A method and system for determining the position of a clamp

By adjusting the tilt angle and feature extraction of the three-dimensional point cloud image of the clamp, combined with SIFT algorithm and model matching, the problem of insufficient accuracy of the traditional feature extraction algorithm in clamp positioning is solved, and more efficient image analysis and real-time Positioning effect.

CN118154686BActive Publication Date: 2024-11-01GUANGZHOU WENRUI SCI INSTR CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410359603.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-27
Publication Date
2024-11-01
Estimated Expiration
2044-03-27

AI Technical Summary

Technical Problem

In the prior art, when analyzing the three-dimensional point cloud image of clamps using traditional feature extraction algorithms, the accuracy and reliability are insufficient, resulting in low accuracy and timeliness of real-time positioning of clamps and poor image analysis effect.

Method used

By acquiring the three-dimensional point cloud image of the vertically placed clamp, adjusting the tilt angle to obtain the two-dimensional point cloud image, analyzing the symmetry and parallel plane features of the two-dimensional point cloud image, determining the optimal angle, based on the SIFT feature extraction algorithm and Model matching algorithm, filtering key feature points to determine the position information of the clamp.

Benefits of technology

It improves the accuracy and timeliness of clamp position information, enhances the accuracy and efficiency of image analysis, avoids the influence of point cloud data misalignment and abnormal points, and achieves more accurate real-time positioning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118154686B_ABST
    Figure CN118154686B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image analysis technology, and specifically to a method and system for determining the position of a clamp. The method includes: obtaining a three-dimensional point cloud image, determining the data points to be measured, and obtaining two-dimensional point cloud images corresponding to the data points to be measured at different inclination angles; performing slope feature analysis based on the straight lines to be analyzed between every two closest data points, determining the symmetry of the two-dimensional point cloud image, and determining the optimal angle based on the symmetry; determining the quantity influence factor based on the number of data points in each plane, and performing feature extraction on each two-dimensional point cloud image and parallel plane at the optimal angle based on the quantity influence factor and symmetry to obtain key feature points; comparing the key feature points with preset standard feature points to determine the position information of the clamp. The present invention can effectively extract key feature points, ensure the accuracy and timeliness of position information, and accurately and efficiently perform real-time positioning of the clamp.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a method and system for determining the position of a clamp. Background Art

[0002] Remotely controlling mechanical equipment to grab objects can improve the efficiency of cargo transportation while also ensuring the safety of users. The goods to be transported can be grabbed by a clamp and placed in the required location. During the operation, the user needs to determine the position information of the clamp, and control the movement of the clamp through the obtained position information to better grab and transport the object. The position determination of the clamp affects the reliability of the clamping process, so the clamp can be positioned in combination with image analysis.

[0003] In the related art, radar scanning is used to acquire a point cloud image of a scene, and the region of interest is extracted to obtain a three-dimensional point cloud image of the clamp. A three-dimensional feature extraction algorithm is used to analyze the feature points and feature descriptors of the three-dimensional point cloud image, thereby achieving the positioning of the clamp. During the analysis of the point cloud image, due to the large amount of point cloud data and the presence of abnormal noise points, the accuracy and reliability of feature point extraction using traditional feature extraction algorithms are insufficient, resulting in low accuracy and timeliness of real-time positioning of the clamp and poor image analysis results. Summary of the invention

[0004] In order to solve the technical problems that the accuracy and reliability of feature point extraction using traditional feature extraction algorithms in related technologies are insufficient, resulting in low accuracy and timeliness of real-time positioning of clamps and poor image analysis effects, the present invention provides a method and system for determining the position of clamps, and the technical solutions adopted are as follows:

[0005] The present invention provides a method for determining the position of a clamp, the method comprising:

[0006] Acquire a three-dimensional point cloud image of the vertically placed clamp, use the coordinate axis corresponding to the vertical direction as a fixed axis, use any data point in the three-dimensional point cloud image as a data point to be measured, adjust the inclination angle on the fixed axis through the data point to be measured, and obtain two-dimensional point cloud images of the data point to be measured corresponding to different inclination angles;

