A method for recognizing irregular steel plate posture and related equipment
Through irregular steel plate pose recognition methods, including taking steel plate pose images, preprocessing and clustering algorithm segmentation, the best fit plane and center point are determined, which solves the problem of inaccurate identification of traditional robot systems and improves steel plate grabbing Efficiency and productivity.
Patent Information
- Application Number
- CN202210736233.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-27
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2042-06-27
AI Technical Summary
When traditional robot systems deal with irregular steel plates stacked indiscriminately, their recognition position is inaccurate, resulting in low production efficiency and unable to meet the intelligent needs of industrial production.
The irregular steel plate position recognition method is adopted, including taking steel plate position image, pre-processing to remove background, extracting point cloud information, segmenting the steel plate through the least squares method and clustering algorithm, determining the best fit plane and center point, using Mean- The shift algorithm calculates the center point and obtains the coordinate information of the steel plate to guide the manipulator to grab it.
It improves the efficiency of the grabbing operation of irregular steel plates by the robot system, improves production efficiency, and meets the intelligent needs of industrial production.
Smart Images

Figure CN115100416B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of automatic optical detection, and in particular relates to an irregular steel plate posture recognition method and related equipment. Background Art
[0002] With the development of science and technology, the robotics industry has continued to progress, allowing robotics technology to improve people's quality of life and be applied to many aspects of life, medical care, manufacturing, etc. Among them, industrial production is one of the main applications of robots. The monotonous and repetitive labor in the factory, such as the alignment, sorting, stacking, loading and unloading of production objects in the production line, has been transformed from the original manual work to the robot, which greatly reduces the workload of factory employees and enables people to better focus on non-mechanical tasks. On the production line, the traditional loading mechanism occupies a large area and has reduced reliability. Manual loading will cause fatigue and danger. In response to this problem, industrial robots' object recognition and classification, posture estimation, and grasping technology have become the research direction and theme of many scholars and technicians. Traditional robot systems are often only suitable for simple structured scenarios and perform single repetitive tasks through offline programming. In a common scenario, for the grasping of irregularly stacked steel plates, traditional machine vision algorithms can only segment and identify the single position features of the steel plates, and the pose calculation effect of randomly stacked workpieces is poor. This results in the system based on the algorithm being unable to efficiently guide the robot to perform grasping operations. In other words, most existing robot systems can no longer meet the increasing demand for intelligence in industrial production. Summary of the invention
[0003] The embodiment of the present invention provides a method for recognizing the position of an irregular steel plate and related equipment, aiming to solve the problem of inaccurate recognition of the position of an irregular steel plate in the existing manufacturing industry, resulting in low production efficiency.
[0004] In a first aspect, an embodiment of the present invention provides a method for recognizing an irregular steel plate posture, the method comprising the following steps:
[0005] S1, taking a steel plate posture image, and removing the background outside the steel plate detection area through preprocessing, and extracting the point cloud information in the steel plate detection area;
[0006] S2. Smoothing the point cloud information by least square method to obtain point cloud origin data;
[0007] S3. Using a preset clustering algorithm, segment the point cloud origin data to obtain separate point cloud information of a single steel plate;
[0008] S4, performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features;
[0009] S5, determining the best fitting plane according to the point cloud data;
[0010] S6. Calculate the center point of the best fit plane using a Mean-shift algorithm;
[0011] S7. Acquire coordinate information of the steel plate according to the best fitting plane and the center point, and grab the steel plate using a manipulator according to the coordinate information.
[0012] Furthermore, in step S1, the preprocessing method includes:
[0013] Removing outliers from the point cloud in the steel plate pose image based on statistical filtering, wherein the outliers include noise around the point cloud;
[0014] The steel plate detection area is identified by using an AABB bounding box combined with a straight-through filter, and the influence of the area boundary is removed.
