Error calibration method and system for three-dimensional laser scanner

Abstract

The invention relates to the technical field of image correction, in particular to an error calibration method and system for a three-dimensional laser scanner, and the method comprises the steps: obtaining three-dimensional point cloud data in each scanning direction through the three-dimensional laser scanner; any three-dimensional point cloud number is subjected to pooling processing, a plurality of boundary feature curves in pooled point cloud data are obtained, the pooled point cloud data are screened to obtain target point cloud data, reference point clouds are further screened out, the positions of feature points in the reference point clouds are adjusted, adjacent feature points are combined to obtain an adjustment trend consistent area, and the adjustment trend consistent area is obtained; and carrying out movement adjustment on data points except the feature points in the reference point clouds by utilizing the overall adjustment condition of the feature points in the adjustment trend consistent region, carrying out position adjustment on all the data points in all the reference point clouds to obtain new reference point clouds, and carrying out registration splicing on all the new reference point clouds. According to the invention, the accuracy of the three-dimensional point cloud data collected by the three-dimensional laser scanner is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image correction, and in particular to an error calibration method and system for a three-dimensional laser scanner. Background Art

[0002] Error calibration in point clouds is crucial in the registration process, as it directly impacts registration accuracy and reliability. By calibrating systematic errors (such as scanner bias) and random errors (such as environmental noise), the convergence of registration algorithms (such as ICP) can be optimized, preventing misalignment caused by error accumulation. Especially in multi-station scanning or large scenes, an accurate error model improves the consistency of point cloud stitching, ensuring the accuracy of subsequent modeling and analysis.

[0003] Traditional methods match and splice point clouds based on the matching degree of point cloud feature curves at multiple angles, without considering the distortion of the point cloud itself during scanning, and the acquisition of the final point cloud of the object is not accurate enough. Summary of the Invention

[0004] The present invention provides an error calibration method and system for a three-dimensional laser scanner to solve the existing problems.

[0005] The error calibration method and system for a three-dimensional laser scanner of the present invention adopt the following technical solutions: One embodiment of the present invention provides an error calibration method for a three-dimensional laser scanner, the method comprising the following steps: Scan the object with a three-dimensional laser scanner in several scanning directions to obtain three-dimensional point cloud data in each scanning direction; Constructing several 3D pooling windows of different sizes to pool any 3D point cloud data, obtaining the pooled point cloud data and several boundary feature curves composed of several feature points in the pooled point cloud data. Analyzing the consistency of the changing trends between the boundary feature curves to obtain the matching relationship between the boundary feature curves, based on the matching relationship between the boundary feature curves in the 3D point cloud data of all scanning directions, the pooled point cloud data is filtered to obtain the target point cloud data. The overall distribution level of trend consistency corresponding to the boundary feature curves with matching relationships in different target point cloud data is used to screen out reference curves. The reference point cloud is screened based on the number and length of reference curves in the target point cloud data. The position of the feature points in the reference point cloud is adjusted based on the matching relationship between the boundary feature curves. Merge adjacent feature points based on their position changes before and after adjustment to obtain several regions with consistent adjustment trends. Use the overall adjustment of feature points in these regions to move and adjust data points other than feature points in the reference point cloud. After adjusting the positions of all data points in all reference point clouds, a new reference point cloud is obtained. All new reference point clouds are registered and stitched.

[0006] Furthermore, the method of constructing a plurality of three-dimensional pooling windows of different sizes to perform pooling processing on any three-dimensional point cloud data to obtain pooled point cloud data and a plurality of boundary feature curves composed of a plurality of feature points in the pooled point cloud data includes the following specific methods: The default side length is The three-dimensional pooling window is used to slide the three-dimensional point cloud data. The three-dimensional point cloud data is downsampled by the maximum pooling operation during the sliding traversal process. The maximum pooling operation of the three-dimensional pooling window with any side length is used as a pooling method to obtain the pooled point cloud data under any pooling method and several feature points contained in the pooled point cloud data. The side length of the three-dimensional pooling window is , and the size of the three-dimensional pooling window is in the interval Internal As the step size increases, the sliding compensation of the three-dimensional pooling window during the sliding traversal process is ,in 、 、 as well as are the preset first side length parameter, second side length parameter, first step length parameter, and second step length parameter respectively; Under any pooling mode, the distance between feature points is obtained, and the feature points whose distance is less than a preset distance threshold are connected to each other to obtain several boundary feature curves under the corresponding pooling mode.

