Automatic steel pipe unstacking method and system based on linear laser projection

Abstract

The present application belongs to but is not limited to the technical field of computer vision, and particularly relates to a steel pipe automatic unstacking method and system based on line laser projection. The method projects line laser to the top of a steel pipe stack, acquires a laser reflection curve image of the surface of the stack, and performs image preprocessing on the image. Then, a deep learning model YOLOv8l is used for inference analysis to accurately identify the position of the top steel pipe in the image. By capturing images at two preset measurement points, a mapping relationship between image pixel coordinates and real world coordinates is established, and the movement distance required for the electromagnetic clamping jaw to reach the top steel pipe can be accurately calculated. The present application can greatly improve the automation and intelligence level of steel pipe unstacking operation, and significantly reduce hardware cost.

Description

Technical Field

[0001] This invention belongs to, but is not limited to, the field of computer vision technology, and particularly relates to an automated destacking method and system for steel pipes based on line laser projection. Background Technology

[0002] Steel pipe destacking, as the final process before steel pipe processing, involves the effective separation of stacked steel pipes. Currently, steel pipe destacking technology still faces some practical problems: 1. In the steel pipe processing industry, some steel pipes exceed 10 meters in length and weigh over 200 kilograms, making manual destacking extremely difficult. 2. For large steel pipes, when using robots in conjunction with gripper manipulator systems for destacking, if the steel pipes are tilted or the stacks are not neatly arranged, accurate individual pipe gripping cannot be ensured, and damage or pressure may occur to the steel pipe surface, affecting its mechanical properties.

[0003] Chinese Patent No. CN107336069A discloses a feeding device and method for a sawing machine. The method involves tilting a rack holding stacked steel pipes, using a lifting mechanism on the rack to lift the pipes to a storage platform, and finally using a flipping mechanism to flip the pipes from the storage platform into a conveyor platform, thus completing the destacking process. This solution improves the speed and efficiency of destacking and feeding, but it suffers from problems such as excessive pressure on the mechanical structure due to the weight of the steel pipes, leading to damage, and excessive noise at the production site. Chinese Patent No. CN201952015U discloses an automatic steel pipe handling and palletizing machine that uses an electromagnet adsorption mechanism and a pallet gripping structure to separate stacked steel pipes. However, this solution cannot achieve accurate steel pipe gripping when dealing with non-orderly arranged stacks of steel pipes.

[0004] Therefore, it is essential to provide an automated destacking method that combines machine vision and robotics. This method uses vision technology to obtain the position of steel pipes and then performs precise clamping and destacking operations to improve destacking efficiency and automation.

[0005] Based on the above analysis, the urgent technical problems that need to be solved in the existing technology are:

[0006] 1. In the steel pipe processing industry, some steel pipes are over 10 meters long and weigh over 200 kilograms, making it extremely difficult to dismantle them manually.

[0007] 2. When using robots in conjunction with clamping manipulator systems to destacking large steel pipes, if the steel pipes are tilted or the stacks are not neatly arranged, it is impossible to ensure accurate single-pipe clamping operations, and damage or pressure may be caused to the surface of the steel pipes, affecting their mechanical properties. Summary of the Invention

[0008] To address the problems in the existing technology, this invention proposes an automated destacking method and system for steel pipes based on line laser projection.

[0009] This invention is achieved through the following method: an automated destacking method for steel pipes based on line laser projection and machine vision, comprising:

[0010] S1. A parallel line laser is projected at a specific angle onto the surface of the steel pipe stack to be destabilized. The intersection of the laser with the surface of the top layer of steel pipe generates a series of laser reflection curves that vary with the contour of the steel pipe surface. An industrial camera is used to acquire the image of the laser line curve.

[0011] S2, perform image preprocessing steps on the obtained laser reflection curve image, including converting it to a grayscale image and performing binarization to segment the laser curve. Then, use the labeling tool anyLabeling to label the laser reflection curve, thereby constructing a training dataset;

[0012] S3. Use the YOLOv8l model to train features on the dataset to ensure that the laser reflection curve of the steel pipe surface in the image can be correctly segmented in order to accurately obtain the target steel pipe area in the image.

[0013] S4, driven by the conveyor belt, the main and secondary cameras move parallel to the first measurement position and project a laser line onto the steel pipe stack. The main and secondary cameras acquire images one after another, and use the trained YOLOv8l model to infer the images to determine the position of the curved laser line and its minimum bounding rectangle.

[0014] S5, the main and secondary cameras move to a preset fixed distance to the second measurement position, and then successively acquire the laser reflection curve image of the steel pipe stack surface. The trained YOLOv8l model is used to infer the image to determine the position of the laser reflection curve and its minimum bounding rectangle.

[0015] S6, match the laser reflection curve position and its minimum bounding rectangle in the images acquired by the main and secondary cameras at the two measurement positions, and determine the minimum bounding rectangle with the smallest pixel Y coordinate in each acquired image, which is the target minimum bounding rectangle;

[0016] S7 records the pixel coordinates of the curve vertex enclosed by the smallest bounding rectangle of the target in the image acquired by the main and secondary cameras at two positions, calculates the distance difference in the X coordinate, and determines the conversion relationship between the pixel coordinate distance and the actual motion distance by combining the preset fixed motion distance.

[0017] S8, based on the conversion relationship between pixel coordinate distance and actual movement distance, obtains the actual movement distance from the current position of the main and secondary cameras to the target steel pipe through the vertex coordinates of the target laser reflection curve at the second measurement position, in order to carry out subsequent destacking operations.

[0018] Furthermore, S6 specifically includes:

[0019] Step S61: Obtain feature information derived by the YOLOv8l model after the main and secondary cameras collect data at two measurement positions. The feature information includes the area of ​​the minimum bounding rectangle and the pixel coordinates of the curve vertex.

[0020] Step S62: Using the image acquired by the main and secondary cameras at the first measurement position as a reference, compare it with the image acquired at the second position to determine whether the area ratio of each group of laser reflection curves exceeds the preset area ratio.

[0021] Step S63: Obtain the X-coordinate translation pixel distance of the images acquired by the main and secondary cameras at the two measurement positions. Translate the minimum bounding rectangle of the laser reflection curve captured by the main and secondary cameras at the second measurement position into the image at the first measurement position. Perform positioning and matching of the vertices of the laser reflection curve. Then repeat step S62 to perform area matching.

[0022] Furthermore, S61 specifically includes: calculating the area S of the minimum bounding rectangle derived from the YOLOv8l model for each image. r The set of areas of the smallest bounding rectangles that form the laser reflection curve, S. i Where i is the minimum bounding rectangle region of the i-th laser reflection curve, and the minimum bounding rectangle areas of the laser reflection curves are combined to form a list S. r ={S0,S1,…,S n}, where n is the total number of minimum bounding rectangles of the laser reflection curve on the image.

[0023] Furthermore, S62 specifically includes: the comparison process starts from the image coordinate origin. Based on the X-coordinate of the upper left vertex of the smallest bounding rectangle of the laser reflection curve, the first smallest bounding rectangle is selected from the images acquired by the main camera at the two measurement positions. Then, its area is compared sequentially with the rectangles in the images acquired by the corresponding secondary cameras at the measurement positions. The comparison process is performed one-to-one, that is, each time a rectangle from the main camera is compared with a rectangle from the secondary camera. The formula is as follows: Among them, S i,m S refers to the area of ​​the smallest bounding rectangle of the laser reflection curve from the main camera. i,s This refers to the area of ​​the minimum bounding rectangle of the laser reflection curve from the secondary camera;

[0024] By setting the area ratio R m To control the area ratio of the minimum bounding rectangle of each group of laser reflection curves, when the preset area ratio is appropriate, the small laser reflection curve area formed by the laser projection on the lower steel pipe or on the material frame can be deleted, allowing the error caused by local changes in the laser reflection curve area;

[0025] When R i <R m When comparing the minimum bounding rectangle of the set of laser reflection curves, the area difference is too large and they do not correspond to the same steel pipe. Therefore, the minimum bounding rectangle with the smaller area is deleted. After deletion, the larger rectangle will continue to be compared with the next rectangle in another camera.

[0026] When R i >R m At that time, the smallest bounding rectangle of the laser reflection curves being compared continues to be compared in a loop because the areas are similar, until all the rectangular information acquired by the main camera or the secondary camera at a measurement position is compared.

[0027] Furthermore, S63 specifically includes: sorting the minimum bounding rectangles of the laser reflection curves acquired by the main and secondary cameras at the two measurement positions according to the X coordinate of their upper left corner vertex in ascending order, and selecting two rectangles sequentially from the sorted rectangle list to form a rectangle pair; for the rectangle features of each pair of rectangles extracted by the main and secondary cameras at the first and second measurement positions, calculating the following parameters:

[0028] 1. Spacing D x The horizontal distance between the center points of two rectangles can be expressed by the following formula: D x =|X 2c -X 1c |, where X 1c and X 2c These are the X coordinates of the center points of the first and second rectangles, respectively.

[0029] 2. Areas A1 and A2: Calculate the area of ​​each rectangle separately using the formulas: A1 = W1 × H1, A2 = W2 × H2, where W1 and H1 are the width and height of the first rectangle, and W2 and H2 are the width and height of the second rectangle.

[0030] 3. Y-direction difference D y The formula for calculating the relative height difference between two rectangles in the vertical direction is: D y =|Y 2c -Y 1c |, where Y 1c and Y 2c These are the Y coordinates of the center points of the first and second rectangles, respectively.