[0015] Furthermore, in step S3, the preset clustering algorithm is specifically:
[0016] S31, sorting the point cloud origin data according to the size of the Z-axis coordinate value, obtaining a seed point P with the largest Z-axis coordinate value, and putting the seed point P into the cluster set Q;
[0017] S32, constructing a kd number about the seed point P, and finding n neighbors of the seed point P, calculating the distance from each neighbor to the seed point P, and putting the points corresponding to the neighbors whose distance is less than a preset distance threshold r into a neighbor array;
[0018] S33, traverse the neighbor array, calculate the angle between the normal vector corresponding to each neighbor and the normal vector of the seed point P, and put the preset points of the neighbors whose angles are less than the preset angle threshold into the cluster set Q. According to the above definition, the cluster set Q satisfies the following relationship:
[0019]
[0020] S34. Take the next point of the seed point P in the cluster set Q as the new seed point, and repeat steps S32-S34 until no new points can be added to the cluster set Q, wherein each point in the cluster set Q is the individual point cloud information of a single steel plate.
[0021] Furthermore, in step S4, the step of performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features includes the following sub-steps:
[0022] S41, calculating the point cloud centroid of a random point through the neighborhood points of the random point in the individual point cloud information;
[0023] S42, calculating a covariance matrix C using the random points and the point cloud centroid;
[0024] S43, calculating the eigenvector of the covariance matrix C by a singular value solution method, wherein the minimum value of the eigenvector is the normal vector feature of the random point;
[0025] S44. Use the corresponding points whose normal vector feature values are greater than a preset feature threshold as the point cloud data.
[0026] Furthermore, in step S5, the step of determining the best fitting plane according to the point cloud data is specifically as follows:
[0027] Using the RANSAC method, three points in the point cloud set data are randomly selected to form a sampling plane, the distance from each point in the point cloud set data to the sampling plane is calculated, and points greater than a preset plane distance threshold are placed in a plane point set. The above steps are repeated until the number of points in the plane point set is the largest, and the sampling plane at this time is used as the best fitting plane.
[0028] Furthermore, in step S6, the step of calculating the center point of the best fitting plane using the Mean-shift algorithm includes the following sub-steps:
[0029] S61, randomly obtain the center point x in the best fitting plane i , use the kd tree to search for the center point x i Neighborhood S k ;
[0030] S62, counting the central point x i With the neighborhood S k The vector formed by the points in the The end point coordinate of the sampling vector is taken as x i+1 , the sampling vector The following relationship is satisfied:
[0031]
[0032] S63, repeating steps S61-S62 until the end point coordinates of the sampling vector no longer change.
[0033] Furthermore, in step S7, the coordinate information of the steel plate takes the best fitting plane as the plane where the steel plate is located, and takes the end point coordinates of the sampling vector as the coordinates of the plane.
[0034] In a second aspect, an embodiment of the present invention further provides an irregular steel plate posture recognition system, comprising:
[0035] An image acquisition module is used to capture the steel plate posture image, and remove the background outside the steel plate detection area through preprocessing to extract the point cloud information in the steel plate detection area;
[0036] A data smoothing processing module is used to smooth the point cloud information by using the least square method to obtain point cloud origin data;
[0037] A segmentation module, used to segment the point cloud origin data using a preset clustering algorithm to obtain separate point cloud information of a single steel plate;
[0038] A posture recognition module, used for performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features;
[0039] A fitting plane calculation module, used to determine the best fitting plane according to the point cloud data;
[0040] A center point calculation module, used for calculating the center point of the best fitting plane using a Mean-shift algorithm;
[0041] A coordinate determination module is used to obtain the coordinate information of the steel plate according to the best fitting plane and the center point, and to grab the steel plate using a manipulator according to the coordinate information.
[0042] In a third aspect, an embodiment of the present invention further provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the irregular steel plate posture recognition method as described in any one of the above embodiments are implemented.
[0043] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the irregular steel plate posture recognition method as described in any one of the above embodiments are implemented.
[0044] The beneficial effect achieved by the present invention is that the irregular position of the steel plate is identified by the method of fitting the plane and then determining the coordinate information, so that the operating efficiency of the robot system for grasping the steel plate in the factory environment can be improved, thereby improving the production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 is a flowchart of the steps of the irregular steel plate posture recognition method provided by an embodiment of the present invention;
[0046] Figure 2 is a structural schematic diagram of an irregular steel plate posture recognition system 200 provided in an embodiment of the present invention;
[0047] Figure 3 It is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0048] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0049] Please refer to Figure 1 , Figure 1 : is a flowchart of the irregular steel plate posture recognition method provided by an embodiment of the present invention, which specifically includes the following steps:
[0050] S1. Capture the steel plate posture image, and remove the background outside the steel plate detection area through preprocessing to extract the point cloud information in the steel plate detection area.