[0007] Furthermore, the specific method for obtaining the degree of consistency of the change trends between the boundary characteristic curves is: For any two three-dimensional point cloud data with different scanning directions, the shortest alignment path between any two boundary feature curves in the two three-dimensional point cloud data and the correspondence between the feature points in the two boundary feature curves are obtained by the DDTW algorithm. For any one of the two boundary feature curves, it is recorded as the first target boundary feature curve, and the other boundary feature curve is recorded as the second target boundary feature curve. The number of feature points on the first target boundary feature curve that have a corresponding relationship on the second target boundary feature curve is counted and recorded as the matching parameter. The number of feature points in the two boundary feature curves whose matching parameters are greater than the preset first parameter is recorded as the overmatching parameter between the two boundary feature curves. According to the length of the shortest alignment path between the two boundary feature curves and the overmatching parameter, the degree of consistency of the change trend between the two boundary feature curves is obtained, wherein the length of the shortest alignment path and the overmatching parameter are both negatively correlated with the degree of consistency of the change trend.

[0008] Furthermore, the target point cloud data is obtained by screening the pooled point cloud data based on the matching relationship between the boundary feature curves in the three-dimensional point cloud data in all scanning directions, including the specific method of: When the degree of consistency of the change trend is greater than a preset consistency threshold, the corresponding second target boundary feature curve is used as the matching curve of the corresponding first target boundary feature curve; under any pooling method, the number of matching curves of any boundary feature curve in the three-dimensional point cloud data in any scanning direction is obtained, and recorded as the matching amount of the boundary feature curve; and the total number of feature points in all three-dimensional point cloud data under any pooling method is counted; Calculate the pooling suitability under any pooling mode according to the side length of the corresponding pooling window, the matching amount of all boundary feature curves in the three-dimensional point cloud data in all scanning directions, and the total number of feature points in the three-dimensional point cloud data in all scanning directions; the side length of the pooling window and the matching amount of all boundary feature curves in the three-dimensional point cloud data in all scanning directions are positively correlated with the pooling suitability, and the total number of feature points in the three-dimensional point cloud data in all scanning directions is negatively correlated with the pooling suitability; The pooling mode corresponding to the maximum pooling fitness under all pooling modes is obtained, recorded as the target pooling mode, and the pooled point cloud data obtained under the target pooling mode is recorded as the target point cloud data.

[0009] Furthermore, the reference curve is screened out by utilizing the overall distribution level of trend consistency corresponding to the boundary feature curves having matching relationships in different target point cloud data, including the specific method of: For target point cloud data in any scanning direction, when any boundary feature curve in the target point cloud data has a matching curve in the target point cloud data in all other scanning directions, the number of corresponding target point cloud data is obtained, and recorded as the matching inclusion amount of the boundary feature curve; the target point cloud data in any scanning direction is recorded as the first point cloud data, and any boundary feature area in the first point cloud data is recorded as the first boundary feature curve; the maximum trend consistency of all matching curves of the first boundary feature curve in any target point cloud data other than the first point cloud data is obtained, and recorded as the maximum trend consistency of the first boundary feature curve in any target point cloud data other than the first point cloud data; the average value of the maximum trend consistency of the first boundary feature curve in any target point cloud data other than the first point cloud data is obtained, and recorded as the consistency parameter of the first boundary feature curve; according to the matching inclusion amount and the consistency parameter of any boundary feature curve, the true conformity of the boundary feature curve is calculated, wherein the matching inclusion amount and the consistency parameter are both positively correlated with the true conformity; When the true conformity is greater than the preset second parameter, the corresponding boundary characteristic curve is recorded as a reference curve.

[0010] Furthermore, the method of selecting the reference point cloud by the number and length of the reference curves in the target point cloud data includes the following specific methods: Obtain the number of reference curves contained in the target point cloud data of any scanning direction, and the number of feature points contained in all the reference curves in the target point cloud data; calculate the reference coefficient of the target point cloud data in the scanning direction based on the number of reference curves in the target point cloud data of the scanning direction and the proportion of the feature points contained in all the reference curves in the target point cloud data in all the target point cloud data, wherein the number of reference curves and the proportion of the feature points are positively correlated with the reference coefficient; and use the target point cloud data with the largest reference coefficient in the target point cloud data of all the scanning directions as the reference point cloud.