[0031] The following ratios are defined using the feature parameters obtained by the main and secondary cameras at the first and second measurement positions, and a total product value is further calculated to evaluate the similarity of the features.

[0032] 1. Spacing ratio The ratio of the rectangular pair spacing at the second measurement position to the rectangular pair spacing at the first measurement position is calculated using the following formula: Among them, D x,M1 It is the rectangular pair spacing D obtained by the camera at the first measurement position. x,M2 The ratio is calculated by dividing the maximum value by the minimum value to ensure that the ratio is always greater than or equal to 1, based on the rectangular pair spacing obtained by the camera at the second measurement position.

[0033] 2. Area ratio R A1 ,R A2 The formulas are as follows: The first ratio is the area of ​​the first rectangle in the middle of the two rectangles, and the second ratio is the area of ​​the second rectangle in the middle of the two rectangles. Among them, A 1,M1 Let A be the area of ​​the first rectangle centered at the first measurement position. 1,M2 Let A be the area of ​​the first rectangle centered at the second measurement position. 2,M1 Let A be the area of ​​the second rectangle centered at the first measurement position rectangle. 2,M2 The area of ​​the second rectangle in the second measurement position rectangle is calculated using the method of dividing the maximum value by the minimum value to ensure that the ratio is always greater than or equal to 1.

[0034] 3. Y-direction difference ratio R Dy The formula for calculating the ratio of the difference in the Y-direction between the center points of two rectangles is: Where D y,M1 D represents the difference in the Y-direction between the rectangles at the first measurement position. y,M2 The second measurement and control position rectangle represents the difference in the Y direction. When calculating this ratio, the method of dividing the maximum value by the minimum value was used to ensure that the ratio is always greater than or equal to 1.

[0035] Based on the above ratios, a total product value P is calculated. This value combines all the ratios through a product method to evaluate the similarity of the features of the rectangle pairs. The formula is: P = R Dx ×R A1 ×R A2 ×R Dy The smaller the total product value, the more similar the rectangular features of the selected first measurement location and the rectangle pair of the second measurement are;

[0036] Select the pair of rectangles with the smallest P-value, and use these four rectangles as a reference to determine the horizontal pixel offset D of the images acquired by the main and secondary cameras at the first and second measurement positions. x The calculation formula is: Where X 1,M1 Let X be the center x-coordinate of the first rectangle in the first measurement position rectangle pair. 1,M2Let X be the center x-coordinate of the first rectangle in the second measurement position rectangle pair. 2,M1 Let X be the X-coordinate of the center of the second rectangle in the current rectangle pair at the first measurement position. 2,M2 The X-coordinate of the center of the second rectangle in the current rectangle pair at the second measurement position;

[0037] Get the horizontal pixel offset value D x Then, the coordinates X of the vertex of each laser reflection curve obtained from the second measurement position of the main and secondary cameras are calculated. i,M2 The coordinates of the curve vertex X′ after translation i,M2 The formula is X′ i,M2 =X i,M2 +D x After completing the offset calculation, check the coordinates X′ of the curve vertex after the second measurement position has been translated. i,M2 Is it the rectangle R corresponding to its first measurement position? i,M1 Match successful. The conditions for a successful match are: in It is a rectangle R i,M1 The top-left X-coordinate For rectangle R i,M1 The X-coordinate of the bottom right corner. If a match fails, then the X-coordinate of the vertex is considered to be... i,M2 The smallest bounding rectangle R in the image of the second measurement location i,M2 If a match is found, it is removed from the list of rectangles at the second measurement location. If a match is found, step S62 is repeated for area matching to determine if the minimum bounding rectangle of the set of laser reflection curves corresponds to the same pipe in the image.

[0038] Through the above matching process, the number of minimum bounding rectangles of the laser reflection curves obtained by the main and secondary cameras at the first and second measurement positions will remain consistent, and they will be sorted in ascending order of X-coordinate. By matching the index, it can be ensured that the rectangles obtained by the main and secondary cameras at the two measurement positions are matched to the same pipe. Accordingly, based on the list of minimum bounding rectangles of the laser reflection curves recorded by the main camera at the first measurement position, the rectangle with the smallest Y-coordinate at the top left corner is selected, and the corresponding minimum bounding rectangle of the target obtained by the main and secondary cameras at the two measurement positions is determined according to its index. This is the topmost steel pipe in the current stack.

[0039] Furthermore, the S7 specifically includes:

[0040] By recording the difference in the vertex X pixel coordinates of the laser reflection curve enclosed by the minimum bounding rectangle of the target in the images acquired by the main and secondary cameras at two measurement positions, and calculating the conversion relationship between the pixel coordinates of the images acquired by the two cameras and the real-world coordinates, the scaling factor K is obtained. m K sThe calculation formula is as follows: Where D real The preset horizontal movement distance ΔX between the two measurement positions. m The difference in vertex X pixel coordinates, ΔX, is obtained from the laser reflection curve bounded by the smallest bounding rectangle of the target at two measurement positions of the main camera. s The difference in X-pixel coordinates of the vertex of the laser reflection curve enclosed by the smallest bounding rectangle of the target is obtained from the two measurement positions of the secondary camera.

[0041] Furthermore, S8 specifically includes:

[0042] Based on the pixel deviation between the vertex X coordinate of the target laser reflection curve obtained by the main and secondary cameras at the second measurement point and the center of the fixed image width, and the scaling factor determined in step S7, combined with the actual distance D between the electromagnetic gripper and the camera center, c The precise movement distance from the second measurement position to the target steel pipe can be calculated. The formula for calculating the actual movement distance depends on the installation position of the electromagnetic gripper and the camera's movement direction. The direction is considered positive if the X-coordinate of the laser reflection curve's vertex exceeds the center of the image width, and negative if it falls below.

[0043] 1. The electromagnetic gripper is installed in the positive direction of the camera's movement:

[0044] When the camera needs to move in the positive direction, the calculation formula is: When the camera needs to move in the negative direction, the calculation formula is:

[0045] 2. The electromagnetic gripper is installed in the negative direction of the camera's movement:

[0046] When the camera needs to move in the positive direction, the calculation formula is: When the camera needs to move in the negative direction, the calculation formula is:

[0047] Where D move Finally, the actual distance required for the destacking operation is given by W, which is a fixed image width, and X. M2 To obtain the X-coordinate of the vertex of the laser reflection curve enclosed by the smallest bounding rectangle of the target at the second measurement position, K is the scaling factor for converting image pixel coordinates and real-world coordinates, and D is the coordinate of the laser reflection curve. c This is the horizontal distance between the electromagnetic gripper and the center of the camera.

[0048] Another objective of this invention is to provide an automated steel pipe destacking system based on line laser projection and machine vision for implementing the aforementioned automated steel pipe destacking method based on line laser projection and machine vision, comprising:

[0049] The curved laser line image acquisition module is used to project a parallel line laser at a specific angle onto the surface of the steel pipe stack to be destabilized. The intersection of the laser line with the surface of the topmost steel pipe generates a series of laser reflection curves that vary with the contour of the steel pipe surface. The image of the curved laser line is then acquired using an industrial camera.

[0050] The image preprocessing module is used to perform image preprocessing steps on the obtained laser reflection curve image, including converting it to a grayscale image and performing binarization to segment the laser curve; subsequently, the laser reflection curve is labeled using the labeling tool anyLabeling, thereby constructing a training dataset;

[0051] The feature training module is used to train the dataset with the YOLOv8l model to ensure that the laser reflection curve of the steel pipe surface in the image can be correctly segmented in order to accurately obtain the target steel pipe area in the image.

[0052] The image reasoning module involves the main and secondary cameras moving parallel to each other on the conveyor belt and projecting a laser line onto the steel pipe stack. The main and secondary cameras acquire images sequentially, and the trained YOLOv8l model is used to reason about the images to determine the position of the curved laser line and its minimum bounding rectangle.

[0053] The second measurement position module moves the main and secondary cameras to a preset fixed distance to the second measurement position, and then successively acquires the laser reflection curve image of the steel pipe stack surface. The trained YOLOv8l model is used to infer the image to determine the position of the laser reflection curve and its minimum bounding rectangle.

[0054] The image matching module is used to match the position of the laser reflection curve and its minimum bounding rectangle in the images acquired by the main and secondary cameras at two measurement positions, and to determine the minimum bounding rectangle with the smallest pixel Y coordinate in each acquired image, which is the target minimum bounding rectangle.

[0055] The conversion relationship determination module is used to record the pixel coordinates of the curve vertex enclosed by the smallest bounding rectangle of the target in the images acquired by the main and secondary cameras at two positions, calculate the X coordinate distance difference, and combine it with the preset fixed motion distance to determine the conversion relationship between the pixel coordinate distance and the actual motion distance.

[0056] The real motion distance acquisition module obtains the real motion distance from the current position of the main and secondary cameras to the target steel pipe by using the vertex coordinates of the target laser reflection curve at the second measurement position, based on the conversion relationship between pixel coordinate distance and actual motion distance, in order to carry out subsequent destacking operations.

[0057] Another object of the present invention is to provide a computer device, the computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the described automated destacking method for steel pipes based on line laser projection and machine vision.

[0058] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the described automated destacking method for steel pipes based on line laser projection and machine vision.

[0059] Another objective of this invention is to provide an information data processing terminal, which includes the aforementioned automated steel pipe destacking system based on line laser projection and machine vision.