[0007] Connecting all the closest data points on the two-dimensional point cloud image to obtain a straight line to be analyzed between every two closest data points, determining the symmetry of the two-dimensional point cloud image according to the slope characteristics of all the straight lines to be analyzed, obtaining at least two parallel planes parallel to the two-dimensional point cloud image, and determining the screening degree of the inclination angle according to the symmetry of the two-dimensional point cloud image and all the parallel planes;

[0008] The tilt angle is adjusted, and the optimal angle is selected according to the screening degree of all tilt angles; the quantity influence factor is determined according to the number of all data points in each two-dimensional point cloud image and parallel plane, and the feature extraction is performed on each two-dimensional point cloud image and parallel plane at the optimal angle according to the quantity influence factor and symmetry to obtain key feature points;

[0009] The position information of the clamp is determined by comparing the key feature points with preset standard feature points.

[0010] Furthermore, the determining the symmetry of the two-dimensional point cloud image according to the slope characteristics of all the straight lines to be analyzed includes:

[0011] The absolute value of the sum of the slopes of all the straight lines to be analyzed is calculated as the slope sum; the variance of the absolute values ​​of the slopes of all the straight lines to be analyzed is calculated as the slope disorder;

[0012] The symmetry of the two-dimensional point cloud image is determined according to the slope sum value and the slope disorder.

[0013] Furthermore, the slope sum value is negatively correlated with the symmetry of the two-dimensional point cloud image, the slope chaos is negatively correlated with the symmetry of the two-dimensional point cloud image, and the symmetry of the two-dimensional point cloud image is a normalized value.

[0014] Furthermore, the determining the screening degree of the tilt angle according to the symmetry between the two-dimensional point cloud image and all the parallel planes includes:

[0015] The mean of the symmetry between the two-dimensional point cloud image and all the parallel planes is calculated as the screening degree corresponding to the tilt angle.

[0016] Further, the selecting the optimal angle according to the screening degree of all the tilt angles includes:

[0017] The inclination angle corresponding to the maximum value of the screening degree is taken as the optimal angle.

[0018] Furthermore, determining the quantity influencing factor according to the quantity of all data points in each two-dimensional point cloud image and parallel planes includes:

[0019] Calculate the sum of the numbers of all data points in the two-dimensional point cloud image and all parallel planes as the total number of influences;

[0020] The ratio of the number of data points in any two-dimensional point cloud image and parallel plane to the total number of influences is calculated as the quantity influence factor.

[0021] Furthermore, the step of extracting features from each two-dimensional point cloud image and parallel plane at the optimal angle according to the quantity influencing factor and the symmetry to obtain key feature points includes:

[0022] Extract the initial feature points of each two-dimensional point cloud image and parallel plane based on SIFT feature extraction algorithm;

[0023] Select any initial feature point as the feature point to be analyzed, and in each two-dimensional point cloud image and parallel plane at the optimal angle, determine the plane closest to the plane where the feature point to be analyzed is located as the adjacent plane;

[0024] On the adjacent plane, determine the initial feature point that is closest to the feature point to be analyzed in the three-dimensional space as the adjacent feature point;

[0025] Connecting the feature point to be analyzed and the adjacent feature points to obtain a feature straight line, and calculating an inversely proportional normalized value of the length of the feature straight line as a feature factor of the feature point to be analyzed;

[0026] Calculating a normalized value of the product of the quantity influence factor and the symmetry as a screening factor;

[0027] Calculating the product of the screening factor and the characteristic factor as a screening coefficient;

[0028] Key feature points are screened out from the initial feature points according to the screening coefficients.

[0029] Further, the filtering out key feature points from the initial feature points according to the filtering coefficients includes:

[0030] The initial feature points whose screening coefficients are greater than a preset screening threshold are taken as key feature points.

[0031] Furthermore, the determining the position information of the clamp by comparing the key feature points with preset standard feature points includes:

[0032] Performing feature analysis on the key feature points based on the SIFT 3D feature extraction algorithm to determine a feature descriptor for each key feature point;

[0033] The feature descriptor and the preset descriptor are matched based on a model matching algorithm to obtain the position information of the clamp, wherein the position information includes an offset distance and an offset direction.

