contours of each proppant particle are extracted, and particle contour data containing information such as contour boundary pixel coordinates and boundary pixel distribution are obtained.

[0032] Step 302: Based on the contour data of each particle, calculate the number of pixels in the region enclosed by each contour to obtain the contour area of ​​each particle. Specifically, this includes: determining the range of the region enclosed by the contour of each proppant particle according to the contour data of each particle, and determining all pixels belonging to the proppant particle within this region; counting the pixels in the region enclosed by each contour one by one, accumulating the total number of pixels in this region, and using the counted number of pixels as the contour area of ​​each particle; to ensure the accuracy of the calculation results, the statistical process strictly follows the boundary range defined by the contour data to avoid including pixels in the contour area, and at the same time, the statistical process is verified point by point to eliminate statistical errors caused by minor deviations in the contour edges, ensuring that the contour area data of each particle truly reflects the actual projected area of ​​the proppant particle.

[0033] Step 303: Based on the contour data of each particle, calculate the length of the pixel sequence on each contour boundary to obtain the contour perimeter of each particle. Specifically, this includes: extracting all pixels on the contour boundary of each proppant particle according to the contour data of each particle, and sorting these pixels sequentially in a clockwise or counterclockwise direction of the contour to form a continuous pixel sequence; calculating the distance between two adjacent pixels in the pixel sequence, and accumulating the distances of all adjacent pixels one by one, the total accumulated distance is the contour perimeter of the particle; during the calculation process, the distance is accumulated in the order of the pixel sequence to avoid pixel omissions or duplicate calculations, and the distance calculation between adjacent pixels is verified to ensure that the calculation of each distance segment is accurate, thereby ensuring the overall accuracy of the contour perimeter calculation.

[0034] Step 304: Based on the contour data of each particle, fit the minimum circumcircle of each contour to obtain the minimum circumcircle radius of each particle. Specifically, this includes: Based on the contour data of each particle, the contour data fully presents the contour shape of each proppant particle, providing sufficient morphological feature information for fitting the minimum circumcircle. Simultaneously, the minimum circumcircle radius is a key parameter for sphericity calculation, and its fitting accuracy directly affects the reliability of subsequent sphericity indices. Through the fitting processing of the contour data, the accurate acquisition of the minimum circumcircle radius of each particle is achieved. During the fitting process, based on the contour data of each particle, extract the coordinates of all pixel points on the contour boundary of each proppant particle, and then... The pixel coordinates serve as the basic data for fitting the minimum circumcircle. The least squares method is used to fit these pixels. By adjusting the center coordinates and radius of the circle, the circle is made to completely enclose the entire proppant particle outline, and the radius of the circle reaches its minimum value. This circle is the minimum circumcircle of the particle outline. After fitting, the radius value of the minimum circumcircle is extracted, which is the minimum circumcircle radius of each particle. During the fitting process, the fitting results are verified multiple times to ensure that the minimum circumcircle can completely enclose the particle outline and has the smallest radius, while avoiding fitting errors caused by small deviations in the outline edge, thus ensuring the accuracy of the minimum circumcircle radius data.

[0035] In this embodiment of the invention, edge detection is performed on the preprocessed binary image. The contour features of each proppant particle in the image are extracted using an edge detection algorithm to obtain contour data containing particle boundary information, achieving effective separation of the target particle from the background and providing complete and accurate basic feature information for subsequent contour-based parameter calculations. Based on the contour data of each particle, the number of pixels in the region enclosed by the contour is calculated to quantify the image space occupied by each proppant particle, determine the contour area of ​​each particle, and establish a quantitative index of the morphological features of the proppant particle, providing core basic parameters for subsequent roundness and sphericity calculations. Based on the contour data of each particle, the length of the contour boundary pixel sequence is calculated to accurately obtain the total boundary length of each proppant particle contour, determine the contour perimeter of each particle, realize the quantitative expression of the boundary morphology of the proppant particle, and provide necessary feature parameters for subsequent roundness and sphericity calculations. Based on the contour data of each particle, the minimum circumscribed circle of each contour is fitted to determine the minimum enclosing circular region of each proppant particle contour, obtain the minimum circumscribed circle radius, and construct the key geometric features of the proppant particle morphology, providing an important reference for subsequent roundness and sphericity calculations and ensuring the integrity and standardization of morphological parameter calculations.

[0036] In a preferred embodiment of the present invention, step 400, which calculates the roundness and sphericity values ​​of each particle based on its outline area, outline perimeter, and minimum circumscribed circle radius, includes: Step 401: Based on the outline area of ​​each particle, calculate the radius of a circle with the same area as the outline area to obtain the equivalent circular radius of each particle. Specifically, this includes: based on the outline area of ​​each particle and the minimum circumscribed circle radius of each particle, providing standardized intermediate parameters for the accurate calculation of subsequent roundness and sphericity values. Simultaneously, considering the technical deficiencies of existing detection methods, such as non-standard parameter calculations and insufficient accuracy of detection data, the equivalent circular radius of each particle is obtained through the transformation of the outline area, thus building a crucial bridge for the subsequent calculation of sphericity values. Specifically, the equivalent circular radius refers to the radius of a standard circle that is completely equal to the outline area of ​​the proppant particle. Its calculation process is completed using a corresponding formula, which is: , where r e Let r represent the equivalent circle radius, A represent the outline area of ​​a single proppant particle calculated in step 302, and π represent pi, with a value of 3.14. During the calculation, the outline area data of each particle is retrieved, and the outline area A of each particle is substituted into the above formula to calculate the ratio of the outline area A to pi π. Then, the square root of this ratio is calculated, and the result is the equivalent circle radius r of the particle. e .

[0037] After the calculation is completed, the equivalent circle radius r of each particle is determined. eIt is associated with and stored with the corresponding particle contour data, contour area data, and minimum circumscribed circle radius data to ensure the traceability of intermediate data and specifically address the problem of lack of traceability of detection results.

[0038] Step 402: Based on the outline area and outline perimeter of each particle, calculate the roundness value of each particle. The roundness value is determined by the square of the outline area and outline perimeter according to the definition of roundness. Specifically, the roundness value is used to characterize the degree of fit between the proppant particle outline and a standard circle. The calculation process is completed using a corresponding formula, which is as follows: Where C represents the roundness value, A represents the outline area of ​​a single proppant particle, L represents the outline perimeter of a single proppant particle calculated in step 303, π represents pi (π), with a value of 3.14, and 4π is a fixed coefficient. 2 The calculation represents the square of the contour perimeter. During the calculation, the contour area A and contour perimeter L of the corresponding particle stored in steps 302 and 303 are first called. A and L of the same particle are then substituted into the above formula to calculate the square value L of the contour perimeter L. 2 Next, calculate the product of the contour area A and pi (π), multiply this product by a fixed coefficient of 4 to obtain the numerator value, and then divide the numerator value by the square of the contour perimeter L. 2 The result of the calculation is the roundness value C of the particle.