[0060] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0061] First, the automated destacking method and system for steel pipes based on line laser projection and machine vision provided by this invention has the following advantages over the prior art:

[0062] 1. To address the issue of complex environments in steel pipe production sites affecting the normal operation of vision systems, this paper addresses the problem of converting linear laser reflection images of steel pipe stacks to grayscale, binarizing them, and then deriving a YOLOv8l model to eliminate noise interference from the harsh industrial environment, thereby achieving accurate segmentation of the curved light rays on the steel pipe surface. This process obtains the minimum bounding rectangle of the steel pipe image and its positional information, facilitating accurate identification of the steel pipe's position coordinates and improving the accuracy of subsequent destacking operations.

[0063] 2. Addressing the problems of excessive mechanical structural pressure, excessive noise, and safety hazards caused by large steel pipes in existing research, this invention configures primary and secondary vision systems and line laser emitters on both sides of the steel pipe frame, along with two sets of electromagnetic grippers. Servo motors drive the conveyor belts at both ends, enabling the horizontal movement of both systems. Through the collaborative work of the line laser projection and the vision system, the electromagnetic grippers can accurately grasp large steel pipes that are tilted or irregularly arranged in the stack. This method significantly reduces the cost of the entire destacking system and ensures the accuracy of the destacking operation.

[0064] 3. To address the calibration problem of image pixel coordinates and real-world coordinates in a vision system, this invention establishes a first measurement point and a second measurement point, enabling the primary and secondary cameras to perform two separate image acquisitions and image processing operations. In the image processed by the YOLOv8l model, the X-coordinate of the vertex pixel enclosed by the bounding rectangle with the smallest Y-coordinate is recorded. By calculating the difference in X-coordinates obtained from the two acquisitions and combining it with a preset fixed motion distance, the mapping relationship between the image pixel coordinates of the primary and secondary cameras and the real-world coordinates can be determined. This method achieves camera calibration by setting up image acquisition at two measurement positions for the monocular camera, significantly reducing hardware investment costs.

[0065] Secondly, as supplementary evidence of the inventive step of the claims of this invention, it is also reflected in the following important aspects:

[0066] (1) The expected benefits and commercial value of the technical solution of this invention after transformation are as follows:

[0067] First, it significantly improves production efficiency by automating steel pipe destacking operations, shortening production cycles and increasing production line throughput. Second, in the long term, automation systems can substantially reduce operating costs, decrease manpower requirements, and reduce material waste. Furthermore, the system's precise control ensures consistent product handling and quality, thereby improving product quality. The application of this technology further enhances a company's market competitiveness, enabling it to provide faster and more reliable service. Simultaneously, the technology's applicability is not limited to steel pipe destacking but can be extended to other industrial automation scenarios, increasing market potential. The introduction of automation systems also helps improve workplace safety and reduce accident risks in high-risk operating environments. In addition, reducing waste and improving material utilization efficiency contributes to environmental sustainability, aligning with environmental trends. Finally, this invention provides solid technical support for companies' transformation towards intelligent manufacturing. Overall, this invention not only optimizes production processes but also provides companies with a sustainable competitive advantage and economic benefits.

[0068] (2) The technical solution of this invention fills a technical gap in the industry both domestically and internationally:

[0069] The technical solution of this invention, through its innovative laser projection and computer vision technologies, effectively fills the gap in the field of efficient and precise automated steel pipe destacking technology both domestically and internationally. Traditional steel pipe processing methods rely on manual operation or simple mechanical devices, resulting in problems such as low efficiency, low accuracy, and safety risks. This invention, however, achieves rapid and accurate identification of the steel pipe's position and precise calculation of the actual movement distance by accurately measuring the laser reflection curve on the surface of the steel pipe stack, thereby significantly improving the performance of automated destacking.

[0070] The implementation of this technology not only improves production efficiency and safety but also offers high flexibility and adaptability, enabling stable operation in diverse working environments. Furthermore, it provides a novel automation solution for related industries, contributing to technological advancements and the transformation of production methods across the sector. Therefore, this invention is not only a technological innovation but also possesses significant practical value in the fields of industrial automation and intelligent manufacturing.

[0071] (3) The technical solution of the present invention solves a technical problem that people have long wanted to solve but have never been able to solve successfully:

[0072] The technical solution of this invention successfully solves a long-standing but unresolved technical problem: how to accurately, efficiently, and safely automate the processing of large quantities of stacked steel pipes. In the past, despite continuous efforts to find solutions, this problem remained unresolved due to technological limitations, particularly challenges in precision, efficiency, and safety.

[0073] This invention utilizes laser projection and computer vision technology to quickly and accurately identify and measure the position and actual movement distance of steel pipes. The implementation of this technology not only improves the efficiency of steel pipe processing but also significantly reduces safety risks for workers during the process due to its automation. Furthermore, it offers high flexibility and adaptability, enabling stable operation in various working environments. Therefore, this invention successfully solves a long-standing technical problem, providing a new and efficient automated solution for steel pipe processing and other related industries.

[0074] (4) The technical solution of the present invention overcomes technical bias:

[0075] The technical solution of this invention successfully overcomes the long-standing technical biases in the field of automated steel pipe processing, which mainly include:

[0076] 1. Incompatibility of Automation with Complex Environments: For a long time, the industry has generally considered implementing high-precision automation technology in complex and ever-changing industrial environments to be extremely challenging. Especially in scenarios like steel pipe stacks, where physical forms are diverse and arrangements are disordered, traditional views hold that achieving efficient and accurate identification and handling through automation is almost impossible. This invention, through advanced laser projection and computer vision algorithms, demonstrates that even in such complex environments, high-efficiency and high-precision automated processing can be achieved.

[0077] 2. Expectations of High Costs and Low Returns: Before automation technology was widely applied to steel pipe processing, a common technological bias was the belief that applying high technology to such traditional industries was too costly and the return on investment was not significant. This invention demonstrates that intelligent transformation can not only significantly improve production efficiency and safety, but also achieve significant economic benefits by reducing reliance on manual labor and improving product quality and consistency, thereby breaking this bias.

[0078] 3. The Conflict Between Safety and Automation: Another technological bias is the belief that automated systems increase workplace safety risks, especially in scenarios involving the handling of heavy objects such as steel pipes. However, this invention, through precise laser positioning and computer vision technology, not only improves operational accuracy but also significantly reduces safety risks by minimizing direct human intervention in heavy object handling, demonstrating that automation technology can coexist with improved workplace safety.

[0079] This invention not only promotes technological progress in the field of steel pipe processing through its innovative technical solutions, but also challenges and overcomes several long-standing technical biases in the field, providing important technical support and practical cases for the automation and intelligent transformation of related industries.

[0080] Third, in existing steel pipe destacking operations, traditional methods mainly rely on manual operation or simple mechanical equipment, which have significant shortcomings in terms of accuracy, efficiency, and safety. Manual operation is not only labor-intensive, but also often leads to low efficiency and damage to steel pipes during the destacking process due to human factors. Simple mechanical equipment is difficult to achieve high-precision steel pipe identification and positioning, and cannot meet the needs of automated destacking. In addition, existing technologies lack precise identification and positioning methods for the surface contours of steel pipes, making it difficult to accurately separate steel pipes in complex stacking environments, which leads to easy errors during the destacking process and reduces destacking efficiency and accuracy.

[0081] This invention proposes an automated destacking method for steel pipes based on line laser projection and machine vision. By introducing line laser projection and advanced machine vision technology, it achieves high-precision identification and positioning of the steel pipe surface contour, significantly improving the efficiency and accuracy of destacking operations. This method employs parallel line laser projection technology to form a series of laser reflection curves on the steel pipe surface. Laser line images are acquired using an industrial camera, and then processed using a YOLOv8l model to accurately segment the laser reflection curves on the steel pipe surface and determine the position of the target steel pipe. Compared to traditional methods, this invention can accurately identify steel pipes even in complex stacking environments, avoiding misoperation.

[0082] 1. Highly efficient and accurate steel pipe identification and positioning:

[0083] This invention utilizes laser projection and machine vision technology to accurately identify and locate the contours of steel pipe surfaces, avoiding errors caused by manual operation or insufficient precision of mechanical equipment in traditional methods. By segmenting and locating the laser reflection curve using the YOLOv8l model, the identification and positioning of steel pipes becomes more precise, improving the efficiency of destacking operations.

[0084] 2. Efficient data processing and analysis:

[0085] This invention achieves efficient data processing and analysis through image preprocessing, labeling, and feature training. Laser reflection curve images are acquired using an industrial camera, and these images are then processed by grayscale conversion, binarization, and labeling to construct a training dataset. This ensures that the YOLOv8l model can accurately segment and identify laser reflection curves, providing reliable data support for subsequent steel pipe positioning and destacking.

[0086] 3. High degree of automation:

[0087] The destacking method of this invention achieves a high degree of automation. Driven by the conveyor belt, the main and auxiliary cameras automatically acquire and process images at two measurement positions. By calculating the conversion relationship between pixel coordinate distance and actual movement distance, the true movement distance of the target steel pipe is accurately determined. The entire process requires no manual intervention, thus improving the automation level of the destacking operation.

[0088] 4. Safety and environmental friendliness:

[0089] This invention employs non-contact line laser projection and machine vision technology, avoiding damage to steel pipes caused by traditional mechanical equipment during destacking and protecting the integrity of the steel pipe surface. Furthermore, this invention does not generate harmful gases or substances during data processing and image analysis, meeting environmental protection requirements and contributing to green production.