[0011] Furthermore, the position adjustment of the feature points in the reference point cloud based on the matching relationship between the boundary feature curves includes the following specific methods: Obtain the maximum value of the adjustment coefficients of the boundary feature curve and all the corresponding matching curves, use the matching point in the matching curve corresponding to the maximum adjustment coefficient as the reference point of the feature point corresponding to the matching point in the boundary feature curve, and adjust the position of the feature point in the boundary feature curve according to the relative position relationship between the reference points in the matching curve corresponding to the boundary feature curve, so that the relative position relationship between the feature points in the boundary feature curve is the same as the relative position relationship between the reference points in the corresponding matching curve, and record the Euler angle before and after the position change of the feature point in the boundary feature curve and the distance before and after the position change, which are respectively recorded as the adjusted Euler angle and the adjusted distance.

[0012] Furthermore, the method of merging adjacent feature points based on the changes before and after the feature point positions are adjusted to obtain several regions with consistent adjustment trends includes: For the reference point cloud after the position of the feature points in the boundary feature curve is adjusted, the absolute value of the difference in the adjusted distance between adjacent feature points in the reference point cloud and the sum of the absolute value of the difference in the adjusted Euler angle are obtained, which are respectively recorded as the first difference and the second difference. According to the first difference and the second difference, the merging feasibility between adjacent feature points is obtained. The first difference and the second difference are both negatively correlated with the merging feasibility. When the merging feasibility is greater than or equal to the preset fourth parameter, the corresponding two feature points are merged. After merging all adjacent feature points whose merging feasibility is greater than or equal to the fourth parameter, several areas with consistent adjustment trends are obtained.

[0013] Furthermore, the overall adjustment of the feature points in the region with the same adjustment trend is used to move and adjust the data points other than the feature points in the reference point cloud, including the following specific methods: For any consistent adjustment trend region, a feature point with the largest adjustment distance in the consistent adjustment trend region is obtained and recorded as a main feature point. The adjustment distance of the main feature point is used as the adjustment length of the consistent adjustment trend region to which it belongs. The adjusted Euler angle of the main feature point is used as the adjustment scanning direction of the consistent adjustment trend region to which it belongs. Data points other than feature points in the reference point cloud are recorded as non-feature points. The consistent adjustment trend region closest to any non-feature point is obtained and recorded as the target region of the non-feature point. For any non-feature point, the midpoint of the boundary feature curve closest to the non-feature point in the target region is obtained. A straight line passing through the midpoint and having a scanning direction in the adjusted scanning direction of the target region is used as the adjustment trend line of the target region to which it belongs. The angle between the line connecting the non-feature point and the midpoint and the adjustment trend line is obtained and recorded as a first adjustment parameter of the non-feature point. The minimum distance between the non-feature point and the adjustment trend line is obtained and recorded as a second adjustment parameter of the non-feature point. The adjustment influence of the non-feature point is obtained based on the first adjustment parameter and the second adjustment parameter. The movement distance of the non-feature point along the adjusted scanning direction of the corresponding target region is obtained by combining the adjustment length of the target region of the non-feature point and the adjustment influence of the non-feature point.

[0014] An error calibration system for a three-dimensional laser scanner includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the error calibration method for a three-dimensional laser scanner are implemented.

[0015] The beneficial effects of the technical solution of the present invention are: by analyzing the changing trends between the boundary feature curves, accurate matching of feature points can be achieved, thereby optimizing the subsequent point cloud data alignment and splicing process; in addition, by utilizing the matching relationship of the boundary feature curves, a reference curve consistent with the overall trend is selected, thereby improving the selection accuracy of the reference point cloud data, and being able to perform effective feature point position adjustment; and by utilizing the adjustment of the feature point positions and the merging of trend-consistent areas, the matching accuracy of adjacent feature points can be effectively improved, making the final reference point cloud more accurate, thereby improving the point cloud data alignment effect; further, by adjusting the positions of all data points in the reference point cloud, the global consistency of all data points is ensured, and the new reference point cloud obtained will be more real and accurate, thereby improving the accuracy of the three-dimensional point cloud data collected by the three-dimensional laser scanner. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 any creative work.