[0039] After the calculation is completed, the roundness value C of each particle is associated with and stored with the corresponding intermediate data such as the contour area, contour perimeter, and equivalent circle radius. This ensures that each roundness value can be associated with a specific particle and related calculation parameters, which facilitates subsequent data traceability and error investigation.

[0040] Step 403: Based on the equivalent circular radius and minimum circumscribed circle radius of each particle, calculate the sphericity value of each particle. The sphericity value is determined by the ratio of the equivalent circular radius to the minimum circumscribed circle radius. Specifically, the sphericity value is used to characterize the degree of fit between the proppant particle profile and the standard sphere. Its calculation process is completed using a corresponding formula, which is: Where S represents the sphericity value, r e r represents the equivalent circle radius of a single proppant particle calculated in step 401. min This represents the minimum circumscribed circle radius of a single proppant particle obtained from the fitting in step 304; during the calculation, the equivalent circle radius r of the corresponding particle stored in steps 401 and 304 is first called. e and the minimum circumcircle radius r min r of the same particle e and r min Substituting this into the above formula, we get the equivalent circle radius r. e Divide by the minimum circumcircle radius rmin The result obtained from the calculation is the sphericity value S of the particle.

[0041] After the calculation is completed, the sphericity value S of each particle is associated with and stored along with all intermediate data such as the corresponding roundness value, profile parameters, and fitting parameters to form a complete single particle detection data chain. This ensures that when the proportion of non-compliance exceeds the standard, the specific particle and related calculation process can be quickly traced.

[0042] In this embodiment of the invention, based on the contour area of ​​each particle, the radius of a circle with the same area as the contour area is calculated to obtain the equivalent circular radius of each particle. The contour area of ​​the proppant particle is transformed into a standardized circular geometric parameter, establishing a correlation between the contour area and the circular feature, providing a unified and standardized basic parameter for subsequent calculation of sphericity values. Based on the contour area and contour perimeter of each particle, the roundness value is determined according to the definition of roundness. Through the correlation calculation of the contour area and the square of the contour perimeter, the degree of fit between the contour of the proppant particle and the standard circle is quantitatively characterized, achieving accurate quantification of the roundness value, determining the circular morphological characteristics of the proppant particle, and providing a core indicator for subsequent quality judgment. Based on the equivalent circular radius and the minimum circumscribed circle radius of each particle, the sphericity value is determined by the ratio of the two. Combining the equivalent circular feature and the minimum circumscribed circle feature of the particle, the degree of fit between the contour of the proppant particle and the standard sphere is quantitatively characterized, achieving accurate quantification of the sphericity value, and improving the quantitative indicators of the morphological characteristics of the proppant particle.

[0043] In a preferred embodiment of the present invention, step 500 above involves extracting the image coordinates of feature points on the calibration reference object from the original image; obtaining spatial transformation parameters based on the correspondence between the image coordinates of the feature points and their actual physical coordinates; and using the spatial transformation parameters to correct the roundness and sphericity values ​​of each particle to obtain calibrated roundness and sphericity values, including: Step 501: Identify the elliptical projection of the calibration reference object in the original image and extract the set of elliptical boundary points. Specifically, this includes: scanning the original image row by row and column by column, recording the grayscale value of each pixel point by point during the scanning process, and comparing the grayscale value of each pixel with the preset grayscale threshold range of the calibration reference object. The preset grayscale threshold range of the calibration reference object is determined after multiple experiments and verifications, taking into account the material characteristics of the calibration reference object, the lighting conditions of the original image acquisition, the intensity of dust interference on site, and the acquisition parameters of the high-definition industrial camera. The preset grayscale threshold range is specifically 180 to 220. This range can accurately lock the projection area of ​​the calibration reference object and effectively distinguish proppant particles and background noise, avoiding recognition deviations caused by grayscale overlap. Pixel areas with grayscale values ​​falling within the range of 180 to 220 are then selected. The domain is initially identified as the area suspected of being the projection of the calibration reference object. By calculating the geometric contour parameters of this suspected projection area, it is determined whether it conforms to the basic characteristics of an ellipse. During the calculation, the major axis length, minor axis length, and eccentricity of the suspected projection area are statistically analyzed. The ratio of the major axis to the minor axis and the range of the eccentricity are used to determine whether the area is elliptical. The threshold for judging the eccentricity is set to 0.6 to 0.8, and the threshold for judging the ratio of the major axis to the minor axis is set to 1.2 to 1.8. When the eccentricity of the suspected projection area is between 0.6 and 0.8 and the ratio of the major axis to the minor axis is between 1.2 and 1.8, the area is determined to be elliptical. At the same time, the influence of small compact areas corresponding to proppant particles, discrete pixel areas corresponding to background noise, and other interference areas are excluded to ensure that the identified area is only the elliptical projection area of ​​the calibration reference object formed by lens distortion.

[0044] After identifying the elliptical projection region, the region is scanned again point by point for fine detail. During the scan, each pixel is determined to be an elliptical boundary pixel. This is done by comparing the grayscale value difference between the pixel and its four adjacent pixels (up, down, left, and right) to calculate the grayscale gradient. A preset threshold for the grayscale gradient is set to 50. When the grayscale gradient reaches 50, the pixel is determined to be an elliptical boundary pixel. All pixels on the elliptical boundary are then selected point by point. Following the clockwise direction of the elliptical contour, the selected boundary pixels are organized in an orderly manner. The horizontal and vertical coordinates of each boundary pixel in the original image are recorded sequentially to form a complete set of elliptical boundary points. During the organization process, each boundary pixel is verified point by point. Discretely distributed abnormal pixels are removed, and missing boundary pixels are added to ensure that the boundary point set can completely and accurately reflect the contour shape of the elliptical projection.