[0034] The present invention also provides a system for determining the position of a clamp, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to implement the steps of the aforementioned method for determining the position of a clamp.

[0035] The present invention has the following beneficial effects:

[0036] Based on the characteristics of the clamp itself that it is vertically downward and symmetrical, the present invention analyzes the three-dimensional point cloud image in the three-dimensional space, divides the three-dimensional point cloud image data into multiple two-dimensional point cloud images, and determines the optimal angle of the clamp according to the symmetry of the multiple two-dimensional point cloud images. The optimal angle can accurately characterize the inclination of the clamp, thereby avoiding the misalignment and displacement of the point cloud data caused by the radar and the clamp being in different horizontal planes, and the clamp being in operation. At the same time, at the optimal angle, according to the number and symmetry of all data points in each two-dimensional point cloud image and parallel plane, feature extraction is performed on each two-dimensional point cloud image and parallel plane to obtain key feature points. Feature points can be extracted more accurately and efficiently, and the influence of abnormal points in the three-dimensional point cloud image on feature point extraction can be avoided. At the same time, the acquisition of key feature points can further improve processing efficiency and ensure the accuracy and timeliness of key feature points. Then, based on the comparison between the key feature points and the preset standard feature points, the position information of the clamp is determined. In summary, the present invention can effectively extract the key feature points in the three-dimensional point cloud image, improve the accuracy and reliability of the key feature point extraction, thereby ensuring the accuracy and timeliness of the position information, and more accurately and efficiently positioning the clamp in real time, thereby improving the image analysis effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 A flow chart of a method for determining the position of a clamp provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a method and system for determining the position of a clamp proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0041] The specific scheme of the method for determining the position of a clamp provided by the present invention is described in detail below with reference to the accompanying drawings.

[0042] See also Figure 1 , which shows a flow chart of a method for determining the position of a clamp provided by an embodiment of the present invention, the method comprising:

[0043] S101: Obtain a three-dimensional point cloud image of a vertically placed clamp, use the coordinate axis corresponding to the vertical direction as a fixed axis, use any data point in the three-dimensional point cloud image as a data point to be measured, adjust the inclination angle on the fixed axis through the data point to be measured, and obtain two-dimensional point cloud images of the data point to be measured corresponding to different inclination angles.

[0044] One application scenario of the present invention can be specifically, for example: automatically positioning the position of the clamp in combination with multiple perspectives. The embodiment of the present invention can configure a laser radar at a certain distance from the clamp to scan and obtain a three-dimensional point cloud image of the clamp and perform positioning. It can be understood that the clamp is normally placed vertically downward, and because of the need to clamp objects, the shape of the clamp usually has a left-right symmetrical feature.

[0045] In an embodiment of the present invention, the coordinate axis corresponding to the vertical direction can be used as a fixed axis. It can be understood that since the shape of the clamp has the characteristic of left-right symmetry, it is also placed vertically, that is, on the front of the clamp, the clamp plane is cut along the vertical direction, and the resulting section has a symmetrical feature. However, in actual scenarios, since the clamp can rotate up and down, left and right, and the laser radar is usually fixed, there is a certain viewing angle difference between the clamp and the laser radar. The existence of this viewing angle difference and the offset of the clamp itself will lead to a decrease in symmetry. Therefore, the position of the clamp can be analyzed based on the viewing angle difference and the symmetrical feature of the section.

[0046] In an embodiment of the present invention, any data point in the three-dimensional point cloud image is used as a data point to be measured, and a vertical line is drawn through the data point to be measured, and the inclination angle of the plane formed by the other two axes is adjusted on the fixed axis. That is to say, the x-axis, y-axis, and z-axis are used as a specific example. The z-axis is used as the fixed axis, and the vertical line is used as the reference line. Any fixed direction is used as the direction with an inclination angle of 0, and a plane with different inclination angles including the vertical line is drawn to obtain a two-dimensional point cloud image corresponding to different inclination angles. The inclination process can be specifically, for example, as follows: the z-axis is fixed, a unit value is increased in the x-axis, and a preset value is increased in the y-axis. The preset value is the value of the corresponding coordinate change of the y-axis at different inclination angles. Different inclination angles are obtained by adjusting the size of the preset value. Thus, the corresponding two-dimensional point cloud image is determined according to different inclination angles.