[0051] In the embodiment of the present invention, the shooting object of the steel plate posture image is the steel plate on the transport plate or transport rack, the placement positions of the steel plates are not completely consistent, and the heights and directions are different.
[0052] Furthermore, in step S1, the preprocessing method includes:
[0053] Removing outliers from the point cloud in the steel plate pose image based on statistical filtering, wherein the outliers include noise around the point cloud;
[0054] The steel plate detection area is identified by using an AABB bounding box combined with a straight-through filter, and the influence of the area boundary is removed.
[0055] S2. Smoothing the point cloud information by least square method to obtain point cloud origin data.
[0056] S3. Use a preset clustering algorithm to segment the point cloud origin data to obtain separate point cloud information of a single steel plate.
[0057] Furthermore, in step S3, the preset clustering algorithm is specifically:
[0058] S31, sorting the point cloud origin data according to the size of the Z-axis coordinate value, obtaining a seed point P with the largest Z-axis coordinate value, and putting the seed point P into the cluster set Q;
[0059] S32, constructing a kd number about the seed point P, and finding n neighbors of the seed point P, calculating the distance from each neighbor to the seed point P, and putting the points corresponding to the neighbors whose distance is less than a preset distance threshold r into a neighbor array;
[0060] S33, traverse the neighbor array, calculate the angle between the normal vector corresponding to each neighbor and the normal vector of the seed point P, and put the preset points of the neighbors whose angles are less than the preset angle threshold into the cluster set Q. According to the above definition, the cluster set Q satisfies the following relationship:
[0061]
[0062] S34. Take the next point of the seed point P in the cluster set Q as the new seed point, and repeat steps S32-S34 until no new points can be added to the cluster set Q, wherein each point in the cluster set Q is the individual point cloud information of a single steel plate.
[0063] S4. Performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features.
[0064] Furthermore, in step S4, the step of performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features includes the following sub-steps:
[0065] S41. Calculate the point cloud centroid of a random point in the individual point cloud information through the neighborhood points of the random point.
[0066] Exemplarily, let the random point be p i , the neighborhood points include at least p i1 、p i2 、p i3 …, then the point cloud centroid satisfies:
[0067]
[0068] S42. Calculate a covariance matrix C using the random points and the point cloud centroid.
[0069] According to step S41, the covariance matrix C satisfies:
[0070]
[0071] S43. Calculate the eigenvector of the covariance matrix C by a singular value solution method, wherein the minimum value of the eigenvector is the normal vector feature of the random point.
[0072] With λ j is the jth eigenvalue, is the jth eigenvector, then the singular value solution method can be expressed as:
[0073]
[0074] S44. Use the corresponding points whose normal vector feature values are greater than a preset feature threshold as the point cloud data.
[0075] S5. Determine the best fitting plane according to the point cloud data.
[0076] Furthermore, in step S5, the step of determining the best fitting plane according to the point cloud data is specifically as follows:
[0077] Using the RANSAC method, three points in the point cloud set data are randomly selected to form a sampling plane, the distance from each point in the point cloud set data to the sampling plane is calculated, and points greater than a preset plane distance threshold are placed in a plane point set. The above steps are repeated until the number of points in the plane point set is the largest, and the sampling plane at this time is used as the best fitting plane.
[0078] S6. Calculate the center point of the best fitting plane using a Mean-shift algorithm.
[0079] Furthermore, in step S6, the step of calculating the center point of the best fitting plane using the Mean-shift algorithm includes the following sub-steps:
[0080] S61, randomly obtain the center point x in the best fitting plane i , use the kd tree to search for the center point x i Neighborhood S k ;
[0081] S62, counting the central point x i With the neighborhood S k The vector formed by the points in the The end point coordinate of the sampling vector is taken as x i+1 , the sampling vector The following relationship is satisfied:
[0082]
[0083] S63, repeating steps S61-S62 until the end point coordinates of the sampling vector no longer change.
[0084] S7. Acquire coordinate information of the steel plate according to the best fitting plane and the center point, and grab the steel plate using a manipulator according to the coordinate information.