[0017] Figure 1 This is a flowchart of the steps of the error calibration method for a three-dimensional laser scanner of the present invention. DETAILED DESCRIPTION

[0018] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of the error calibration method and system for a 3D laser scanner proposed in accordance with the present invention. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0019] 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.

[0020] The specific scheme of the error calibration method and system for a three-dimensional laser scanner provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] See also Figure 1 , which shows a flowchart of a method for error calibration of a three-dimensional laser scanner provided by an embodiment of the present invention, the method comprising the following steps: Step S001: Scan an object using a three-dimensional laser scanner in several scanning directions to obtain three-dimensional point cloud data in each scanning direction.

[0022] It should be noted that calibrating errors in the point cloud is crucial in the registration process, as it directly impacts registration accuracy and result reliability. By calibrating systematic errors (such as inherent scanner bias) and random errors (such as environmental noise), the convergence of registration algorithms (such as ICP) can be optimized, avoiding misalignment due to error accumulation. Especially in multi-station scanning or large scenes, an accurate error model can improve the consistency of point cloud stitching and ensure the accuracy of subsequent modeling and analysis. Lack of error calibration can lead to registration failure or error propagation, affecting overall data quality. The present invention determines the most data-efficient downsampling method based on the matching of point clouds at different angles under different downsampling conditions. A reference point cloud is selected based on the degree of match between the boundary feature curves of a single-angle point cloud and those of other point clouds, and the positions of the boundary feature points in the reference point cloud are adjusted. The positions of other points in the point cloud are adjusted based on the adjusted performance of the feature points in the boundary feature curves, allowing for accurate stitching of multi-angle point clouds.

[0023] Specifically, in order to implement the error calibration method for a 3D laser scanner proposed in this embodiment, it is first necessary to collect 3D point cloud data. The specific process is as follows: First, set up several scanning points and install a 3D laser scanner at each scanning point. Scan the object with the 3D laser scanner. During the scanning process, ensure that there is at least The above overlaps to obtain corresponding three-dimensional point cloud data, and the three-dimensional point cloud data is denoised by statistical filtering, where A is a preset scale parameter.

[0024] It should be noted that the ratio parameter A is preset to 30 based on experience and can be adjusted according to actual conditions, and is not specifically limited in the embodiment of the present invention.

[0025] Then, the preset side length is The three-dimensional pooling window is used to perform convolution processing on the three-dimensional point cloud data, perform local area product summation, and follow the convolution with a nonlinear activation function to enhance the model's expression ability and obtain the feature map of the convolution layer.

[0026] It should be noted that the side length of the three-dimensional pooling window is preset based on experience. is 3, which can be adjusted according to actual conditions; in addition, the nonlinear activation function described in the embodiment of the present invention may include at least ReLU and PReLU, which can be adjusted according to actual conditions and is not specifically limited in the embodiment of the present invention.

[0027] Finally, the feature map is max-pooled through a multi-level pooling window, and the pooling results of different levels are concatenated according to the channel dimension to form a feature vector of fixed length. The fixed-length feature vector is input into the fully connected layer to obtain the feature points.

[0028] At this point, three-dimensional point cloud data and several feature points in the three-dimensional point cloud data are obtained through the above method.

[0029] Step S002: Construct several three-dimensional pooling windows of different sizes to pool any three-dimensional point cloud data respectively, obtain pooled point cloud data and several boundary feature curves composed of several feature points in the pooled point cloud data, analyze the consistency of the change trends between the boundary feature curves to obtain the matching relationship between the boundary feature curves, and based on the matching relationship between the boundary feature curves in the three-dimensional point cloud data of all scanning directions, filter the pooled point cloud data to obtain the target point cloud data.

[0030] It should be noted that when directly performing point cloud registration on an object's original 3D point cloud acquired from different angles, the system requires a significant amount of computing power to match the large number of feature points contained in the 3D point cloud. To reduce point cloud registration time, downsampling can be used to reduce the number of feature points in the 3D point cloud.

[0031] Specifically, in step S201, for the three-dimensional point cloud data under any scanning direction, several three-dimensional pooling windows with different side lengths are constructed, and the three-dimensional point cloud data are pooled respectively to obtain the pooled point cloud data under the three-dimensional pooling windows with corresponding side lengths and several boundary feature curves composed of several feature points in the pooled point cloud data, and analyze the degree of consistency of the change trends between the boundary feature curves obtained by the pooling windows with the same side length under any two scanning directions.