[0090] Fourth, this invention primarily addresses the problems of low recognition accuracy, complex operation, and insufficient safety in the automated destacking process of steel pipes in existing technologies by introducing line laser projection, machine vision technology, and advanced image processing algorithms. Traditional methods rely on manual operation or simple mechanical equipment, which cannot achieve high-precision steel pipe identification and positioning in complex stacking environments, resulting in low destacking efficiency, frequent misoperations, and equipment wear and tear. This invention, through scientific parameter settings and algorithm optimization, achieves accurate identification and positioning of the steel pipe surface, greatly improving the automation level and safety of the destacking operation.

[0091] First, this invention utilizes line laser projection technology to project parallel line lasers at a specific angle onto the surface of the steel pipe stack to be destabilized, and then uses an industrial camera to acquire an image of the laser reflection curve on the steel pipe surface. This method ensures clear and accurate laser reflection curves even in complex lighting environments, overcoming the difficulty of traditional visual recognition methods in recognizing changes in light and surface reflection. By optimizing the laser projection angle and camera position, a clear and stable reflection curve is ensured on the steel pipe surface by the laser line, providing a reliable data foundation for subsequent image processing and feature extraction.

[0092] Secondly, this invention employs the YOLOv8l model for image processing and feature extraction. A training dataset is constructed by converting, binarizing, and labeling the laser reflection curve images. The YOLOv8l model is then used to train features on this dataset, ensuring the model can accurately segment and identify the laser reflection curves on the steel pipe surface, further improving the accuracy and stability of the identification. This model boasts high detection speed and accuracy, enabling real-time processing and analysis of acquired image data, significantly improving the efficiency of destacking operations.

[0093] During the identification and positioning process, this invention accurately determines the position of the target steel pipe through multiple measurements and comparisons. By calculating the area of ​​the minimum bounding rectangle of the laser reflection curve and the pixel coordinates in images acquired by the main and secondary cameras at different locations, the actual movement distance of the steel pipe is determined. Using mathematical models and algorithms, the pixel coordinate distance is converted to the actual movement distance, ensuring accurate positioning of the target steel pipe even in complex stacking environments. This approach avoids errors and damage to the steel pipe caused by inaccurate identification in traditional methods, improving the safety and reliability of the destacking operation.

[0094] Ultimately, this invention represents a significant technological advancement in addressing the problems of low efficiency and poor safety in automated steel pipe destacking in existing technologies. By combining line laser projection and machine vision technologies, and optimizing laser projection parameters and image processing algorithms, high-precision identification and positioning of the steel pipe surface are achieved, significantly improving the automation level and efficiency of the destacking operation. Furthermore, this invention achieves non-contact identification and positioning while avoiding damage to the steel pipe surface, protecting the integrity of the steel pipe, meeting environmental protection requirements, and promoting the realization of green production. Attached Figure Description

[0095] Figure 1 This is a flowchart of an automated steel pipe destacking method based on line laser projection provided in an embodiment of the present invention.

[0096] Figure 2 This is an example diagram of obtaining laser reflection curves provided in an embodiment of the present invention.

[0097] Figure 3This is a flowchart of object detection inference using YOLOv8l provided in an embodiment of the present invention.

[0098] Figure 4 This is an example diagram showing the results of inference using the YOLOv8l model, as provided in this embodiment of the invention.

[0099] Figure 5 This is an example image of the main and secondary cameras after image feature matching at two measurement positions, provided in an embodiment of the present invention.

[0100] Figure 6 This is a structural diagram of an electronic device provided in an embodiment of the present invention.

[0101] Figure 7 This is a structural diagram of an automated steel pipe destacking system based on line laser projection provided in an embodiment of the present invention.

[0102] Figure 8 This is a schematic diagram of the host computer software for the automated steel pipe destacking system based on line laser projection provided in an embodiment of the present invention.

[0103] Figure 9 This is a schematic diagram of a destacking operation performed at the work site according to an embodiment of the present invention.

[0104] In the diagram: 1. Electronic device; 11. Processor; 12. Communication bus; 13. User interface; 14. Network interface; 15. Memory. Detailed Implementation

[0105] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0106] Before introducing the embodiments of the present invention, some terms and their abbreviations involved in the embodiments of the present invention will be defined and explained.

[0107] The OTSU algorithm (Otsu's Method) is a classic image segmentation algorithm used to automatically determine the optimal threshold for dividing an image into two categories: foreground and background. The key idea of ​​this algorithm is to find the optimal threshold by maximizing the inter-class variance based on the image's gray-level histogram, thus achieving effective image segmentation. The advantages of the OTSU algorithm lie in its simplicity and effectiveness, automatically determining the threshold without requiring manual setting by the user. Therefore, it is widely used in various tasks in the field of image processing, such as object detection, edge detection, and document image processing.

[0108] The YOLOv8l model is a variant of the eighth generation of the YOLO (You Only Look Once) series. Building on the success of previous YOLO versions, YOLOv8l features architectural optimizations, employing a new backbone network, a new Ancher-Free detection head, and a new loss function. This model combines CNN (Convolutional Neural Network) and Transformer structures to improve feature extraction capabilities and overall model performance. The "l" stands for "large," indicating that YOLOv8l is a larger model in the YOLOv8 series, with more parameters and higher computational complexity, aiming to achieve higher detection accuracy. Despite its higher computational resource requirements, it maintains the fast inference capabilities of the YOLO series. YOLOv8l can simultaneously perform multi-task learning, including object detection, image segmentation, and keypoint detection. This multi-task learning capability makes it perform exceptionally well in handling complex scenes.

[0109] The Minimum Bounding Rectangle (MBR) is a technique commonly used in computational geometry and image processing. It is used to delineate the smallest bounding region of an object or set of points to be processed, in the form of a rectangle. In machine vision software such as Halcon, OpenCV, and Matlab, various operators and functions are frequently used to calculate the minimum bounding rectangle and perform further image processing. In image processing, the minimum bounding rectangle (MBR) is the smallest rectangle that completely encloses a given set of points or geometry. The sides of this rectangle are parallel to the coordinate axes. Calculating the minimum bounding rectangle involves finding the minimum and maximum x and y coordinates of the point set and using these coordinates to construct the rectangle's boundary. The minimum bounding rectangle (MBR) is an important and practical technique widely used in various fields requiring rapid computation and processing of geometric shapes.

[0110] This application discloses an automated destacking method for steel pipes based on line laser projection and machine vision, such as... Figure 1 As shown, the method includes steps S1-S8.

[0111] Step S1: A parallel line laser is projected at a specific angle onto the surface of the steel pipe stack to be destabilized. The intersection of the laser line with the surface of the uppermost steel pipe generates a series of laser reflection curves that vary with the contour of the steel pipe surface. An industrial camera is used to acquire an image of the laser reflection curve.

[0112] Specifically, by fixing an industrial camera at both ends of the material frame and adjusting the camera's focal length and aperture, an image of the curved light rays projected by a line laser onto the top layer of the steel pipe stack is obtained. Before acquiring the image, ensure the camera is fixed and its parameters are properly adjusted, with the exposure time of the industrial camera set to 5000ms and the acquired image resolution at 5472 pixels × 3000 pixels. The acquired image signal is saved as a BMP format image data file, numbered incrementally according to the shooting time. Figure 2 As shown, this image displays a curved light pattern formed by a line laser projected onto the top layer of a stack of steel pipes.

[0113] Step S2 involves performing image preprocessing on the obtained laser reflection curve image, including converting it to a grayscale image and performing binarization to segment the laser curve. Subsequently, the laser reflection curve is labeled using the labeling tool anyLabeling to construct a training dataset.

[0114] Step S3: Use the YOLOv8l model to train the features of the dataset to ensure that the laser reflection curve of the steel pipe surface in the image can be correctly segmented in order to accurately obtain the target steel pipe area in the image.

[0115] Preferably, in this embodiment, the RGB format three-channel color image acquired by the industrial camera is converted into a single-channel grayscale image to reduce computational complexity and data volume, while retaining the main structural information of the image. The grayscale conversion formula is Y = 0.299 × R + 0.587 × G + 0.144 × B, where Y is the grayscale value, and R, G, and B are the pixel values ​​of the red, green, and blue channels, respectively. Furthermore, to segment continuous laser reflection curves and reduce the limitations of manually adjusting the threshold, thereby improving the overall segmentation efficiency of the image processing algorithm, this embodiment uses the OTSU algorithm for adaptive recognition and segmentation of the laser reflection curves in the image. Further, the anyLabeling tool is used to annotate the laser reflection curve image after threshold segmentation to record the target location information and category in the image. The annotated data is saved in a JSON format text file. During the annotation process, the polygonal bounding boxes used should closely fit the target edges for accurate annotation. Furthermore, the labeled images were divided into training, validation, and test datasets in a 6:2:2 ratio, and the YOLOv8l model was used to train features on these datasets.

[0116] In step S4, the main and secondary cameras, driven by the conveyor belt, move parallel to the first measurement position and project a line laser onto the steel pipe stack. The main and secondary cameras acquire images one after the other, and use the trained YOLOv8l model to infer the image to determine the position of the curved laser line and its minimum bounding rectangle.