[0085] Furthermore, in step S7, the coordinate information of the steel plate takes the best fitting plane as the plane where the steel plate is located, and takes the end point coordinates of the sampling vector as the coordinates of the plane.
[0086] The beneficial effect achieved by the present invention is that the irregular position of the steel plate is identified by the method of fitting the plane and then determining the coordinate information, so that the operating efficiency of the robot system for grasping the steel plate in the factory environment can be improved, thereby improving the production efficiency.
[0087] The embodiment of the present invention also provides an irregular steel plate posture recognition system, please refer to Figure 2 , Figure 2 : is a structural schematic diagram of an irregular steel plate posture recognition system 200 provided in an embodiment of the present invention, wherein the irregular steel plate posture recognition system 200 comprises:
[0088] The image acquisition module 201 is used to capture the steel plate posture image, and remove the background outside the steel plate detection area through preprocessing, and extract the point cloud information in the steel plate detection area;
[0089] A data smoothing processing module 202 is used to smooth the point cloud information by using the least square method to obtain point cloud origin data;
[0090] The segmentation module 203 is used to segment the point cloud origin data using a preset clustering algorithm to obtain separate point cloud information of a single steel plate;
[0091] A posture recognition module 204 is used to perform posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features;
[0092] A fitting plane calculation module 205 is used to determine the best fitting plane according to the point cloud data;
[0093] A center point calculation module 206, configured to calculate the center point of the best fitting plane using a Mean-shift algorithm;
[0094] The coordinate determination module 207 is used to obtain the coordinate information of the steel plate according to the best fitting plane and the center point, and to grab the steel plate using a manipulator according to the coordinate information.
[0095] The irregular steel plate posture recognition system 200 can implement the steps in the irregular steel plate posture recognition method in the above embodiment, and can achieve the same technical effects. Please refer to the description in the above embodiment and will not be repeated here.
[0096] The embodiment of the present invention also provides a computer device, please refer to Figure 3 , Figure 3300 is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. The computer device 300 includes: a memory 302, a processor 301, and a computer program stored in the memory 302 and executable on the processor 301.
[0097] The processor 301 calls the computer program stored in the memory 302 to execute the steps of the irregular steel plate posture recognition method provided in the embodiment of the present invention. Figure 1 , specifically including:
[0098] S1. Capture the steel plate posture image, and remove the background outside the steel plate detection area through preprocessing to extract the point cloud information in the steel plate detection area.
[0099] Furthermore, in step S1, the preprocessing method includes:
[0100] Removing outliers from the point cloud in the steel plate pose image based on statistical filtering, wherein the outliers include noise around the point cloud;
[0101] The steel plate detection area is identified by using an AABB bounding box combined with a straight-through filter, and the influence of the area boundary is removed.
[0102] S2. Smoothing the point cloud information by least square method to obtain point cloud origin data.
[0103] S3. Use a preset clustering algorithm to segment the point cloud origin data to obtain separate point cloud information of a single steel plate.
[0104] Furthermore, in step S3, the preset clustering algorithm is specifically:
[0105] S31, sorting the point cloud origin data according to the size of the Z-axis coordinate value, obtaining a seed point P with the largest Z-axis coordinate value, and putting the seed point P into the cluster set Q;
[0106] S32, constructing a kd number about the seed point P, and finding n neighbors of the seed point P, calculating the distance from each neighbor to the seed point P, and putting the points corresponding to the neighbors whose distance is less than a preset distance threshold r into a neighbor array;
[0107] S33, traverse the neighbor array, calculate the angle between the normal vector corresponding to each neighbor and the normal vector of the seed point P, and put the preset points of the neighbors whose angles are less than the preset angle threshold into the cluster set Q. According to the above definition, the cluster set Q satisfies the following relationship:
[0108]
[0109] S34. Take the next point of the seed point P in the cluster set Q as the new seed point, and repeat steps S32-S34 until no new points can be added to the cluster set Q, wherein each point in the cluster set Q is the individual point cloud information of a single steel plate.
[0110] S4. Performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features.
[0111] Furthermore, in step S4, the step of performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features includes the following sub-steps:
[0112] S41, calculating the point cloud centroid of a random point through the neighborhood points of the random point in the individual point cloud information;
[0113] S42, calculating a covariance matrix C using the random points and the point cloud centroid;
[0114] S43, calculating the eigenvector of the covariance matrix C by a singular value solution method, wherein the minimum value of the eigenvector is the normal vector feature of the random point;
[0115] S44. Use the corresponding points whose normal vector feature values are greater than a preset feature threshold as the point cloud data.