[0032] First, the default side length is The three-dimensional pooling window is used to slide the three-dimensional point cloud data. The three-dimensional point cloud data is downsampled by the maximum pooling operation during the sliding traversal process. The maximum pooling operation of the three-dimensional pooling window with any side length is used as a pooling method to obtain the pooled point cloud data under any pooling method and several feature points contained in the pooled point cloud data. The side length of the three-dimensional pooling window is , and the size of the three-dimensional pooling window is in the interval Internal As the step size increases, the sliding compensation of the three-dimensional pooling window during the sliding traversal process is ,in 、 、 as well as They are the preset first side length parameter, second side length parameter, first step length parameter and second step length parameter respectively.

[0033] It should be noted that the first side length parameter, the second side length parameter, the first step length parameter and the second step length parameter are preset to 3, 15, 1 and 1 respectively based on experience. They can be adjusted according to actual conditions and are not specifically limited in the embodiment of the present invention.

[0034] Then, under any pooling mode, the distance between feature points is obtained, and feature points whose distance is less than a preset distance threshold are connected to each other to obtain a number of boundary feature curves under the corresponding pooling mode.

[0035] It should be noted that in the process of obtaining the boundary characteristic curve, the distance between multiple feature points in the local area may be less than the distance threshold, resulting in the appearance of a non-single-chain graph structure in the boundary characteristic curve, making it impossible to use the DDTW algorithm for subsequent analysis. Therefore, in an embodiment of the present invention, for the boundary characteristic curve with a non-single-chain structure, the feature points in the non-single-chain structure where the number of connected feature points is greater than 2 are used as breakpoints, thereby segmenting the boundary characteristic curve to avoid the appearance of a non-single-chain structure in the boundary characteristic curve.

[0036] It should be noted that, in the embodiment of the present invention, the side length of the three-dimensional pooling window under the corresponding pooling mode is combined with the side length of the three-dimensional pooling window under the corresponding pooling mode. , based on experience, the distance threshold is preset to , and its value can be adjusted according to actual conditions, and is not specifically limited in the embodiment of the present invention.

[0037] Finally, for any two three-dimensional point cloud data with different scanning directions, the shortest alignment path between any two boundary feature curves in the two three-dimensional point cloud data and the correspondence between the feature points in the two boundary feature curves are obtained by the DDTW algorithm. For any one of the two boundary feature curves, it is recorded as the first target boundary feature curve, and the other boundary feature curve is recorded as the second target boundary feature curve. The number of feature points on the first target boundary feature curve that have a corresponding relationship on the second target boundary feature curve is counted and recorded as the matching parameter. The number of feature points in the two boundary feature curves with a matching parameter greater than the preset first parameter is recorded as the over-matching parameter between the two boundary feature curves. According to the length of the shortest alignment path between the two boundary feature curves and the over-matching parameter, the degree of consistency of the change trend between the two boundary feature curves is obtained, wherein the length of the shortest alignment path and the over-matching parameter are both negatively correlated with the degree of consistency of the change trend.

[0038] It should be noted that the DDTW (Derivative Dynamic Time Warping) algorithm is an existing sequence analysis algorithm, and therefore will not be described in detail in the embodiments of the present invention.

[0039] It should also be noted that, in the embodiment of the present invention, the first parameter is preset to 5 based on experience, and it can be adjusted according to actual conditions, and is not specifically limited in the embodiment of the present invention.

[0040] As an optional embodiment, a specific method for calculating the degree of consistency of the change trends between any two boundary characteristic curves is as follows: in, Represents the boundary characteristic curve and boundary characteristic curve The degree of consistency of the changing trends between Represents the boundary characteristic curve and boundary characteristic curve Overmatch parameters between ; Represents the boundary characteristic curve and boundary characteristic curve The length of the shortest alignment path between Represents the sigmoid normalization function.

[0041] It should be noted that when the overmatch parameter between the two boundary characteristic curves is smaller and the length of the shortest alignment path is shorter, the change trends of the two boundary characteristic curves are more consistent, and the corresponding degree of consistency of the change trends is greater.

[0042] In step S202, under any pooling method, the matching relationship between the boundary feature curves in the pooled point cloud data under any two scanning directions is obtained based on the consistency of the change trend, and the number of boundary feature curves with matching relationships in the three-dimensional point cloud data of all scanning directions is counted, and the pooling suitability is calculated in combination with the number of feature points in the three-dimensional point cloud data of all scanning directions to filter out the target point cloud data.