[0117] Step S5: The main and secondary cameras move to a preset fixed distance to the second measurement position, and then successively acquire the laser reflection curve image of the steel pipe stack surface. The trained YOLOv8l model is used to infer the image to determine the position of the laser reflection curve and its minimum bounding rectangle.

[0118] In this embodiment, since a monocular industrial camera is used for image acquisition, it is impossible to directly obtain the depth information of the steel pipe stack to determine the specific position of the top steel pipe. Therefore, this embodiment designs two measurement positions and performs two image acquisitions. By analyzing the image information acquired in these two acquisitions, the position of the topmost steel pipe in the stack is determined in the image at a two-dimensional level.

[0119] Specifically, in this embodiment, the image is subjected to object detection using the YOLOv8l model to obtain image information. The flowchart of the YOLOv8l model's object detection task is as follows: Figure 3 As shown, this model employs CSPDarknet (CrossStage Partial networks with Darknet), an improved backbone network that combines the advantages of cross-stage partial connectivity techniques, achieving efficient feature extraction while reducing computational costs. Compared to the Darknet backbone network in previous YOLO models, CSPDarknet demonstrates superior performance and efficiency. For the neck network, YOLOv8l adopts the PANet (Path Aggregation Network) design, which facilitates the effective aggregation of multi-scale features, thereby optimizing detection performance for targets of various sizes. Furthermore, YOLOv8l introduces the concept of a decoupled head, reducing interference between classification and localization tasks by processing them separately, thus improving the model's detection accuracy. YOLOv8l has undergone numerous architectural optimizations, making the model more lightweight and faster inference. Figure 4 As shown, the YOLOv8l model can quickly and accurately identify the position of the laser reflection curve and its minimum bounding rectangle in the image, thereby determining the exact position of the top layer of steel pipe in the current steel pipe stack in the image.

[0120] Step S6: Match the laser reflection curve position and its minimum bounding rectangle in the images acquired by the main and secondary cameras at the two measurement positions, and determine the minimum bounding rectangle with the smallest pixel Y coordinate in each acquired image, which is the target minimum bounding rectangle.

[0121] Specifically, when a line laser is projected onto the surface of a stack of steel pipes at a certain angle, due to the stacking characteristics, the reflected image formed by the laser line on the surface of the stack presents curves of light with different curvatures. Since the laser is at a certain angle to the camera, according to the principle of laser triangulation, when the laser shines on the object surface at a non-orthogonal angle, the laser reflection curve will form a displacement on the camera sensor. The magnitude of this displacement depends on the height of the object surface. In this case, the tallest steel pipe will cause the laser line to have the largest displacement on the camera sensor, and therefore its position on the image will be closest to the top edge of the image. In computer vision and image processing environments, the origin of the image is located at the top left corner. In this coordinate system, the X-coordinate increases from left to right, while the Y-coordinate increases from top to bottom. The smaller the Y-coordinate value, the closer the point is to the top of the image. Therefore, in the image, the taller steel pipe in the stack will cause the smallest bounding rectangle of the laser reflection curve to have a smaller Y-coordinate in the image.

[0122] In this embodiment, the relative height of the steel pipes in the stack can be determined by analyzing the reflection position of the laser reflection curve on the surface of the stack. However, in certain specific scenarios, when acquiring images of ultra-long steel pipe stacks, such as when the steel pipes are excessively deformed or the top layer of steel pipes in the stack is arranged very regularly, the target laser reflection curves acquired by the main and secondary cameras may correspond to different steel pipes, which will prevent the destabilization operation from being performed normally. Therefore, in this embodiment, it is necessary to accurately match the laser reflection curves and their minimum bounding rectangle position information acquired by the main and secondary camera modules at two measurement points to determine that the target laser reflection curves acquired by the main and secondary cameras are for the same steel pipe, thereby improving the accuracy and stability of the destabilization operation.

[0123] In this embodiment, the matching process in step S6 includes steps S61 to S63.

[0124] Step S61: Obtain feature information derived from the YOLOv8l model after the main and secondary cameras have acquired data at two measurement positions. The feature information includes the area of ​​the minimum bounding rectangle and the pixel coordinates of the curve vertex.

[0125] In this step, the area S of the minimum bounding rectangle derived from the YOLOv8l model is calculated for each image. r The set of areas of the smallest bounding rectangles that form the laser reflection curve, S. i Where i is the minimum bounding rectangle region of the i-th laser reflection curve, and the minimum bounding rectangle areas of the laser reflection curves are combined to form a list S. r ={S0,S1,…,S n}, where n is the total number of minimum bounding rectangles of the laser reflection curve on the image.

[0126] Step S62: Using the image acquired by the main and secondary cameras at the first measurement position as a reference, compare it with the image acquired at the second position to determine whether the area ratio of each group of laser reflection curves exceeds the preset area ratio.

[0127] In this step, the comparison process begins at the image coordinate origin. Based on the X-coordinate of the top-left vertex of the smallest bounding rectangle of the laser reflection curve, the first smallest bounding rectangle is selected from the images acquired by the main camera at the two measurement positions. Then, its area is compared sequentially with rectangles in the images acquired by the secondary camera at the corresponding measurement positions. The comparison process is performed one-to-one, meaning that each time a rectangle from the main camera is compared with a rectangle from the secondary camera. The formula is as follows: Among them, S i,m S refers to the area of ​​the smallest bounding rectangle of the laser reflection curve from the main camera. i,s This refers to the area of ​​the smallest bounding rectangle of the laser reflection curve from the secondary camera.

[0128] By setting the area ratio R m This controls the area ratio of the minimum bounding rectangle of each group of laser reflection curves. When the preset area ratio is appropriate, the small laser reflection curve area formed by the laser projection on the lower steel pipe or on the material frame can be deleted, allowing the error caused by local changes in the laser reflection curve area.

[0129] When R i <R m When comparing the minimum bounding rectangles of the laser reflection curves, the smaller rectangle is deleted because the area difference is too large and it does not correspond to the same steel pipe. After deletion, the larger rectangle will be compared with the next rectangle in another camera.

[0130] When R i >R m At that time, the smallest bounding rectangle of the laser reflection curves being compared continues to be compared in a loop because the areas are similar, until all the rectangular information acquired by the main camera or the secondary camera at a measurement position is compared.

[0131] Step S63: Obtain the X-coordinate translation pixel distance of the images acquired by the main and secondary cameras at the two measurement positions. Translate the minimum bounding rectangle of the laser reflection curve captured by the main and secondary cameras at the second measurement position into the image at the first measurement position. Perform positioning and matching of the vertices of the laser reflection curve. Then repeat step S62 to perform area matching.

[0132] In this step, the minimum bounding rectangles of the laser reflection curves acquired by the primary and secondary cameras at the two measurement positions are sorted from smallest to largest according to the X-coordinate of their top-left vertex. Two rectangles are then selected sequentially from the sorted list to form a rectangle pair. For the rectangular features of each pair of rectangles extracted by the primary and secondary cameras at the first and second measurement positions, the following parameters are calculated:

[0133] 1. Spacing (D) x The horizontal distance between the center points of two rectangles can be calculated using the following formula: D x =|X 2c -X 1c |, where X 1c and X 2c These are the X coordinates of the center points of the first and second rectangles, respectively.

[0134] 2. Area (A1 and A2): Calculate the area of ​​each rectangle separately using the formulas: A1 = W1 × H1, A2 = W2 × H2, where W1 and H1 are the width and height of the first rectangle, and W2 and H2 are the width and height of the second rectangle.

[0135] 3. Y-direction difference (D) y To calculate the relative height difference between two rectangles in the vertical direction, the formula is: D y =|Y 2c -Y 1c |, where Y 1c and Y 2c These are the Y-coordinates of the center points of the first and second rectangles, respectively.

[0136] The following ratios are defined using the feature parameters obtained by the main and secondary cameras at the first and second measurement positions, and a total product value is further calculated to evaluate the similarity of the features.

[0137] 1. Spacing ratio The ratio of the rectangular pair spacing at the second measurement position to the rectangular pair spacing at the first measurement position is calculated using the following formula: Among them, D x,M1 It is the rectangular pair spacing D obtained by the camera at the first measurement position. x,M2 The ratio is calculated using the maximum value divided by the minimum value to ensure that the ratio is always greater than or equal to 1, based on the rectangular pair spacing obtained by the camera at the second measurement position.

[0138] 2. Area ratio (R) A1 ,R A2 ): These are two area ratios. The first ratio is the area of ​​the first rectangle in the middle of the two rectangles, and the second ratio is the area of ​​the second rectangle in the middle of the two rectangles. The formula is: Among them, A 1,M1 Let A be the area of ​​the first rectangle centered at the first measurement position. 1,M2 Let A be the area of ​​the first rectangle centered at the second measurement position. 2,M1 Let A be the area of ​​the second rectangle centered at the first measurement position rectangle. 2,M2 The area of ​​the second rectangle is calculated by dividing the maximum value by the minimum value to ensure that the ratio is always greater than or equal to 1.

[0139] 3. Y-direction difference ratio (R) Dy ): The ratio of the difference in the Y-direction between the center points of two rectangles is calculated using the following formula: Where D y,M1 D represents the difference in the Y-direction between the rectangles at the first measurement position. y,M2 The value is the difference between the rectangle at the second measurement and control position and the value in the Y direction. When calculating this ratio, the maximum value is divided by the minimum value to ensure that the ratio is always greater than or equal to 1.