[0047] S102: Connect all the closest data points on the two-dimensional point cloud image to obtain the straight line to be analyzed between every two closest data points, determine the symmetry of the two-dimensional point cloud image according to the slope characteristics of all the straight lines to be analyzed, obtain at least two parallel planes parallel to the two-dimensional point cloud image, and determine the screening degree of the inclination angle according to the symmetry of the two-dimensional point cloud image and all the parallel planes.

[0048] In the embodiment of the present invention, any two-dimensional point cloud image can be selected for analysis, and all the closest data points on the two-dimensional point cloud image are connected to obtain a straight line to be analyzed between every two closest data points.

[0049] Optionally, in some embodiments of the present invention, the symmetry of the two-dimensional point cloud image is determined based on the slope characteristics of all the straight lines to be analyzed, including: calculating the absolute value of the sum of the slopes of all the straight lines to be analyzed as the slope sum; calculating the variance of the absolute values ​​of the slopes of all the straight lines to be analyzed as the slope chaos; and determining the symmetry of the two-dimensional point cloud image based on the slope sum and the slope chaos.

[0050] Among them, the slope feature can specifically include the numerical feature of the slope and the distribution feature of the slope. Therefore, the present invention calculates the sum of the slopes of all the straight lines to be analyzed on the two-dimensional point cloud image, and obtains the absolute value to obtain the sum of the slopes, which is the numerical feature of the slope; calculates the variance of the absolute values ​​of the slopes of all the straight lines to be analyzed on the two-dimensional point cloud image, and obtains the slope chaos, which is the distribution feature of the slope.

[0051] Among them, symmetry is the symmetry effect of the two-dimensional point cloud image itself. In the embodiment of the present invention, any vertical straight line is used as the symmetry axis, and then the axial symmetry effect of the clamp image on both sides of the symmetry axis is analyzed and used as the symmetry.

[0052] In an embodiment of the present invention, the distribution characteristics of the slope and the numerical characteristics of the slope can effectively perform symmetry analysis on all straight lines to be analyzed on the two-dimensional point cloud image. It can be understood that when the clamp is in the positive direction, the shape of the clamp on the corresponding plane has the highest symmetry. Therefore, the symmetry of the two-dimensional point cloud image can be determined by combining the slope and value and the slope chaos.

[0053] Furthermore, in some embodiments of the present invention, the slope and value are negatively correlated with the symmetry of the two-dimensional point cloud image, the slope chaos is negatively correlated with the symmetry of the two-dimensional point cloud image, and the symmetry of the two-dimensional point cloud image is a normalized value.

[0054] Among them, a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship can be a multiplication relationship, an addition relationship, or the power of an exponential function, which is determined by actual application; a negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtraction relationship, a division relationship, etc., which is determined by actual application.

[0055] That is to say, in some embodiments of the present invention, the inversely proportional normalized value of the product of the slope sum value and the slope chaos can be calculated as the symmetry. The closer the slope sum value is to 0, the more symmetrical the corresponding two-dimensional point cloud image is, that is, the greater the symmetry. When the slope chaos is greater, it can be said that the fluctuation of the slope of the straight line in the two-dimensional point cloud image is more drastic, that is, the more irregular the distribution of the three-dimensional point cloud image is, the lower its credibility is, and the worse the symmetry is.

[0056] In the two-dimensional point cloud image, the embodiment of the present invention can more accurately and simply analyze the symmetry effect of the overall two-dimensional point cloud image by combining the slope characteristics of the straight line formed by adjacent point cloud data, thereby ensuring the accuracy and reliability of symmetry acquisition.

[0057] In an embodiment of the present invention, the acquisition step size can be preset, and then, multiple parallel planes are acquired parallel to the two-dimensional point cloud image. It can be understood that the parallel plane is parallel to the two-dimensional point cloud image. When the two-dimensional point cloud image can be parallel to the positive direction of the clamp, the corresponding parallel plane is also parallel to the positive direction of the clamp, that is, the parallel plane also has a high degree of symmetry. In an embodiment of the present invention, each inclination angle can be analyzed by acquiring the two-dimensional point cloud image of the parallel plane and the symmetry of all parallel planes.