[0045] Step 502: Based on the set of elliptical boundary points, calculate the intersection points of the tangents corresponding to the elliptical boundary points to obtain the coordinates of the three intersection points. Specifically, this includes: randomly selecting 12 sets of non-overlapping boundary points from the set of elliptical boundary points. Each set of boundary points contains two adjacent elliptical boundary pixels. Each set of boundary points is used to calculate an elliptical tangent. During the selection process, a set is selected every 30 boundary pixels in a clockwise direction along the elliptical contour to ensure that the 12 sets of boundary points are evenly distributed on the elliptical contour, avoiding tangent calculation deviations caused by the concentration of boundary points in a certain area of ​​the ellipse. At the same time, the 12 sets of boundary points are checked one by one, and abnormal sets with large coordinate deviations are removed to ensure that each set of boundary points can accurately reflect the local direction of the elliptical contour.

[0046] For each selected set of boundary points, the x-coordinates and y-coordinates of two adjacent boundary points in that set are recorded. By comparing the coordinate changes of the two adjacent boundary points, the difference in coordinate changes corresponding to the boundary points in that set is calculated. The direction of the ellipse tangent corresponding to that boundary point is determined based on the difference in coordinate changes, thus determining the specific direction of each tangent. During the calculation process, the coordinate change trends of adjacent boundary points are compared point by point. When the difference in x-coordinate changes is greater than the difference in y-coordinate changes, the tangent direction is determined to be a horizontal skew direction. When the difference in y-coordinate changes is greater than the difference in x-coordinate changes, the tangent direction is determined to be a vertical skew direction. At the same time, the inclination angle of the tangent is calculated to ensure that the direction of each tangent accurately corresponds to the local tangent features of the ellipse contour, avoiding inaccurate calculation of subsequent intersection points due to deviations in the tangent direction.

[0047] The 12 calculated elliptical tangents are subjected to pairwise intersection operations. The coordinates of the intersection point of each pair of tangents are calculated one by one. During the calculation, the direction parameter of each tangent is first determined. By comparing the direction parameters of two tangents, the intersection position of the two tangents is found, and the x-coordinate and y-coordinate of the intersection position are recorded as the intersection point coordinates of the two tangents. The intersection operations of all pairwise tangents are completed in sequence, resulting in a total of 66 intersection point coordinates. From the 66 intersection point coordinates, three intersection points with uniform distribution and no obvious deviation are selected. During the selection, the average x-coordinate and y-coordinate of all intersection points are calculated first. Three intersection points with a deviation of less than 5 pixels from the average value and approximately equal distances between the three intersection points are selected. The image coordinates of these three intersection points are recorded to form the three intersection point coordinate data. After the selection is completed, the three intersection point coordinates are verified point by point, and intersection points that are too far from the center of the ellipse projection are removed to ensure that the three intersection point coordinates can provide an accurate reference for the subsequent feature point coordinate extraction. This effectively avoids subsequent data processing errors caused by tangent calculation deviations and improper intersection point selection, and specifically solves the problem of insufficient positioning accuracy of calibration feature points.

[0048] Step 503: Based on the coordinates of the three intersection points and the set of elliptical boundary points, and combining the harmonic conjugate relation in projective geometry, construct constraint equations. Solve these constraint equations to obtain the image coordinates of the feature points on the calibration reference. Specifically, the harmonic conjugate relation in projective geometry refers to the specific positional association and proportional constraint between geometric elements formed by the ellipse and its tangents within the same projective plane. This relationship maintains the invariance of the relative positions between geometric elements even under conditions of lens distortion causing projection deformation, thus providing a stable and reliable geometric basis for solving the feature point coordinates. (Specific implementation details follow.) Cheng Wei, within the plane formed by elliptical projection, uses the intersection of the elliptical boundary points and tangents as basic geometric elements to determine each set of geometric points participating in harmonic conjugate relations. Based on the geometric characteristics of elliptical projection, he determines the positional conditions and correspondence rules that satisfy harmonic conjugate between each set of geometric points, ensuring that each set of geometric points maintains a stable constraint relationship under projective transformation. He also determines the core logic of harmonic conjugate relations in projective geometry, associates this logic with the set of elliptical boundary points and the coordinates of the three intersection points, uses the coordinates of the three intersection points as a basic reference, and combines the elliptical projection characteristics reflected by the set of elliptical boundary points to construct constraint equations for solving the image coordinates of feature points.

[0049] The constraint equations specifically include geometric constraints of the elliptical projection, positional constraints of the tangent intersection points, and numerical constraints of the harmonic conjugate relation. The constraint equations are constructed using a linear correlation, unifying the coordinate variables of the elliptical boundary points, the tangent intersection points, and the feature point coordinates. The equations also incorporate projection deviation compensation terms caused by lens distortion, ensuring that the equations accurately represent the actual projected positions of the feature points in the image. After the constraint equations are constructed, they are iteratively calculated by substituting the coordinates of the elliptical boundary points and the three intersection points. During the calculation, the compliance of each set of substituted data is verified according to the rules for determining the harmonic conjugate relation, eliminating abnormal data interference, and gradually converging the calculation results. The solution process follows the principles of projective geometry, progressively checking for possible deviations to ensure the accuracy of the solution results. Finally, by solving the constraint equations, the specific coordinates of each feature point on the calibration reference in the original image are obtained, i.e., the image coordinates of the feature points.

[0050] Step 504: Obtain the pre-calibrated actual physical coordinates of the feature points. Specifically, this includes: calling the actual physical coordinates of each feature point measured and stored during the pre-calibration of the calibration reference in step 103. During the calling process, ensure that the image coordinates of each feature point correspond one-to-one with the corresponding actual physical coordinates to avoid coordinate confusion or correspondence errors. After the calling is completed, verify the obtained actual physical coordinates one by one, check whether the actual physical coordinates of each feature point are consistent with the original data calibrated in step 103, and investigate any errors that may occur during data storage or calling to ensure that the obtained actual physical coordinates are accurate.

[0051] Step 505: Based on the correspondence between the image coordinates and the actual physical coordinates of the feature points, solve for the spatial transformation parameters between the image coordinate system and the physical coordinate system. Specifically, this includes: pairing the image coordinates of each feature point with its corresponding actual physical coordinates to form several sets of coordinate correspondences. These coordinate correspondences can fully reflect the coordinate deviation pattern caused by lens distortion. Based on these coordinate correspondences, analyze the mapping relationship between the image coordinate system and the actual physical coordinate system. Through quantitative analysis of the coordinate deviation, solve for the spatial transformation parameters that can eliminate this deviation. The solution process for the spatial transformation parameters must fully consider the nonlinear deviation caused by lens distortion to ensure that the solved parameters can characterize the transformation pattern between the two coordinate systems. After the solution is completed, store the spatial transformation parameters in the detection system and associate them with the corresponding original image and feature point coordinate data for easy subsequent parameter retrieval and error tracking.