[0116] S5. Determine the best fitting plane according to the point cloud data.
[0117] Furthermore, in step S5, the step of determining the best fitting plane according to the point cloud data is specifically as follows:
[0118] Using the RANSAC method, three points in the point cloud set data are randomly selected to form a sampling plane, the distance from each point in the point cloud set data to the sampling plane is calculated, and points greater than a preset plane distance threshold are placed in a plane point set. The above steps are repeated until the number of points in the plane point set is the largest, and the sampling plane at this time is used as the best fitting plane.
[0119] S6. Calculate the center point of the best fitting plane using a Mean-shift algorithm.
[0120] Furthermore, in step S6, the step of calculating the center point of the best fitting plane using the Mean-shift algorithm includes the following sub-steps:
[0121] S61, randomly obtain the center point x in the best fitting plane i , use the kd tree to search for the center point x i Neighborhood Sk ;
[0122] S62, counting the central point x i With the neighborhood S k The vector formed by the points in the The end point coordinate of the sampling vector is taken as x i+1 , the sampling vector The following relationship is satisfied:
[0123]
[0124] S63, repeating steps S61-S62 until the end point coordinates of the sampling vector no longer change.
[0125] S7. Acquire coordinate information of the steel plate according to the best fitting plane and the center point, and grab the steel plate using a manipulator according to the coordinate information.
[0126] Furthermore, in step S7, the coordinate information of the steel plate takes the best fitting plane as the plane where the steel plate is located, and takes the end point coordinates of the sampling vector as the coordinates of the plane.
[0127] The computer device 300 provided in the embodiment of the present invention can implement the steps in the irregular steel plate posture recognition method in the above embodiment, and can achieve the same technical effect. Please refer to the description in the above embodiment and will not be repeated here.
[0128] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes and steps in the irregular steel plate posture recognition method provided in the embodiment of the present invention are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0129] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).
[0130] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.
[0131] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.
[0132] The embodiments of the present invention are described above in conjunction with the accompanying drawings. What is disclosed is only the preferred embodiment of the present invention. However, the present invention is not limited to the above-mentioned specific implementation manner. The above-mentioned specific implementation manner is only illustrative rather than restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms and equivalent changes without departing from the scope of protection of the purpose of the present invention and the claims, all of which are within the protection of the present invention.
Claims
1. A method for recognizing the position and posture of an irregular steel plate, characterized in that: The method comprises the following steps: S1, taking a steel plate posture image, and removing the background outside the steel plate detection area through preprocessing, and extracting the point cloud information in the steel plate detection area; S2. Smoothing the point cloud information by least square method to obtain point cloud origin data; S3. Using a preset clustering algorithm, segment the point cloud origin data to obtain separate point cloud information of a single steel plate; S4, performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features; S5, determining the best fitting plane according to the point cloud data; S6. Calculate the center point of the best fit plane using a Mean-shift algorithm; S7, acquiring coordinate information of the steel plate according to the best fitting plane and the center point, and grabbing the steel plate using a manipulator according to the coordinate information; In step S3, the preset clustering algorithm is specifically: S31, sorting the point cloud origin data according to the size of the Z-axis coordinate value, obtaining a seed point P with the largest Z-axis coordinate value, and putting the seed point P into the cluster set Q; S32, constructing a kd number about the seed point P, and finding n neighbors of the seed point P, calculating the distance from each neighbor to the seed point P, and putting the points corresponding to the neighbors whose distance is less than a preset distance threshold r into a neighbor array; S33, traverse the neighbor array, calculate the angle between the normal vector corresponding to each neighbor and the normal vector of the seed point P, and put the preset points of the neighbors whose angles are less than the preset angle threshold into the cluster set Q. According to the above definition, the cluster set Q satisfies the following relationship: Q={P i (φ,d|φ i ≤θ,d i ≤r)} S34. Take the next point of the seed point P in the cluster set Q as the new seed point, and repeat steps S32-S34 until no new points can be added to the cluster set Q, wherein each point in the cluster set Q is the individual point cloud information of a single steel plate.