[0058] Furthermore, in some embodiments of the present invention, the screening degree of the tilt angle is determined based on the symmetry between the two-dimensional point cloud image and all parallel planes, including: calculating the mean of the symmetry between the two-dimensional point cloud image and all parallel planes as the screening degree of the corresponding tilt angle.

[0059] Among them, since the symmetry of all parallel planes and the two-dimensional point cloud image is large, it can be said that all parallel planes and the two-dimensional point cloud image are facing the front plane of the clamp. Therefore, in an embodiment of the present invention, the mean of the symmetry of the two-dimensional point cloud image and all parallel planes is calculated as the screening degree of the corresponding inclination angle. The larger the mean, the greater the screening degree, which means that the closer it is to the front plane of the clamp at the corresponding inclination angle.

[0060] Then, the tilt angle can be optimized and analyzed in combination with the screening degree.

[0061] S103: Adjust the tilt angle and select the optimal angle according to the screening degree of all tilt angles; determine the quantity influence factor according to the number of all data points in each two-dimensional point cloud image and parallel plane, and extract features of each two-dimensional point cloud image and parallel plane at the optimal angle according to the quantity influence factor and symmetry to obtain key feature points.

[0062] Furthermore, in some embodiments of the present invention, the optimal angle is selected according to the screening degree of all the inclination angles, including: taking the inclination angle corresponding to the maximum value of the screening degree as the optimal angle.

[0063] That is to say, when the screening degree is the largest, the corresponding representation is closer to the front plane of the clamp at the corresponding inclination angle, which facilitates the positioning and analysis of the clamp according to the front plane, that is, the inclination angle with the largest screening degree is regarded as the optimal angle.

[0064] Furthermore, in some embodiments of the present invention, a quantity influence factor is determined based on the number of all data points in each two-dimensional point cloud image and parallel plane, including: calculating the sum of the numbers of all data points in the two-dimensional point cloud image and all parallel planes as the total number of influences; calculating the ratio of the number of data points in any two-dimensional point cloud image and parallel plane to the total number of influences as the quantity influence factor.

[0065] Since the point cloud data is basically evenly distributed, when the number of data points in any plane in the two-dimensional point cloud image and all parallel planes is large, it can be said that there are more point cloud data related to the clamp contained therein, that is, the corresponding plane is more important. Therefore, the present invention uses the number of data points on each plane as the weight of each plane, and normalizes the number of data points on each plane. The normalization process can be specifically calculated as the ratio of the number of data points in any two-dimensional point cloud image and parallel plane to the total number of influences, that is, the quantity influence factor is a normalized value.

[0066] Furthermore, in an embodiment of the present invention, feature extraction is performed on each two-dimensional point cloud image and parallel plane at an optimal angle according to the quantity influence factor and symmetry to obtain key feature points, including: extracting initial feature points of each two-dimensional point cloud image and parallel plane based on the SIFT feature extraction algorithm; randomly selecting a certain initial feature point as the feature point to be analyzed, and in each two-dimensional point cloud image and parallel plane at the optimal angle, determining the plane closest to the plane where the feature point to be analyzed is located as an adjacent plane; on the adjacent plane, determining the initial feature point closest to the feature point to be analyzed in three-dimensional space as an adjacent feature point; connecting the feature point to be analyzed and the adjacent feature points to obtain a feature straight line, and calculating the inversely proportional normalized value of the length of the feature straight line as the feature factor of the feature point to be analyzed; calculating the normalized value of the product of the quantity influence factor and the symmetry as a screening factor; calculating the product of the screening factor and the feature factor as a screening coefficient; and screening out key feature points from the initial feature points according to the screening coefficient.

[0067] Among them, the Scale Invariant Feature Transform (SIFT) algorithm is a commonly used feature extraction algorithm. The SIFT algorithm can extract feature points in a plane image. It can be understood that the SIFT algorithm is a well-known algorithm in the art and will not be further elaborated or limited.