[0052] Step 506: Using the spatial transformation parameters, perform geometric correction on the contour area, contour perimeter, and minimum circumscribed circle radius of each particle to obtain the corrected contour area, corrected contour perimeter, and corrected minimum circumscribed circle radius of each particle. Specifically, this includes: calling up the contour area, contour perimeter, and minimum circumscribed circle radius data of each proppant particle, and simultaneously calling up the spatial transformation parameters to associate the relevant geometric parameters of each particle with the spatial transformation parameters one by one, ensuring that the geometric parameters of each particle can be matched with the corresponding spatial transformation parameters, and performing geometric correction on each particle one by one to avoid parameter confusion or correction omissions.

[0053] During the correction process, spatial transformation parameters are used to transform the coordinates of each pixel on the particle's contour one by one. The coordinate transformation eliminates the contour stretching deformation caused by lens distortion, ensuring that the transformed pixel coordinates correspond to the actual physical coordinates of the particle. The contour area of ​​each particle is recalculated based on the transformed pixel coordinates. During the calculation, the number of all pixels enclosed by the transformed contour is counted point by point, and the total number of pixels is the corrected contour area. The contour perimeter of each particle is recalculated. According to the order of the pixels on the boundary of the transformed contour, the distance between each pair of adjacent pixels is accumulated, and the total distance is the corrected contour perimeter. The minimum circumcircle of each particle is refitted. The coordinates of all pixels on the boundary of the transformed contour are extracted. By comparing and adjusting the center position and radius of the circle point by point, the circle that can completely enclose the contour with the smallest radius is found. The radius of this circle is the corrected minimum circumcircle radius. Finally, the corrected contour area, corrected contour perimeter, and corrected minimum circumcircle radius of each particle are obtained.

[0054] After calibration, the calibrated geometric parameters are associated and stored with the original geometric parameters, spatial transformation parameters, and the contour data of the corresponding particles to establish a complete calibration data chain. This ensures that the calibration process for each particle is traceable, facilitating subsequent error investigation and data traceability, and further solving the problem of lack of traceability of test results.

[0055] Step 507: Based on the corrected contour area, corrected contour perimeter, and corrected minimum circumscribed circle radius of each particle, the roundness and sphericity values ​​of each particle are corrected to obtain calibrated roundness and sphericity values. Specifically, this includes: calling the original roundness value data of each proppant particle, and simultaneously retrieving the corresponding original calculation logic parameters from step 402, i.e., the ratio of the square of the contour area to the square of the contour perimeter. Then, the corrected contour area and corrected contour perimeter of the particle are obtained simultaneously. The original roundness values ​​and corrected geometric parameters are matched one by one to ensure that each set of calculated data corresponds to a unique particle identifier. According to the established roundness calculation logic, the corrected contour area and corrected contour perimeter are re-substituted to recalculate the roundness value of each particle. During the calculation process, the ratio of the contour area to the square of the contour perimeter is followed. The corrected parameters are substituted one by one to complete the numerical calculation to obtain the calibrated roundness value of the particle.

[0056] The original sphericity value data of each proppant particle is retrieved, and the corresponding original calculation logic parameters from step 403 are retrieved simultaneously, namely the ratio of the equivalent circle radius to the minimum circumscribed circle radius. At the same time, the equivalent circle radius data and the corrected minimum circumscribed circle radius corresponding to the particle are obtained, and the original sphericity value is associated with the corrected geometric parameters. According to the established sphericity calculation logic, the equivalent circle radius and the corrected minimum circumscribed circle radius are re-substituted to recalculate the sphericity value of each particle. During the calculation process, the numerical solution is completed according to the operation rule of dividing the equivalent circle radius by the corrected minimum circumscribed circle radius, and the calibrated sphericity value of the particle is obtained.

[0057] After all the correction calculations are completed, the calibrated roundness value, calibrated sphericity value, and the corresponding corrected contour area, corrected contour perimeter, corrected minimum circumscribed circle radius, as well as the original morphological parameters, the contour data of the corresponding particles, and the spatial transformation parameters of step 505 are stored in a full-dimensional association to build a complete single-particle detection data link, ensuring that each calibration index can be traced back to the specific original data and calculation process.

[0058] In this embodiment of the invention, the elliptical projection of the calibration reference object is identified in the original image, the set of elliptical boundary points is extracted, the projection shape of the calibration reference object in the image is captured, and the basic feature data of the elliptical boundary is obtained, providing a complete and reliable boundary basis for the subsequent extraction of feature point image coordinates. Based on the set of elliptical boundary points, the intersection points of the tangents corresponding to the elliptical boundary points are calculated, obtaining the coordinates of the three intersection points. Key feature points are located through the tangent intersection points, and the core coordinate data used for feature point calculation is selected, providing necessary coordinate support for the subsequent construction of constraint equations and simplifying the feature point image coordinate extraction process. Based on the coordinates of the three intersection points and the set of elliptical boundary points, constraint equations are constructed using the harmonic conjugate relation in projective geometry. The feature point image coordinates on the calibration reference object are obtained by solving the constraint equations. The construction and solution of constraint equations are standardized based on the principles of projective geometry, achieving accurate extraction of feature point image coordinates and ensuring the accuracy of the correspondence between image coordinates and actual physical coordinates. The pre-calibrated actual physical coordinates of the feature points are obtained, and the previously pre-calibrated physical coordinate data is called. A baseline is established to correspond the image coordinates of feature points to their actual physical coordinates, providing core comparative data for solving subsequent spatial transformation parameters. Based on the correspondence between the image coordinates and the actual physical coordinates of feature points, the spatial transformation parameters between the image coordinate system and the physical coordinate system are solved, establishing the association logic between the two coordinate systems, quantifying the deviation between the coordinate systems, and obtaining standardized parameters that can be used for geometric correction, providing a standardized and unified basis for subsequent correction of detection data. Using the spatial transformation parameters, the contour area, contour perimeter, and minimum circumscribed circle radius of each particle are geometrically corrected to eliminate parameter errors caused by lens distortion and coordinate deviation during image acquisition, restore the true values ​​of particle geometric parameters, and ensure the accuracy of geometric parameters. Based on the corrected contour area, corrected contour perimeter, and corrected minimum circumscribed circle radius of each particle, the roundness and sphericity values ​​of each particle are corrected. Combined with the corrected geometric parameters, the calculation results of roundness and sphericity values ​​are improved to obtain calibration indicators that fit the actual shape of the particles, ensuring the authenticity and standardization of morphological detection data.