[0140] By combining the above ratios, a total product value (P) is calculated. This value combines all the ratios through a product to evaluate the similarity of the rectangle pairs' features. The formula is: P = R Dx ×R A1 ×R A2 ×R Dy The smaller the total product value, the more similar the rectangular features of the selected first measurement location and the rectangle pair of the second measurement are.

[0141] Select the pair of rectangles with the smallest P-value, and use these four rectangles as a reference to determine the horizontal pixel offset (D) of the images acquired by the main and secondary cameras at the first and second measurement positions. x The calculation formula is: Where X 1,M1 Let X be the center x-coordinate of the first rectangle in the first measurement position rectangle pair. 1,M2 Let X be the center x-coordinate of the first rectangle in the second measurement position rectangle pair. 2,M1 Let X be the X-coordinate of the center of the second rectangle in the current rectangle pair at the first measurement position. 2,M2 The X-coordinate of the center of the second rectangle in the current rectangle pair at the second measurement position.

[0142] Get the horizontal pixel offset value (D) x After that, calculate the X coordinates of the vertex of each laser reflection curve obtained from the second measurement position of the main and secondary cameras. i,M2 The coordinates of the curve vertex X′ after translation i,M2 The formula is X′ i,M2 =X i,M2 +D xAfter completing the offset calculation, check the coordinates X′ of the curve vertex after the second measurement position has been translated. i,M2 Is it the rectangle R corresponding to its first measurement position? i,M1 Match successful. The conditions for a successful match are: in It is a rectangle R i,M1 The top-left X-coordinate For rectangle R i,M1 The X-coordinate of the bottom right corner. If a match fails, then the X-coordinate of the vertex is considered to be... i,M2 The smallest bounding rectangle R in the image of the second measurement location i,M2 If a match is found, it is removed from the list of rectangles at the second measurement location. If a match is found, step S62 is repeated for area matching to determine if the minimum bounding rectangle of the set of laser reflection curves corresponds to the same pipe in the image.

[0143] Through the matching process described above, the number of minimum bounding rectangles of the laser reflection curves obtained by the primary and secondary cameras at the first and second measurement positions will remain consistent, and they will be sorted in ascending order of their X-coordinates. By using index correspondence, it can be ensured that the rectangles obtained by the primary and secondary cameras at both measurement positions match the same pipe. Therefore, based on the list of minimum bounding rectangles of the laser reflection curve recorded by the primary camera at the first measurement position, the rectangle with the smallest Y-coordinate at its top-left corner is selected, and its index is used to determine the corresponding minimum bounding rectangle of the target obtained by the primary and secondary cameras at both measurement positions, which is the topmost steel pipe in the current stack.

[0144] Step S7: Record the pixel coordinates of the curve vertices enclosed by the smallest bounding rectangle of the target in the images obtained by the main and secondary cameras at the two positions, calculate the X-coordinate distance difference, and combine it with the preset fixed motion distance to determine the conversion relationship between pixel coordinate distance and actual motion distance.

[0145] In this step, the difference in the vertex X pixel coordinates of the laser reflection curve enclosed by the smallest bounding rectangle of the target in the images acquired by the main and secondary cameras at the two measurement positions is recorded. This difference is then used to calculate the conversion relationship between the pixel coordinates of the images acquired by the two cameras and the real-world coordinates. These two scaling factors (K...) are then used to determine the conversion. m K s The formula for calculating ) is: Where D real The preset horizontal movement distance ΔX between the two measurement positions. m The difference in vertex X pixel coordinates, ΔX, is obtained from the laser reflection curve bounded by the smallest bounding rectangle of the target at two measurement positions of the main camera. s The difference in X-pixel coordinates of the vertex of the laser reflection curve enclosed by the smallest bounding rectangle of the target is obtained from the two measurement positions of the secondary camera.

[0146] Step S8: Based on the conversion relationship between pixel coordinate distance and actual movement distance, the actual movement distance from the current position of the main and secondary cameras to the target steel pipe is obtained through the vertex coordinates of the target laser reflection curve at the second measurement position, so as to carry out subsequent destacking operations.

[0147] In this step, based on the pixel deviation between the X-coordinate of the target laser reflection curve vertex obtained by the main and secondary cameras at the second measurement point and the center of the fixed image width, and the scaling factor determined in step S7, combined with the actual distance (D) between the electromagnetic gripper and the camera center, the calculation is performed. c The precise distance of movement from the second measurement position to the target steel pipe can be calculated. The formula for calculating the actual distance of movement depends on the installation position of the electromagnetic gripper and the direction of camera movement. The direction is considered positive if the X-coordinate of the laser reflection curve vertex exceeds the center of the image width, and negative if it is below.

[0148] 1. The electromagnetic gripper is installed in the positive direction of the camera's movement:

[0149] When the camera needs to move in the positive direction, the calculation formula is: When the camera needs to move in the negative direction, the calculation formula is:

[0150] 2. The electromagnetic gripper is installed in the negative direction of the camera's movement:

[0151] When the camera needs to move in the positive direction, the calculation formula is: When the camera needs to move in the negative direction, the calculation formula is:

[0152] Where D move Finally, the actual distance required for the destacking operation is given by W, which is a fixed image width, and X. M2 To obtain the X-coordinate of the vertex of the laser reflection curve enclosed by the smallest bounding rectangle of the target at the second measurement position, K is the scaling factor for converting image pixel coordinates and real-world coordinates, and D is the coordinate of the laser reflection curve. c This is the horizontal distance between the electromagnetic gripper and the center of the camera.

[0153] Based on the above method, this application discloses an automated steel pipe destacking system based on line laser projection and machine vision, such as... Figure 6 As shown, electronic device 1 may include: at least one processor 11, at least one network interface 14, user interface 13, memory 15, and at least one communication bus 12.

[0154] The communication bus 12 is used to enable communication between these components.

[0155] The user interface 13 may include a display screen and a camera. Optionally, the user interface 13 may also include a standard wired interface and a wireless interface.

[0156] The network interface 14 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0157] The processor 11 may include one or more processing cores. The processor 11 connects to various parts of the server via various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 15, and by calling data stored in the memory 15. Optionally, the processor 11 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 11 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 11.

[0158] The memory 15 may include random access memory (RAM) or read-only memory. Optionally, the memory 15 may include non-transitory computer-readable storage medium. The memory 15 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 15 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 15 may also be at least one storage device located remotely from the aforementioned processor 11. Figure 6As shown, the memory 15, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for an automatic steel pipe destacking method based on line laser projection and machine vision.

[0159] exist Figure 6 In the electronic device 1 shown, the user interface 13 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 11 can be used to call the application stored in the memory 15 for the automated destacking method of steel pipe based on line laser projection and machine vision. When executed by one or more processors, the electronic device performs one or more methods as described in the above embodiments.

[0160] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0161] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0162] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0163] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0164] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0165] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0166] like Figure 7 As shown, the automated steel pipe destacking system based on line laser projection and machine vision provided in this embodiment of the invention is characterized by comprising:

[0167] The curved laser line image acquisition module is used to project a parallel line laser at a specific angle onto the surface of the steel pipe stack to be destabilized. The intersection of the laser line with the surface of the topmost steel pipe generates a series of laser reflection curves that vary with the contour of the steel pipe surface. The image of the curved laser line is then acquired using an industrial camera.

[0168] The image preprocessing module is used to perform image preprocessing steps on the obtained laser reflection curve image, including converting it to a grayscale image and performing binarization to segment the laser curve; subsequently, the laser reflection curve is labeled using the labeling tool anyLabeling, thereby constructing a training dataset;

[0169] The feature training module is used to train the dataset with the YOLOv8l model to ensure that the laser reflection curve of the steel pipe surface in the image can be correctly segmented in order to accurately obtain the target steel pipe area in the image.

[0170] The image reasoning module involves the main and secondary cameras moving parallel to each other on the conveyor belt and projecting a laser line onto the steel pipe stack. The main and secondary cameras acquire images sequentially, and the trained YOLOv8l model is used to reason about the images to determine the position of the curved laser line and its minimum bounding rectangle.

[0171] The second measurement position module moves the main and secondary cameras to a preset fixed distance to the second measurement position, and then successively acquires the laser reflection curve image of the steel pipe stack surface. The trained YOLOv8l model is used to infer the image to determine the position of the laser reflection curve and its minimum bounding rectangle.

[0172] The image matching module is used to match the position of the laser reflection curve and its minimum bounding rectangle in the images acquired by the main and secondary cameras at two measurement positions, and to determine the minimum bounding rectangle with the smallest pixel Y coordinate in each acquired image, which is the target minimum bounding rectangle.

[0173] The conversion relationship determination module is used to record the pixel coordinates of the curve vertex enclosed by the smallest bounding rectangle of the target in the images acquired by the main and secondary cameras at two positions, calculate the X coordinate distance difference, and combine it with the preset fixed motion distance to determine the conversion relationship between the pixel coordinate distance and the actual motion distance.

[0174] The real motion distance acquisition module obtains the real motion distance from the current position of the main and secondary cameras to the target steel pipe by using the vertex coordinates of the target laser reflection curve at the second measurement position, based on the conversion relationship between pixel coordinate distance and actual motion distance, in order to carry out subsequent destacking operations.

[0175] like Figure 8 As shown, an application embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of an automated steel pipe destacking method based on line laser projection and machine vision.

[0176] An application embodiment of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of an automated steel pipe destacking method based on line laser projection and machine vision.