[0068] In an embodiment of the present invention, based on the SIFT algorithm, initial feature points in the corresponding plane can be extracted from each two-dimensional point cloud image and parallel plane respectively, and then, an initial feature point is randomly selected from all the initial feature points as the feature point to be analyzed, and the plane closest to the plane where the feature point to be analyzed is located is determined as the adjacent plane.

[0069] In an embodiment of the present invention, on an adjacent plane, an initial feature point that is closest to the feature point to be analyzed in three-dimensional space is determined as an adjacent feature point. That is to say, for example, in an embodiment of the present invention, a feature point n to be analyzed can be selected from a two-dimensional point cloud image and a parallel plane, and the plane where point n is located is taken as N, where N represents the index of the plane. In an embodiment of the present invention, a plane can be selected from the N+1th plane and the N-1th plane as an adjacent plane. The present invention uses the N+1th plane as the adjacent plane, and then selects the feature point that is closest to the feature point n to be analyzed from the N+1th plane as the adjacent feature point. Of course, in other embodiments of the present invention, the adjacent feature points corresponding to the N+1th plane and the N-1th plane can also be calculated separately, and analyzed in combination with these two adjacent feature points, and there is no limitation on this.

[0070] Among them, in a normal three-dimensional point cloud image, the closest distance means that the distribution is relatively similar. In the embodiment of the present invention, the feature points with relatively similar distribution are analyzed. It can be understood that abnormal data in the three-dimensional point cloud image usually leads to a disordered distribution of data points in the region. Therefore, in an abnormal situation, the feature points may be offset due to the disordered distribution, that is, the length of the characteristic straight line obtained by connecting the feature point to be analyzed and the adjacent feature point is longer, while in a normal situation, the distribution of feature points is more regular, that is, the length of the obtained characteristic straight line is shorter. Therefore, based on this rule, the embodiment of the present invention analyzes the feature point to be analyzed through the length of the characteristic straight line to obtain a characteristic factor. Since the longer the length of the characteristic straight line, the greater the degree of abnormality, the embodiment of the present invention calculates the inverse proportional normalized value of the length of the characteristic straight line as the characteristic factor.

[0071] Among them, the characteristic factor can represent the abnormality degree of the corresponding feature point to be analyzed. The smaller the characteristic factor, the more likely the corresponding feature point to be analyzed is to be offset due to abnormal noise, and the larger the characteristic factor, the smaller the corresponding abnormality degree.

[0072] In the embodiment of the present invention, the normalized value of the product of the quantity influence factor and the symmetry is calculated as the screening factor. Since the greater the symmetry, the higher the symmetry of the corresponding two-dimensional point cloud image and the parallel plane, and the greater the quantity influence factor, the higher the importance of the corresponding two-dimensional point cloud image and the parallel plane, therefore, in the embodiment of the present invention, the normalized value of the product of the quantity influence factor and the symmetry is used as the screening factor. The larger the screening factor, the more important the corresponding two-dimensional point cloud image and the parallel plane are, and the higher the importance of the initial feature points in the corresponding plane.

[0073] The present invention combines the screening factor and the characteristic factor to obtain the screening coefficient, wherein the screening coefficient is the screening weight coefficient of the feature point to be analyzed. It can be understood that the screening factor is the screening factor of the plane where the feature point to be analyzed is located, and the characteristic factor is the characteristic value of the feature point to be analyzed itself. By combining the screening factor and the characteristic factor, data analysis can be performed based on the overall characteristics of the entire three-dimensional point cloud image, thereby ensuring the reliability of the screening coefficient and facilitating the effective screening of the initial feature points according to the screening coefficient.

[0074] Furthermore, in some embodiments of the present invention, selecting key feature points from initial feature points according to a screening coefficient includes: taking initial feature points whose screening coefficients are greater than a preset screening threshold as key feature points.

[0075] Among them, the preset screening threshold is the threshold value of the screening coefficient. In the embodiment of the present invention, the preset screening threshold can be set to 0.8, that is, when the screening coefficient of the initial feature point is greater than 0.8, the initial feature point is used as a key feature point.

[0076] In the embodiment of the present invention, by screening the initial feature points through screening coefficients, abnormal initial feature points in the three-dimensional point cloud image can be effectively avoided. At the same time, the number of feature points can be further reduced, thereby improving processing efficiency while ensuring the effect of feature point analysis.