[0059] In a preferred embodiment of the present invention, step 600, based on the roundness and sphericity values ​​of all calibrated particles, calculates the average roundness value, average sphericity value, and percentage of defective particles, and compares the percentage of defective particles with a threshold specified in the industry standard to obtain a trigger judgment signal, including: Step 601 involves obtaining the calibrated roundness and sphericity values ​​for each particle. This includes: establishing a data association mechanism with Step 507; retrieving the calibrated roundness and sphericity values ​​of each proppant particle from the database of the detection system batch by batch, according to the unique particle identifier. During the retrieval process, the integrity of the data link is ensured, meaning that each calibrated morphological indicator can correspond to the specific original image, contour data, and spatial transformation parameters, avoiding data omissions or correspondence errors. All retrieved calibrated roundness and sphericity values ​​are uniformly classified and archived, arranged in an orderly manner according to the particle acquisition time sequence or image number sequence, forming a complete particle detection data set. This provides a clear and standardized data source for subsequent threshold comparison and average value calculation, while also ensuring the complete retrieval and archiving of intermediate data.

[0060] Step 602: Based on the roundness and sphericity values ​​of all calibrated particles, compare the roundness value of each particle with the roundness threshold specified in the industry standard, and compare the sphericity value of each particle with the sphericity threshold specified in the industry standard. Count the number of particles that meet at least one of the following criteria: roundness value below the roundness threshold, sphericity value below the sphericity threshold, to obtain the number of unqualified particles. Calculate the proportion of unqualified particles based on the number of unqualified particles and the total number of particles. Specifically, this includes: determining the roundness threshold and sphericity threshold specified in industry standard SY / T5108-2014, where both the roundness threshold and sphericity threshold are 0.80. The threshold is used as the criterion for determining whether a single particle is qualified or not. The calibrated roundness value and sphericity value of each particle are compared one by one. During the comparison, the calibrated roundness value of a single particle is compared with the roundness threshold of 0.80, and the calibrated sphericity value of the particle is compared with the sphericity threshold of 0.80. It is determined whether the calibrated roundness value or the calibrated sphericity value of a single particle is lower than 0.80. If either condition is met, the particle is determined to be unqualified. The comparison results of each particle are recorded point by point during the comparison process to ensure that the judgment basis of each particle is traceable.

[0061] During the statistical process, a dedicated counting variable is established, initially set to 0. All particles deemed unqualified are enumerated one by one. For each confirmed unqualified particle, the count variable is incremented by 1. After enumeration, the final value of the count variable represents the number of unqualified particles. Simultaneously, the total number of all proppant particles involved in this test is counted. During the count, each particle in the data set integrated in step 601 is counted individually to ensure no omissions or duplications in the total particle count. The percentage of unqualified particles is calculated by dividing the number of unqualified particles by the total number of particles. The specific values ​​of the number of unqualified particles and the total number of particles are then used as the dividend and the total number of particles as the divisor, performing the division operation digit by digit to obtain the specific percentage of unqualified particles. The calculation results are verified step-by-step during the calculation process to avoid deviations in the percentage due to numerical statistical errors or calculation mistakes.

[0062] After the statistics are completed, the number of unqualified particles, the total number of particles, and the calculated percentage of unqualified particles are linked and stored with the corresponding particle identification information, calibrated roundness value, calibrated sphericity value, and comparison results. This ensures that the judgment result of each unqualified particle can be traced back to the specific comparison process, calculation process, and original data, and that the judgment result of each qualified particle can also correspond to specific numerical basis, further improving the standardization and reliability of quality statistics.

[0063] Step 603 involves comparing the percentage of non-conforming particles with the threshold value of the allowable range for non-conforming particles specified in the industry standard to obtain a trigger judgment signal indicating whether the percentage of non-conforming particles exceeds the standard. Specifically, this includes: obtaining the threshold value of the allowable range for non-conforming particles specified in the industry standard, using this threshold value as the critical standard for judging whether the quality of the proppant batch is qualified, and directly comparing the calculated percentage of non-conforming particles with this allowable range threshold value; during the comparison process, if the percentage of non-conforming particles is within the allowable range threshold value, a trigger judgment signal indicating that the percentage of non-conforming particles has not exceeded the standard is obtained; if the percentage of non-conforming particles exceeds the allowable range threshold value, a trigger judgment signal indicating that the percentage of non-conforming particles has exceeded the standard is generated; after the signal is generated, the trigger judgment signal is associated and stored with the corresponding percentage of non-conforming particles data and the original particle detection data to form a complete quality judgment data chain, ensuring that the quality judgment result is traceable and has instruction attributes, and providing standardized signal support for feedback and adjustment in subsequent production processes.

[0064] Step 604: Based on the roundness and sphericity values ​​of all calibrated particles, calculate the average roundness and average sphericity values ​​of all particles. Specifically, this includes: summing the roundness values ​​of all calibrated particles to obtain the total roundness value of all particles, and then dividing this total by the total number of particles to obtain the average roundness value of all particles. During the calculation process, the data is substituted one by one and verified according to the logic of numerical summation and division to ensure the accuracy of the calculation results. The same summation and division operation is performed on the sphericity values ​​of all particles to obtain the average sphericity value of all particles. After the calculation is completed, the average roundness and average sphericity values ​​are associated and stored with the corresponding particle detection data and the trigger judgment signal of step 603 to construct a complete data system containing single particle data, statistical data, and quality signals. This provides comprehensive core indicators for comprehensively evaluating the overall quality performance of proppant batches and guiding the continuous optimization of production processes, further improving the indicator system and data application value of proppant quality testing.

[0065] In this embodiment of the invention, the roundness and sphericity values ​​of each particle after calibration are obtained, morphological index data are called, and the calibrated test data of all particles are integrated to provide complete and unified basic data for subsequent statistical analysis, threshold comparison and quality judgment. Based on the roundness and sphericity values ​​of all particles after calibration, the roundness and sphericity values ​​of each particle are compared with the corresponding thresholds specified in the industry standard. The number of particles that meet at least one of the standards is counted, and the proportion of unqualified particles is calculated to quantitatively characterize the overall quality status of the proppant particles, determine the distribution of unqualified particles, and establish standardized quality statistical logic to provide core statistical data for subsequent quality judgment and trigger signal acquisition. By comparing the percentage of non-compliant particles with the allowable threshold for non-compliant percentages specified in industry standards, it is determined whether the overall quality of the proppant meets industry specifications. This provides a trigger signal indicating whether the percentage of non-compliant particles exceeds the standard, offering a basis for subsequent quality control and process adjustments. This ensures that the test results conform to industry standard requirements and guarantees the standardization of proppant quality assessment. Based on the calibrated roundness and sphericity values ​​of all particles, the average roundness and sphericity values ​​of all particles are calculated to comprehensively reflect the overall morphological characteristics of the proppant particle group. This supplements the deficiencies of single-particle test data, improves the indicator system for proppant quality testing, and provides data support for proppant production process optimization and overall quality assessment.