[0177] An application embodiment of the present invention provides an information data processing terminal, which includes an automated steel pipe destacking system based on line laser projection and machine vision.

[0178] This invention is primarily applied to the destacking process of steel pipes in the steel pipe manufacturing and processing industry. Firstly, it achieves high-precision positioning of steel pipes by accurately measuring the laser reflection curve of the pipe stack surface. This is crucial for automated handling systems, ensuring that steel pipes are accurately grasped and placed, significantly reducing the risk of production interruptions or product damage due to positioning errors. Secondly, this invention accelerates the processing speed of steel pipes. The automated system can quickly identify and process steel pipes, significantly improving production efficiency. Furthermore, it reduces labor intensity and human error, minimizing workers' direct contact with heavy steel pipes, thereby improving workplace safety. In the long term, automated processing can also save on wages and human resource costs, while further reducing production costs due to increased processing speed and reduced scrap rates. This invention also improves product quality and consistency, adapts to different working environments, and is easily integrated with existing production lines and automated systems, providing enterprises with an easy-to-implement upgrade path. These combined advantages not only enhance the automation and intelligence level of the steel pipe industry but also bring considerable economic benefits and competitive advantages to enterprises.

[0179] The potential applications of this invention are not limited to these. Below are some specific application areas or related product examples, which further demonstrate the inventiveness and technical value of this invention:

[0180] 1. Logistics and Warehousing Automation:

[0181] This invention can be applied to automated warehouse management systems, particularly in logistics centers handling long goods such as steel bars, timber, and pipes. Through laser scanning and visual recognition technology, it enables automatic sorting, stacking, storage, retrieval, and inventory counting of goods, improving logistics efficiency and reducing manpower requirements.

[0182] 2. Intelligent Manufacturing System:

[0183] In the manufacturing industry, especially on production lines involving large mechanical parts and components, this invention can be used for precise part positioning, assembly quality inspection, and optimization of motion trajectories. This not only improves production efficiency but also significantly enhances product quality and manufacturing precision.

[0184] 3. Construction industry:

[0185] In the handling and assembly of building materials, such as heavy materials like steel bars and beams, this invention can help automated machinery identify and accurately place these materials, reducing manpower requirements and safety risks during the construction process.

[0186] 4. Robot navigation and collaboration:

[0187] This invention can be integrated into robot systems to provide high-precision environmental perception capabilities, enabling robots to navigate and perform tasks more accurately in complex environments, especially in the fields of industrial automation and high-precision manufacturing.

[0188] 5. Quality Control and Testing:

[0189] By using laser projection and computer vision technology, non-contact size and shape inspection can be performed on finished or semi-finished products to ensure that product quality meets stringent industry standards.

[0190] like Figure 9 As shown, this system mainly consists of a high-power line laser emitter, an industrial camera, an electromagnetic gripper, a programmable logic controller (PLC), and a high-precision servo motor. The industrial camera, combined with the high-power line laser emitter and electromagnetic gripper, forms two independent vision-gripping modules. These modules are distributed on both sides above the steel pipe stack, enabling the projection of line laser light onto both sides of the stack at a specific angle, and the acquisition of images via the industrial camera.

[0191] The PLC receives instructions from the host computer and controls the conveyor belt via servo motors. This drives two independent vision gripping modules above the material frame to move along the horizontal axis to two measurement positions to acquire images. The actual movement distance of the top steel pipe is then derived by the host computer software algorithm, thus completing the precise gripping operation.

[0192] The invention is realized through hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or contained in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0193] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An automated destacking method for steel pipes based on line laser projection and machine vision, characterized in that, include: S1. A parallel line laser is projected at a specific angle onto the surface of the steel pipe stack to be destabilized. The intersection of the laser with the surface of the top layer of steel pipe generates a series of laser reflection curves that vary with the contour of the steel pipe surface. An industrial camera is used to acquire the image of the laser line curve. S2, perform image preprocessing steps on the obtained laser reflection curve image, including converting it to a grayscale image and performing binarization to segment the laser curve; then, use the labeling tool anyLabeling to label the laser reflection curve, thereby constructing a training dataset; S3. Use the YOLOv8l model to train features on the dataset to ensure that the laser reflection curve of the steel pipe surface in the image can be correctly segmented in order to accurately obtain the target steel pipe area in the image. S4, driven by the conveyor belt, the main and secondary cameras move parallel to the first measurement position and project a laser line onto the steel pipe stack. The main and secondary cameras acquire images one after another, and use the trained YOLOv8l model to infer the images to determine the position of the curved laser line and its minimum bounding rectangle. S5, the main and secondary cameras move to a preset fixed distance to the second measurement position, and then successively acquire the laser reflection curve image of the steel pipe stack surface. The trained YOLOv8l model is used to infer the image to determine the position of the laser reflection curve and its minimum bounding rectangle. S6, match the laser reflection curve position and its minimum bounding rectangle in the images acquired by the main and secondary cameras at the two measurement positions, and determine the minimum bounding rectangle with the smallest pixel Y coordinate in each acquired image, which is the target minimum bounding rectangle; S7 records the pixel coordinates of the curve vertex enclosed by the smallest bounding rectangle of the target in the image acquired by the main and secondary cameras at two positions, calculates the distance difference in the X coordinate, and determines the conversion relationship between the pixel coordinate distance and the actual motion distance by combining the preset fixed motion distance. S8, based on the conversion relationship between pixel coordinate distance and actual movement distance, obtains the actual movement distance from the current position of the main and secondary cameras to the target steel pipe through the vertex coordinates of the target laser reflection curve at the second measurement position, in order to carry out subsequent destacking operations.

2. The automated destacking method for steel pipes based on line laser projection and machine vision as described in claim 1, characterized in that, S6 specifically includes: Step S61: Obtain feature information derived by the YOLOv8l model after the main and secondary cameras collect data at two measurement positions. The feature information includes the area of ​​the minimum bounding rectangle and the pixel coordinates of the curve vertex. Step S62: Using the image acquired by the main and secondary cameras at the first measurement position as a reference, compare it with the image acquired at the second position to determine whether the area ratio of each group of laser reflection curves exceeds the preset area ratio. Step S63: Obtain the X-coordinate translation pixel distance of the images acquired by the main and secondary cameras at the two measurement positions. Translate the minimum bounding rectangle of the laser reflection curve captured by the main and secondary cameras at the second measurement position into the image at the first measurement position. Perform positioning and matching of the vertices of the laser reflection curve. Then repeat step S62 to perform area matching.

3. The automated destacking method for steel pipes based on line laser projection and machine vision as described in claim 2, characterized in that, S61 specifically includes: calculating the area S of the smallest bounding rectangle derived from the YOLOv8l model for each image. r The set of areas of the smallest bounding rectangles that form the laser reflection curve, S. i Where i is the minimum bounding rectangle region of the i-th laser reflection curve, and the minimum bounding rectangle areas of the laser reflection curves are combined to form a list S. r ={S0,S1,…,S n }, where n is the total number of minimum bounding rectangles of the laser reflection curve on the image.

4. The automated destacking method for steel pipes based on line laser projection and machine vision as described in claim 2, characterized in that, S62 specifically includes: The comparison process starts from the image coordinate origin. Based on the X-coordinate of the upper left vertex of the smallest bounding rectangle of the laser reflection curve, the first smallest bounding rectangle is selected from the images acquired by the main camera at the two measurement positions. Then, the area is compared with the rectangles in the images acquired by the corresponding secondary cameras at the measurement positions. The comparison process is performed one-to-one, that is, one rectangle from the main camera is compared with one rectangle from the secondary camera each time. The formula is as follows: Among them, S i,m S refers to the area of ​​the smallest bounding rectangle of the laser reflection curve from the main camera. i,s This refers to the area of ​​the minimum bounding rectangle of the laser reflection curve from the secondary camera; By setting the area ratio R m To control the area ratio of the minimum bounding rectangle of each group of laser reflection curves, when the preset area ratio is appropriate, the small laser reflection curve area formed by the laser projection on the lower steel pipe or on the material frame can be deleted, allowing the error caused by local changes in the laser reflection curve area; When R i <R m When comparing the minimum bounding rectangle of the set of laser reflection curves, the area difference is too large and they do not correspond to the same steel pipe. Therefore, the minimum bounding rectangle with the smaller area is deleted. After deletion, the larger rectangle will continue to be compared with the next rectangle in another camera. When R i >R m At that time, the smallest bounding rectangle of the laser reflection curves being compared continues to be compared in a loop because the areas are similar, until all the rectangular information acquired by the main camera or the secondary camera at a measurement position is compared.