[0077] S104: Determine the position information of the clamp by comparing the key feature points with the preset standard feature points.

[0078] Furthermore, in some embodiments of the present invention, the position information of the clamp is determined by comparing the key feature points with preset standard feature points, including: performing feature analysis on the key feature points based on the SIFT 3D feature extraction algorithm to determine the feature descriptor of each key feature point; matching the feature descriptor with the preset descriptor based on the model matching algorithm to obtain the position information of the clamp, wherein the position information includes the offset distance and the offset direction.

[0079] In an embodiment of the present invention, feature analysis is performed on all key feature points obtained as a whole through the SIFT 3D feature extraction algorithm. It can be understood that the SIFT 3D feature extraction algorithm is the application of the SIFT feature extraction algorithm in a three-dimensional stereoscopic scene. It can be understood that since different two-dimensional point cloud images and parallel planes have corresponding key feature points respectively, that is, the distribution of key feature points is a stereoscopic distribution in a three-dimensional scene. Therefore, the present invention uses the SIFT 3D feature extraction algorithm for processing. The SIFT 3D feature extraction algorithm is an algorithm well known in the art and will not be further limited or elaborated on based on the SIFT 3D feature extraction algorithm.

[0080] Among them, the preset descriptor is a descriptor of a preset standard three-dimensional point cloud image. In an embodiment of the present invention, a SIFT 3D feature extraction algorithm can be used to extract feature points of a standard three-dimensional point cloud image in a pre-stored clamp model library, and the feature descriptor of the feature points in the standard three-dimensional point cloud image can be used as the preset descriptor.

[0081] In the embodiment of the present invention, the feature descriptor of the key feature point can be used to perform matching analysis with the preset descriptor. According to the matching results, the best matching situation is selected, and the offset distance and offset direction of the clamp corresponding to the preset descriptor in the best matching situation are used as the position information of the three-dimensional point cloud image to be determined. The matching process of the feature descriptor and the preset descriptor is a technology well known in the art, that is, the data difference between the feature descriptor and the preset descriptor at the same spatial position is comprehensively calculated, which will not be further described. Subsequently, the position information can be combined with the position data of the clamp obtained by other sensors for statistics, and the clamp can be positioned according to the statistical results.

[0082] In an embodiment of the present invention, based on the vertically downward and symmetrical characteristics of the clamp itself, the three-dimensional point cloud image data is analyzed in the three-dimensional space, the three-dimensional point cloud image data is divided into multiple two-dimensional point cloud images, and the optimal angle of the clamp is determined according to the symmetry of the multiple two-dimensional point cloud images. The optimal angle can accurately characterize the inclination of the clamp, thereby avoiding the misalignment and displacement of the point cloud data caused by the laser radar and the clamp being in different horizontal planes, and the clamp being in operation. At the same time, at the optimal angle, according to the number and symmetry of all data points in each two-dimensional point cloud image and parallel plane, feature extraction is performed on each two-dimensional point cloud image and parallel plane to obtain key feature points, which can extract feature points more accurately and efficiently and avoid the influence of abnormal points in the three-dimensional point cloud image on the feature point extraction. At the same time, the acquisition of key feature points can further improve the processing efficiency and ensure the accuracy and timeliness of the key feature points. Then, based on the comparison between the key feature points and the preset standard feature points, the position information of the clamp is determined. In summary, the present invention can effectively extract the key feature points in the three-dimensional point cloud image, improve the accuracy and reliability of the key feature point extraction, thereby ensuring the accuracy and timeliness of the position information, and more accurately and efficiently positioning the clamp in real time, thereby improving the image analysis effect.

[0083] The present invention also provides a system for determining the position of a clamp, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to implement the steps of a method for determining the position of a clamp as described above.