[0066] In a preferred embodiment of the present invention, step 700, based on a trigger judgment signal, triggers the alarm device and control valve to perform operations, and obtains a quality inspection report containing detection data and conclusions, which is then synchronized to the production management platform, including: Step 701: According to the trigger judgment signal, determine whether the proportion of unqualified particles exceeds the threshold of the allowable range of the unqualified proportion specified in the industry standard, and obtain the judgment result. Specifically, it includes: analyzing the trigger judgment signal, extracting the proportion value of unqualified particles carried in the signal and the threshold of the allowable range of the unqualified proportion specified in the industry standard. This allowable range threshold strictly follows the 5% specified in the industry standard SY / T5108-2014. Then, directly and precisely compare the two values. During the comparison process, check one by one the size relationship between the specific value of the proportion of unqualified particles and the threshold of the allowable range. If the value of the proportion of unqualified particles is within the threshold of the allowable range, it is determined that the quality inspection result of this batch of proppants is qualified, and a corresponding qualified judgment result is generated. If the value of the proportion of unqualified particles exceeds the threshold of the allowable range, it is determined that the quality inspection result of this batch of proppants is unqualified, and a corresponding unqualified judgment result is generated. After the judgment result is generated, it is associated and stored with the corresponding proportion data of unqualified particles and the industry standard threshold, ensuring that the quality judgment process is well-documented and data is traceable, and further solving the problem of lack of traceability of inspection results.

[0067] Step 702: If the judgment result is yes, send an alarm trigger instruction to the alarm device and at the same time send a valve control instruction to the control valve, so that the alarm device performs an alarm operation and the control valve performs a preset control action. Specifically, it includes: performing real-time verification on the judgment result. If the judgment result is that the unqualified proportion exceeds the standard, two standardized instructions are generated simultaneously, namely the alarm trigger instruction and the valve control instruction. The alarm trigger instruction is transmitted to the alarm device supporting the detection system. The instruction contains alarm type and alarm level parameters. After receiving the instruction, the alarm device performs a preset alarm operation, including audible and visual alarms and information prompts, to timely remind the on-site staff to handle the quality abnormality.

[0068] Simultaneously, valve control commands are transmitted to the control valves on the production line. These commands include parameters for valve opening / closing duration and opening / closing range. Upon receiving the command, the control valve, based on the deviation value of the non-conforming particle percentage carried in the command, invokes the preset control actions and switching conditions in the detection system to execute the corresponding preset control actions. The preset control actions and preset methods of the control valves are as follows: The preset control actions mainly include two switchable action modes: emergency shutdown and material conveying adjustment. The emergency shutdown action completely closes the proppant discharge channel, while the material conveying adjustment action reduces the material conveying speed to a preset minimum value and maintains stable conveying. The preset method was determined after multiple production tests and equipment debugging, taking into account proppant production process requirements, non-conforming batch handling procedures, the model parameters of the control valves on the production line, and relevant requirements for non-conforming product control in industry standards. Simultaneously, the switching conditions for the two action modes are preset; when the non-conforming particle percentage exceeds... When the proportion of non-conforming particles exceeds the industry standard threshold of 5% and the deviation is within 2%, a material conveying adjustment action is executed. When the proportion of non-conforming particles exceeds the industry standard threshold of 5% and the deviation is 2% or higher, an emergency shutdown action is executed. After the preset is completed, the control action parameters and switching conditions are stored in the detection system for easy retrieval when the command is issued. In specific execution, if the proportion of non-conforming particles exceeds 5% and the deviation is within 2%, a material conveying adjustment action is executed to reduce the material conveying speed to the preset minimum value and maintain stability. If the proportion of non-conforming particles exceeds 5% and the deviation is 2% or higher, an emergency shutdown action is executed to completely close the discharge channel and promptly stop the continuous production and outflow of non-conforming proppant batches. During the command transmission and equipment execution process, an association storage is established between the command sending record and the device operation record to ensure that every alarm and valve control operation can be traced back to the specific judgment result and trigger signal, providing complete operational data support for subsequent production problem investigation and equipment maintenance.

[0069] Step 703: Based on the average roundness value, average sphericity value, percentage of non-conforming particles, and the judgment result, a quality inspection report containing test data and test conclusions is obtained. Specifically, this includes: retrieving the average roundness value, average sphericity value, percentage of non-conforming particles, and judgment result; establishing a data association relationship according to the particle test data number and batch collection number to ensure accurate and unconfused correspondence of various data types; and calling a preset quality inspection report template in the testing system. The preset quality inspection report template and its preset method are as follows: the preset quality inspection report template is a standardized fixed format template, which sequentially includes five core modules: a basic information module, a testing equipment information module, a core test data module, a quality judgment conclusion module, and a traceability association module. The basic information module is used to fill in the support batch number, production batch, test batch, test date, and testing personnel information. The testing equipment information module is used to fill in the support batch number, production batch, test batch, test date, and testing personnel information. The information module is used to fill in the industrial camera model, calibration reference parameters, and detection system version information used in the test. The core detection data module is used to fill in the average roundness value, average sphericity value, percentage of non-conforming particles, and total number of particles. The quality judgment conclusion module is used to fill in the judgment result and judgment basis. The traceability association module is used to fill in the original data number of the corresponding particle detection, the spatial transformation parameter number, and the trigger judgment signal number. The preset method is determined after multiple format debugging, data adaptation tests, and production verifications, based on the relevant provisions of industry standard SY / T5108-2014, the format requirements of the production management platform for quality inspection reports, the full-process data traceability requirements of proppant quality testing, and the actual operating habits of on-site testing and production control. After the template is preset, it is stored in the detection system, and a data entry interface is reserved to facilitate the rapid entry and format unification of subsequent test data.