[0043] First, when the degree of consistency of the change trend is greater than a preset consistency threshold, the corresponding second target boundary feature curve is used as the matching curve of the corresponding first target boundary feature curve; under any pooling method, the number of matching curves of any boundary feature curve in the three-dimensional point cloud data in any scanning direction is obtained, which is recorded as the matching amount of the boundary feature curve; and the total number of feature points in all three-dimensional point cloud data under any pooling method is counted.

[0044] Then, the pooling suitability under any pooling method is calculated based on the side length of the corresponding pooling window, the matching amount of all boundary feature curves in the three-dimensional point cloud data of all scanning directions, and the total number of feature points in the three-dimensional point cloud data of all scanning directions; the side length of the pooling window and the matching amount of all boundary feature curves in the three-dimensional point cloud data of all scanning directions are positively correlated with the pooling suitability, and the total number of feature points in the three-dimensional point cloud data of all scanning directions is negatively correlated with the pooling suitability.

[0045] As an optional embodiment, the specific calculation method of the pooling suitability is: in, Indicates that the window side length is When , the pooling suitability of the corresponding pooling method; Indicates that the window side length is When , it corresponds to the matching amount of all boundary feature curves in the 3D point cloud data of all scanning directions under the pooling method; Indicates that the window side length is When , it corresponds to the total number of feature points in the 3D point cloud data of all scanning directions under the pooling method; Indicates the side length of the 3D pooling window.

[0046] It should be noted that when the pooling window size is When , the larger the matching amount of all boundary feature curves and the total number of feature points in the three-dimensional point cloud data of all scanning directions under the corresponding pooling method, the smaller the total number of feature points in the three-dimensional point cloud data of all scanning directions under the corresponding pooling method, and the larger the value of a corresponding to the pooling window, this pooling method can not only ensure the correspondence between the three-dimensional point cloud data of different scanning directions, but also reduce the feature points used for analysis in the three-dimensional point cloud data, and is the most suitable pooling method.

[0047] Finally, the pooling method corresponding to the maximum pooling fitness under all pooling methods is obtained, recorded as the target pooling method, and the pooled point cloud data obtained under the target pooling method is recorded as the target point cloud data.

[0048] At this point, the target point cloud data is obtained through the above method.

[0049] Step S003: Filter out reference curves using the overall distribution level of trend consistency corresponding to the boundary feature curves with matching relationships in different target point cloud data, and filter out reference point clouds based on the number and length of reference curves in the target point cloud data; adjust the positions of feature points in the reference point clouds based on the matching relationships between the boundary feature curves.

[0050] It should be noted that the distance and angle measurement accuracy of 3D laser scanners may cause distortion in the position of certain points in the point cloud, making it difficult to align 3D point clouds of an object acquired from different angles. To simplify 3D point cloud registration and minimize adjustments to positions in different 3D point clouds, the 3D point cloud that best matches the boundary feature curves of multiple 3D point clouds should be selected as the reference.

[0051] Specifically, in step S301, reference curves are screened out using the overall distribution level of trend consistency corresponding to boundary feature curves with matching relationships in different target point cloud data, and the reference point cloud is screened out based on the number and length of reference curves in the target point cloud data.

[0052] First, the true conformity of the boundary feature curve is calculated according to the overall distribution level of the trend consistency corresponding to the boundary feature curves with matching relationships in different target point cloud data.

[0053] As a preferred embodiment, the calculation of the true conformity of the boundary characteristic curve includes the following specific methods: For target point cloud data in any scanning direction, when any boundary feature curve in the target point cloud data has a matching curve in the target point cloud data in all other scanning directions, the number of corresponding target point cloud data is obtained, and recorded as the matching inclusion amount of the boundary feature curve; the target point cloud data in any scanning direction is recorded as the first point cloud data, and any boundary feature area in the first point cloud data is recorded as the first boundary feature curve; the maximum trend consistency of all matching curves of the first boundary feature curve in any target point cloud data other than the first point cloud data is obtained, and recorded as the maximum trend consistency of the first boundary feature curve in any target point cloud data other than the first point cloud data; the average value of the maximum trend consistency of the first boundary feature curve in any target point cloud data other than the first point cloud data is obtained, and recorded as the consistency parameter of the first boundary feature curve; according to the matching inclusion amount and the consistency parameter of any boundary feature curve, the true conformity of the boundary feature curve is calculated, wherein the matching inclusion amount and the consistency parameter are both positively correlated with the true conformity.