[0084] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0085] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for determining the position of a clamp, characterized in that: The method comprises: Acquire a three-dimensional point cloud image of the vertically placed clamp, use the coordinate axis corresponding to the vertical direction as a fixed axis, use any data point in the three-dimensional point cloud image as a data point to be measured, adjust the inclination angle on the fixed axis through the data point to be measured, and obtain two-dimensional point cloud images of the data point to be measured corresponding to different inclination angles; Connecting all the closest data points on the two-dimensional point cloud image to obtain a straight line to be analyzed between every two closest data points, determining the symmetry of the two-dimensional point cloud image according to the slope characteristics of all the straight lines to be analyzed, obtaining at least two parallel planes parallel to the two-dimensional point cloud image, and determining the screening degree of the inclination angle according to the symmetry of the two-dimensional point cloud image and all the parallel planes; The tilt angle is adjusted, and the optimal angle is selected according to the screening degree of all tilt angles; the quantity influence factor is determined according to the number of all data points in each two-dimensional point cloud image and parallel plane, and the feature extraction is performed on each two-dimensional point cloud image and parallel plane at the optimal angle according to the quantity influence factor and symmetry to obtain key feature points; Determine the position information of the clamp by comparing the key feature points with preset standard feature points; Determining the symmetry of the two-dimensional point cloud image according to the slope characteristics of all the straight lines to be analyzed includes: The absolute value of the sum of the slopes of all the straight lines to be analyzed is calculated as the slope sum; the variance of the absolute values ​​of the slopes of all the straight lines to be analyzed is calculated as the slope disorder; Determining the symmetry of the two-dimensional point cloud image according to the slope sum value and the slope disorder; The slope sum value is negatively correlated with the symmetry of the two-dimensional point cloud image, the slope chaos is negatively correlated with the symmetry of the two-dimensional point cloud image, and the symmetry of the two-dimensional point cloud image is a normalized value; wherein the inversely proportional normalized value of the product of the slope sum value and the slope chaos is calculated as the symmetry; Determining the screening degree of the tilt angle according to the symmetry between the two-dimensional point cloud image and all the parallel planes includes: Calculating the mean of the symmetry between the two-dimensional point cloud image and all the parallel planes as the screening degree corresponding to the tilt angle; Determining the quantity influencing factor according to the quantity of all data points in each two-dimensional point cloud image and parallel planes includes: Calculate the sum of the numbers of all data points in the two-dimensional point cloud image and all parallel planes as the total number of influences; Calculate the ratio of the number of data points in any two-dimensional point cloud image and parallel plane to the total number of influences as the quantity influence factor; The method of selecting the optimal angle according to the screening degree of all tilt angles includes: The tilt angle corresponding to the maximum value of the screening degree is taken as the optimal angle; The method of extracting features from each two-dimensional point cloud image and parallel plane at the optimal angle according to the quantity influencing factor and the symmetry to obtain key feature points includes: Extract the initial feature points of each two-dimensional point cloud image and parallel plane based on SIFT feature extraction algorithm; Select any initial feature point as the feature point to be analyzed, and in each two-dimensional point cloud image and parallel plane at the optimal angle, determine the plane closest to the plane where the feature point to be analyzed is located as the adjacent plane; On the adjacent plane, determine the initial feature point that is closest to the feature point to be analyzed in the three-dimensional space as the adjacent feature point; Connecting the feature point to be analyzed and the adjacent feature points to obtain a feature straight line, and calculating an inversely proportional normalized value of the length of the feature straight line as a feature factor of the feature point to be analyzed; Calculating a normalized value of the product of the quantity influence factor and the symmetry as a screening factor; Calculating the product of the screening factor and the characteristic factor as a screening coefficient; Filtering key feature points from the initial feature points according to the filtering coefficient; The step of comparing the key feature points with preset standard feature points to determine the position information of the clamp includes: Performing feature analysis on the key feature points based on the SIFT 3D feature extraction algorithm to determine a feature descriptor for each key feature point; The feature descriptor and the preset descriptor are matched based on a model matching algorithm to obtain the position information of the clamp, wherein the position information includes an offset distance and an offset direction.

2. A method for determining the position of a clamp according to claim 1, characterized in that: The step of selecting key feature points from the initial feature points according to the selection coefficients includes: The initial feature points whose screening coefficients are greater than a preset screening threshold are taken as key feature points.

3. A clamp position determination system, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: The processor executes the computer program to implement the steps of a method for determining the position of a clamp as described in any one of claims 1 to 2.