2. The irregular steel plate posture recognition method according to claim 1, characterized in that: In step S1, the preprocessing method includes: Removing outliers from the point cloud in the steel plate pose image based on statistical filtering, wherein the outliers include noise around the point cloud; The steel plate detection area is identified by using an AABB bounding box combined with a straight-through filter, and the influence of the area boundary is removed.
3. The irregular steel plate posture recognition method according to claim 1, characterized in that: In step S4, the step of performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features includes the following sub-steps: S41, calculating the point cloud centroid of a random point through the neighborhood points of the random point in the individual point cloud information; S42, calculating a covariance matrix C using the random points and the point cloud centroid; S43, calculating the eigenvector of the covariance matrix C by a singular value solution method, wherein the minimum value of the eigenvector is the normal vector feature of the random point; S44. Use the corresponding points whose normal vector feature values are greater than a preset feature threshold as the point cloud data.
4. The irregular steel plate posture recognition method according to claim 3, characterized in that: In step S5, the step of determining the best fitting plane according to the point cloud data is specifically as follows: Using the RANSAC method, three points in the point cloud set data are randomly selected to form a sampling plane, the distance from each point in the point cloud set data to the sampling plane is calculated, and points greater than a preset plane distance threshold are placed in a plane point set. The above steps are repeated until the number of points in the plane point set is the largest, and the sampling plane at this time is used as the best fitting plane.
5. The irregular steel plate posture recognition method according to claim 4, characterized in that: In step S6, the step of calculating the center point of the best fitting plane using the Mean-shift algorithm includes the following sub-steps: S61, randomly obtaining the center point in the best fitting plane x i , use kd tree to search for the center point x i Neighborhood S k ; S62: Count the central points x i With the neighborhood S k The vector formed by the points in the , the end point coordinates of the sampling vector are taken as x i+1 , the sampling vector The following relationship is satisfied: S63, repeating steps S61-S62 until the end point coordinates of the sampling vector no longer change.
6. The irregular steel plate position recognition method according to claim 5, characterized in that: In step S7, the coordinate information of the steel plate is obtained by taking the best fitting plane as the plane where the steel plate is located and taking the end point coordinates of the sampling vector as the coordinates of the plane where the steel plate is located.
7. An irregular steel plate posture recognition system, characterized in that: include: An image acquisition module is used to capture the steel plate posture image, and remove the background outside the steel plate detection area through preprocessing to extract the point cloud information in the steel plate detection area; A data smoothing processing module is used to smooth the point cloud information by using the least square method to obtain point cloud origin data; A segmentation module, used to segment the point cloud origin data using a preset clustering algorithm to obtain separate point cloud information of a single steel plate; A posture recognition module, used for performing posture recognition on the individual point cloud information to obtain point cloud set data with normal vector features; A fitting plane calculation module, used to determine the best fitting plane according to the point cloud data; A center point calculation module, used for calculating the center point of the best fitting plane using a Mean-shift algorithm; A coordinate determination module, used for acquiring coordinate information of the steel plate according to the best fitting plane and the center point, and grabbing the steel plate using a manipulator according to the coordinate information; The preset clustering algorithm is specifically: Sort the point cloud origin data according to the size of the Z-axis coordinate value, obtain a seed point P with the largest Z-axis coordinate value, and put the seed point P into the cluster set Q; Construct a kd number about the seed point P, find n neighbors of the seed point P, calculate the distance from each neighbor to the seed point P, and put the points corresponding to the neighbors whose distance is less than a preset distance threshold r into a neighbor array; Traverse the neighbor array, calculate the angle between the normal vector corresponding to each neighbor and the normal vector of the seed point P, and put the preset points of the neighbors whose angles are less than the preset angle threshold into the cluster set Q. According to the above definition, the cluster set Q satisfies the following relationship: Q={P i (φ,d|φ i ≤θ,d i ≤r)} The next point of the seed point P in the cluster set Q is used as the new seed point, and steps S32-S34 are repeated until no new points can be added to the cluster set Q, wherein each point in the cluster set Q is the individual point cloud information of a single steel plate.
8. A computer device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the irregular steel plate posture recognition method as described in any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps in the irregular steel plate posture recognition method according to any one of claims 1 to 6 are implemented.