5. The automated destacking method for steel pipes based on line laser projection and machine vision as described in claim 2, characterized in that, S63 specifically includes: sorting the minimum bounding rectangles of the laser reflection curves acquired by the main and secondary cameras at the two measurement positions according to the X coordinate of their upper left corner vertex in ascending order, and selecting two rectangles from the sorted list to form a rectangle pair. For the rectangle features of each pair of rectangles extracted by the main and secondary cameras at the first and second measurement positions, the following parameters are calculated: Spacing D x To calculate the horizontal distance between the center points of two rectangles, use the following formula: D x =|X 2c -X 1c |, where X 1c and X 2c These are the X coordinates of the center points of the first and second rectangles, respectively. Areas A1 and A2: Calculate the area of ​​each rectangle separately using the formulas: A1 = W1 × H1, A2 = W2 × H2, where W1 and H1 are the width and height of the first rectangle, and W2 and H2 are the width and height of the second rectangle. Y-direction difference D y The formula for calculating the relative height difference between two rectangles in the vertical direction is: D y =|Y 2c -Y 1c |, where Y 1c and Y 2c These are the Y coordinates of the center points of the first and second rectangles, respectively. The following ratios are defined using the feature parameters obtained by the main and secondary cameras at the first and second measurement positions, and a total product value is further calculated to evaluate the similarity of the features. Spacing ratio The ratio of the rectangular pair spacing at the second measurement position to the rectangular pair spacing at the first measurement position is calculated using the following formula: Among them, D x,M1 It is the rectangular pair spacing D obtained by the camera at the first measurement position. x,M2 The ratio is calculated by dividing the maximum value by the minimum value to ensure that the ratio is always greater than or equal to 1, based on the rectangular pair spacing obtained by the camera at the second measurement position. Area ratio R A1 ,R A2 The formulas are as follows: The first ratio is the area of ​​the first rectangle in the middle of the two rectangles, and the second ratio is the area of ​​the second rectangle in the middle of the two rectangles. Among them, A 1,M1 Let A be the area of ​​the first rectangle centered at the first measurement position. 1,M2 Let A be the area of ​​the first rectangle centered at the second measurement position. 2,M1 Let A be the area of ​​the second rectangle centered at the first measurement position rectangle. 2,M2 The area of ​​the second rectangle in the second measurement position rectangle is calculated using the method of dividing the maximum value by the minimum value to ensure that the ratio is always greater than or equal to 1. Y-direction difference ratio R Dy The formula for calculating the ratio of the difference in the Y-direction between the center points of two rectangles is: Where D y,M1 D represents the difference in the Y-direction between the rectangles at the first measurement position. y,M2 The second measurement and control position rectangle represents the difference in the Y direction. When calculating this ratio, the method of dividing the maximum value by the minimum value was used to ensure that the ratio is always greater than or equal to 1. Based on the above ratios, a total product value P is calculated. This value combines all the ratios through a product method to evaluate the similarity of the features of the rectangle pairs. The formula is: P = R Dx ×R A1 ×R A2 ×R Dy The smaller the total product value, the more similar the rectangular features of the selected first measurement location and the rectangle pair of the second measurement are; Select the pair of rectangles with the smallest P-value, and use these four rectangles as a reference to determine the horizontal pixel offset D of the images acquired by the main and secondary cameras at the first and second measurement positions. x The calculation formula is: Where X 1,M1 Let X be the center x-coordinate of the first rectangle in the first measurement position rectangle pair. 1,M2 Let X be the center x-coordinate of the first rectangle in the second measurement position rectangle pair. 2,M1 Let X be the X-coordinate of the center of the second rectangle in the current rectangle pair at the first measurement position. 2,M2 The X-coordinate of the center of the second rectangle in the current rectangle pair at the second measurement position; Get the horizontal pixel offset value D x Then, the coordinates X of the vertex of each laser reflection curve obtained from the second measurement position of the main and secondary cameras are calculated. i,M2 The coordinates of the curve vertex X after translation i ′ ,M2 The formula is X i ′ ,M2 =X i,M2 +D x After completing the offset calculation, check the X-coordinates of the curve vertex after the second measurement position has been translated. i ′ ,M2 Is it the rectangle R corresponding to its first measurement position? i,M1 Match successful. The conditions for a successful match are: in It is a rectangle R i,M1 The top-left X-coordinate For rectangle R i,M1 The X-coordinate of the bottom right corner; if the match fails, then the X-coordinate of the vertex is considered to be... i,M2 The smallest bounding rectangle R in the image of the second measurement location i,M2 If a match is found, the rectangle is removed from the list of rectangles at the second measurement location; if a match is found, step S62 is repeated to perform area matching to determine if the minimum bounding rectangle of the set of laser reflection curves corresponds to the same pipe in the image. Through the above matching process, the number of minimum bounding rectangles of the laser reflection curves obtained by the main and secondary cameras at the first and second measurement positions will remain consistent, and they will be sorted in ascending order of X-coordinate. By matching the index, it is ensured that the rectangles obtained by the main and secondary cameras at the two measurement positions are matched to the same pipe. Accordingly, based on the list of minimum bounding rectangles of the laser reflection curves recorded by the main camera at the first measurement position, the rectangle with the smallest Y-coordinate at the top left corner is selected, and the corresponding minimum bounding rectangle of the target obtained by the main and secondary cameras at the two measurement positions is determined according to its index. This is the topmost steel pipe in the current stack.

6. The automated destacking method for steel pipes based on line laser projection and machine vision as described in claim 1, characterized in that, S7 specifically includes: By recording the difference in the vertex X pixel coordinates of the laser reflection curve enclosed by the minimum bounding rectangle of the target in the images acquired by the main and secondary cameras at two measurement positions, and calculating the conversion relationship between the pixel coordinates of the images acquired by the two cameras and the real-world coordinates, the scaling factor K is obtained. m K s The calculation formula is as follows: Where D real The preset horizontal movement distance ΔX between the two measurement positions. m The difference in vertex X pixel coordinates, ΔX, is obtained from the laser reflection curve bounded by the smallest bounding rectangle of the target at two measurement positions of the main camera. s The difference in X-pixel coordinates of the vertex of the laser reflection curve enclosed by the smallest bounding rectangle of the target is obtained from the two measurement positions of the secondary camera.

7. The automated destacking method for steel pipes based on line laser projection and machine vision as described in claim 1, characterized in that, S8 specifically includes: Based on the pixel deviation between the vertex X coordinate of the target laser reflection curve obtained by the main and secondary cameras at the second measurement point and the center of the fixed image width, and the scaling factor determined in step S7, combined with the actual distance D between the electromagnetic gripper and the camera center, c The precise movement distance from the second measurement position to the target steel pipe can be calculated. The formula for calculating the actual movement distance depends on the installation position of the electromagnetic gripper and the camera's movement direction. The X-coordinate of the laser reflection curve vertex is considered positive if it exceeds the center of the image width, and negative if it is below. The electromagnetic gripper is installed in the positive direction of the camera's movement: When the camera needs to move in the positive direction, the calculation formula is: When the camera needs to move in the negative direction, the calculation formula is: The electromagnetic gripper is installed in the negative direction of the camera's movement: When the camera needs to move in the positive direction, the calculation formula is: When the camera needs to move in the negative direction, the calculation formula is: Where D move Finally, the actual distance required for the destacking operation is given by W, which is a fixed image width, and X. M2 To obtain the X-coordinate of the vertex of the laser reflection curve enclosed by the smallest bounding rectangle of the target at the second measurement position, K is the scaling factor for converting image pixel coordinates and real-world coordinates, and D is the coordinate of the laser reflection curve. c This is the horizontal distance between the electromagnetic gripper and the center of the camera.

8. An automated steel pipe destacking system based on line laser projection and machine vision, implementing the automated steel pipe destacking method based on line laser projection and machine vision as described in any one of claims 1 to 7, characterized in that, include: The curved laser line image acquisition module is used to project a parallel line laser at a specific angle onto the surface of the steel pipe stack to be destabilized. The intersection of the laser line with the surface of the topmost steel pipe generates a series of laser reflection curves that vary with the contour of the steel pipe surface. The image of the curved laser line is then acquired using an industrial camera. The image preprocessing module is used to perform image preprocessing steps on the obtained laser reflection curve image, including converting it to a grayscale image and performing binarization to segment the laser curve; Subsequently, the laser reflection curves were labeled using the labeling tool anyLabeling to construct the training dataset; The feature training module is used to train the dataset with the YOLOv8l model to ensure that the laser reflection curve of the steel pipe surface in the image can be correctly segmented in order to accurately obtain the target steel pipe area in the image. The image reasoning module involves the main and secondary cameras moving parallel to each other on the conveyor belt and projecting a laser line onto the steel pipe stack. The main and secondary cameras acquire images sequentially, and the trained YOLOv8l model is used to reason about the images to determine the position of the curved laser line and its minimum bounding rectangle. The second measurement position module moves the main and secondary cameras to a preset fixed distance to the second measurement position, and then successively acquires the laser reflection curve image of the steel pipe stack surface. The trained YOLOv8l model is used to infer the image to determine the position of the laser reflection curve and its minimum bounding rectangle. The image matching module is used to match the position of the laser reflection curve and its minimum bounding rectangle in the images acquired by the main and secondary cameras at two measurement positions, and to determine the minimum bounding rectangle with the smallest pixel Y coordinate in each acquired image, which is the target minimum bounding rectangle. The conversion relationship determination module is used to record the pixel coordinates of the curve vertex enclosed by the smallest bounding rectangle of the target in the images acquired by the main and secondary cameras at two positions, calculate the X coordinate distance difference, and combine it with the preset fixed motion distance to determine the conversion relationship between the pixel coordinate distance and the actual motion distance. The real motion distance acquisition module obtains the real motion distance from the current position of the main and secondary cameras to the target steel pipe by using the vertex coordinates of the target laser reflection curve at the second measurement position, based on the conversion relationship between pixel coordinate distance and actual motion distance, in order to carry out subsequent destacking operations.

9. A computer device comprising a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the automated destacking method for steel pipes based on line laser projection and machine vision as described in any one of claims 1 to 7.

10. An information data processing terminal, comprising the automated steel pipe destacking system based on line laser projection and machine vision as described in claim 8.

Images