[0070] Following the five core modules pre-set in the template, the basic information of the proppant batch, testing time, and testing equipment parameters are integrated sequentially. Core testing data such as average roundness value, average sphericity value, and percentage of non-conforming particles are entered. The judgment result is marked in the quality judgment conclusion module, and the judgment basis is supplemented, namely the roundness and sphericity threshold and the non-conforming percentage threshold specified by industry standards, which serve as the final conclusion of this batch's quality testing. During the report generation process, the format of each entered data is standardized and the content is verified. The completeness of data entry, numerical accuracy, and standardized expression of each module are checked, and errors or format deviations are corrected in a timely manner to ensure that the data expression is standardized, the numerical values ​​are accurate, the conclusions are clearly expressed, and the judgment basis is clear, ultimately forming a complete quality inspection report. After the quality inspection report is generated, it is associated and stored with the corresponding original particle testing data, spatial transformation parameters, and trigger judgment signals. The corresponding data number is entered in the traceability association module of the template to construct a complete data chain containing testing data, statistical data, and judgment conclusions, realizing full-process traceability of the quality inspection report. This provides comprehensive and standardized core information for data synchronization with the production management platform and subsequent production process optimization, further improving the closed-loop management system for proppant quality testing.

[0071] Step 704 involves synchronizing the quality inspection report to the production management platform. This includes: establishing a standardized data interface between the testing system and the production management platform; determining the interface transmission protocol and data format requirements to ensure that the quality inspection report can be transmitted in accordance with the platform's required format; transmitting the quality inspection report to the production management platform in real time through the interface, including the report generation time, batch number, and data verification identifier during transmission to ensure the integrity and accuracy of the report transmission; after receiving the quality inspection report, the platform automatically parses, archives, and displays the report, allowing production management personnel to view the quality testing data and conclusions of this batch of proppant in real time. It also supports the production management platform in retrieving, statistically analyzing, and subsequently accessing the quality inspection report, providing data support for production process optimization, batch quality traceability, and production plan adjustments; after data synchronization is complete, the synchronization time, synchronization status, and platform feedback are recorded to form a complete synchronization operation record, ensuring information exchange between testing data and production management processes, and improving the synergy and efficiency of production management.

[0072] In this embodiment of the invention, based on the trigger judgment signal, it is determined whether the proportion of unqualified particles exceeds the threshold of the allowable unqualified proportion specified in the industry standard. This determines the core judgment result of the proppant batch quality, connects the previous statistical data with the subsequent execution operation, ensures the consistency of the judgment logic, and provides a decision basis for subsequent alarms, valve control, and quality inspection report generation. If the judgment result is that the unqualified proportion exceeds the standard, an alarm trigger command is issued to the alarm device, and a valve control command is issued to the control valve to achieve real-time response to quality anomalies. This drives the alarm device to perform alarm operation and the control valve to perform preset control actions, promptly reminding staff to handle quality problems and preventing the continuous production of unqualified products. This enhances the timeliness and proactivity of quality control. Based on average roundness, average sphericity, the proportion of non-conforming particles, and judgment results, a quality inspection report containing test data and conclusions is generated. The system systematically sorts out the core data of the entire testing process, determines the core information of the test results, and forms a complete quality inspection certificate. This achieves systematic integration and standardized presentation of test data, and improves the closed-loop management of the testing process. The quality inspection report is synchronized to the production management platform to achieve real-time sharing of test data and conclusions. This allows the production management end to keep abreast of the batch quality status of the proppant, connects the quality testing and production management links, and provides timely and comprehensive data support for production process adjustment and quality control optimization.

[0073] like Figure 2 As shown, embodiments of the present invention also provide a proppant roundness and sphericity detection system based on image analysis, comprising: The acquisition module is used to acquire original images containing multiple independent proppant particles and calibration references on the proppant particle conveying line, wherein the actual physical coordinates of the calibration references are pre-calibrated. The preprocessing module is used to perform grayscale conversion, binarization, morphological erosion and dilation on the original image in sequence to obtain a preprocessed binarized image; The extraction module is used to perform edge detection on the preprocessed binarized image and extract the contour data of each particle; based on the contour data of each particle, the minimum circumcircle of each contour is calculated, and the contour area, contour perimeter and minimum circumcircle radius of each particle are obtained. The calculation module is used to calculate the roundness and sphericity values ​​of each particle based on its outline area, outline perimeter, and minimum circumscribed circle radius. The correction module is used to extract the image coordinates of feature points on the calibration reference in the original image; obtain spatial transformation parameters based on the correspondence between the image coordinates of the feature points and the actual physical coordinates; and use the spatial transformation parameters to correct the roundness and sphericity values ​​of each particle to obtain calibrated roundness and sphericity values. The judgment module is used to calculate the average roundness value, average sphericity value and the percentage of unqualified particles based on the roundness and sphericity values ​​of all calibrated particles, and compare the percentage of unqualified particles with the threshold specified in the industry standard to obtain the trigger judgment signal; The control module is used to trigger the alarm device and control valve to perform operations based on the trigger judgment signal, and to synchronize the quality inspection report containing the detection data and conclusions to the production management platform.

[0074] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0075] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0076] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0077] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0078] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting the roundness and sphericity of proppant based on image analysis, characterized in that, The method includes: Step 100: On the proppant particle conveying line, acquire original images containing multiple independent proppant particles and calibration references, wherein the actual physical coordinates of the calibration references are pre-calibrated; Step 200: Perform grayscale conversion, binarization, morphological erosion and dilation on the original image in sequence to obtain a preprocessed binarized image; Step 300: Perform edge detection on the preprocessed binarized image and extract the contour data of each particle; based on the contour data of each particle, calculate the minimum circumcircle of each contour and obtain the contour area, contour perimeter and minimum circumcircle radius of each particle. Step 400: Calculate the roundness and sphericity values ​​of each particle based on its outline area, outline perimeter, and minimum circumscribed circle radius. Step 500: Extract the image coordinates of feature points on the calibration reference in the original image; obtain the spatial transformation parameters based on the correspondence between the image coordinates of the feature points and the actual physical coordinates; use the spatial transformation parameters to correct the roundness and sphericity values ​​of each particle to obtain the calibrated roundness and sphericity values. Step 600: Based on the roundness and sphericity values ​​of all calibrated particles, calculate the average roundness value, average sphericity value, and percentage of unqualified particles, and compare the percentage of unqualified particles with the threshold specified in the industry standard to obtain the trigger judgment signal; Step 700: Based on the trigger judgment signal, the alarm device and control valve are triggered to perform operations, and a quality inspection report containing test data and conclusions is synchronized to the production management platform.

2. The proppant roundness and sphericity detection method based on image analysis according to claim 1, characterized in that, Step 100 includes: Obtain a sample of the proppant particles to be tested from the finished proppant discharge end; The sample of proppant particles to be tested is subjected to vibration sieving to obtain particles that meet the preset particle size requirements; the particles that meet the preset particle size requirements are then evenly spread on the surface of a transparent conveyor belt. A calibration reference is set on the surface of a transparent conveyor belt with proppant particles evenly spread, and the actual physical coordinates of the calibration reference are pre-calibrated. When the transparent conveyor belt carrying proppant particles and calibration references is synchronously transported to the image acquisition area, the image acquisition device acquires original images containing multiple independent proppant particles and calibration references.

3. The proppant roundness and sphericity detection method based on image analysis according to claim 2, characterized in that, Step 200 includes: The original image is converted to grayscale to obtain a grayscale image; The grayscale image is binarized to obtain an initial binarized image; The initial binarized image is subjected to morphological erosion and dilation to obtain a preprocessed binarized image.

4. The proppant roundness and sphericity detection method based on image analysis according to claim 3, characterized in that, Step 300 includes: Edge detection is performed on the preprocessed binarized image to extract the contours of each proppant particle, thus obtaining the contour data of each particle. Based on the contour data of each particle, the number of pixels in the region enclosed by each contour is calculated to obtain the contour area of ​​each particle. Based on the contour data of each particle, the length of the pixel sequence on each contour boundary is calculated to obtain the contour perimeter of each particle. Based on the contour data of each particle, the minimum circumcircle of each contour is fitted to obtain the minimum circumcircle radius of each particle.

5. The proppant roundness and sphericity detection method based on image analysis according to claim 4, characterized in that, Step 400 includes: Based on the outline area of ​​each particle, the radius of a circle with the same area as the outline area is calculated to obtain the equivalent circle radius of each particle. Based on the outline area and outline perimeter of each particle, the roundness value of each particle is calculated. The roundness value is determined by the square of the outline area and the outline perimeter according to the definition of roundness. Based on the equivalent circle radius and the minimum circumscribed circle radius of each particle, the sphericity value of each particle is calculated, and the sphericity value is determined by the ratio of the equivalent circle radius to the minimum circumscribed circle radius.

6. The proppant roundness and sphericity detection method based on image analysis according to claim 5, characterized in that, Step 500 includes: Identify the elliptical projection of the calibration reference in the original image and extract the set of elliptical boundary points; Based on the set of elliptical boundary points, the intersection points of the tangents corresponding to the elliptical boundary points are calculated to obtain the coordinates of the three intersection points; Based on the coordinates of the three intersection points and the set of elliptical boundary points, a constraint equation is constructed using the harmonic conjugate relation in projective geometry. The image coordinates of the feature points on the calibration reference are obtained by solving the constraint equation. Obtain the pre-calibrated actual physical coordinates of the feature points; Based on the correspondence between the image coordinates and the actual physical coordinates of the feature points, the spatial transformation parameters between the image coordinate system and the physical coordinate system are solved. Using the aforementioned spatial transformation parameters, the contour area, contour perimeter, and minimum circumscribed circle radius of each particle are geometrically corrected to obtain the corrected contour area, corrected contour perimeter, and corrected minimum circumscribed circle radius of each particle. Based on the corrected contour area, corrected contour perimeter, and corrected minimum circumscribed circle radius of each particle, the roundness and sphericity values ​​of each particle are corrected to obtain the calibrated roundness and sphericity values.

7. The proppant roundness and sphericity detection method based on image analysis according to claim 6, characterized in that, Step 600 includes: Obtain the roundness and sphericity values ​​of each particle after calibration; Based on the roundness and sphericity values ​​of all calibrated particles, the roundness value of each particle is compared with the roundness threshold specified in the industry standard, and the sphericity value of each particle is compared with the sphericity threshold specified in the industry standard. The number of particles that meet at least one of the following criteria is counted: roundness value below the roundness threshold and sphericity value below the sphericity threshold. The number of unqualified particles is obtained, and the percentage of unqualified particles is calculated based on the number of unqualified particles and the total number of particles. The percentage of non-conforming particles is compared with the threshold of the allowable range of non-conforming percentage specified in the industry standard to obtain a trigger judgment signal indicating whether the percentage of non-conforming particles exceeds the standard. Based on the roundness and sphericity values ​​of all particles after calibration, calculate the average roundness and average sphericity values ​​of all particles.

8. The proppant roundness and sphericity detection method based on image analysis according to claim 7, characterized in that, Step 700 includes: Based on the trigger judgment signal, determine whether the proportion of unqualified particles exceeds the threshold of the allowable range of unqualified proportion specified in the industry standard, and obtain the judgment result; If the judgment result is yes, an alarm trigger command is sent to the alarm device, and a valve control command is sent to the control valve, so that the alarm device performs the alarm operation and the control valve performs the preset control action. Based on the average roundness value, average sphericity value, percentage of non-conforming particles, and the judgment result, a quality inspection report containing test data and test conclusions is obtained. The quality inspection report will be synchronized to the production management platform.

9. A proppant roundness and sphericity detection system based on image analysis, wherein the system implements the method as described in any one of claims 1 to 8, characterized in that, include: The acquisition module is used to acquire original images containing multiple independent proppant particles and calibration references on the proppant particle conveying line, wherein the actual physical coordinates of the calibration references are pre-calibrated. The preprocessing module is used to perform grayscale conversion, binarization, morphological erosion and dilation on the original image in sequence to obtain a preprocessed binarized image; The extraction module is used to perform edge detection on the preprocessed binarized image and extract the contour data of each particle; based on the contour data of each particle, the minimum circumcircle of each contour is calculated, and the contour area, contour perimeter and minimum circumcircle radius of each particle are obtained. The calculation module is used to calculate the roundness and sphericity values ​​of each particle based on its outline area, outline perimeter, and minimum circumscribed circle radius. The correction module is used to extract the image coordinates of feature points on the calibration reference in the original image; obtain spatial transformation parameters based on the correspondence between the image coordinates of the feature points and the actual physical coordinates; and use the spatial transformation parameters to correct the roundness and sphericity values ​​of each particle to obtain calibrated roundness and sphericity values. The judgment module is used to calculate the average roundness value, average sphericity value and the percentage of unqualified particles based on the roundness and sphericity values ​​of all calibrated particles, and compare the percentage of unqualified particles with the threshold specified in the industry standard to obtain the trigger judgment signal; The control module is used to trigger the alarm device and control valve to perform operations based on the trigger judgment signal, and to synchronize the quality inspection report containing the detection data and conclusions to the production management platform.

10. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 8.

Images