[0054] As an optional embodiment, the specific calculation method of the true compliance is: in, Indicates the The target point cloud data in the scanning direction The true conformity of the boundary characteristic curve; Indicates the The target point cloud data in the scanning direction The matching inclusion amount of the boundary characteristic curve; Indicates the The target point cloud data in the scanning direction Consistent parameters of the boundary characteristic curves; Represents the sigmoid normalization function.

[0055] It should be noted that when The target point cloud data in the scanning direction The greater the matching inclusion amount of a boundary feature curve, the more three-dimensional point clouds there are when the boundary feature curve has a matching relationship with the boundary feature curve in other three-dimensional point clouds, and the greater the average value of the corresponding maximum trend consistency in different three-dimensional point clouds, the greater the matching degree of the boundary feature curve c in the three-dimensional point clouds at other angles, and the less likely it is that there is a distortion problem itself.

[0056] Then, when the true conformity is greater than the preset second parameter, the corresponding boundary characteristic curve is recorded as a reference curve.

[0057] It should be noted that the second parameter is preset to 0.7 based on experience and can be adjusted according to actual conditions, and is not specifically limited in the embodiment of the present invention.

[0058] Finally, the number of reference curves contained in the target point cloud data of any scanning direction and the number of feature points contained in all the reference curves in the target point cloud data are obtained; the reference coefficient of the target point cloud data in the scanning direction is calculated based on the number of reference curves in the target point cloud data of the scanning direction and the proportion of the feature points contained in all the reference curves in the target point cloud data in all the target point cloud data, wherein the number of reference curves and the proportion of the feature points are positively correlated with the reference coefficient; the target point cloud data with the largest reference coefficient in the target point cloud data of all the scanning directions is used as the reference point cloud.

[0059] As an optional embodiment, the specific calculation method of the reference coefficient is: in, Indicates the Reference coefficient of target point cloud data in each scanning direction; Indicates the The number of reference curves contained in the target point cloud data in each scanning direction; Indicates the The number of feature points contained in all reference curves in the target point cloud data in each scanning direction; Indicates the number of feature points contained in all reference curves in all target point cloud data.

[0060] It should be noted that the reference coefficient is used to describe the reference value of the corresponding target point cloud data for subsequent point cloud registration; for target point cloud data in any scanning direction, the more reference curves the target point cloud data contains, and the proportion of the number of feature points on the reference curve of the object is greater. The larger it is, the smaller the difference between the boundary feature curve in the target point cloud data and the boundary of the actual scanned object is, and the greater the reference value of the target point cloud data as a reference point cloud for subsequent point cloud registration.

[0061] Step S302 : adjusting the positions of the feature points in the reference point cloud based on the matching relationship between the boundary feature curves.

[0062] It should be noted that the reference point cloud for point cloud registration may also have certain position distortions. In order to make the positions of each point in the reference point cloud more consistent with the performance of the actual object, the feature points in different boundary feature curves should be restored first, and the positions of points outside the boundary feature curves should be restored with reference to the distance change of the boundary feature curve during restoration, so that the positions of each point in the reference point cloud are more referenceable.

[0063] First, obtain the feature points in the boundary feature curve and the corresponding matching curve in any reference point cloud, which are in a matching relationship, and record them as the matching points of the feature points in the boundary feature curve in the matching curve. Obtain the Euclidean distance and Euler angle between any feature point in the boundary feature curve and the matching point in the corresponding matching curve. Obtain the distance between any feature point and the matching point, and record it as the matching distance of the feature point, as well as the Euler angle of the connection path, and record it as the matching angle of the feature point. Calculate the difference in the corresponding matching distance between the feature point and other feature points on the same boundary feature curve, as well as the cumulative sum of the differences in the matching angles at different angles. According to the difference in the matching distance between the feature point and all other feature points on the same feature curve, as well as the cumulative sum of the differences in the corresponding matching angles at different angles, obtain the relationship performance coefficient of the feature point.

[0064] As an optional embodiment, the specific calculation method of the relationship performance coefficient is: in , Indicates the The boundary characteristic curve The relationship performance coefficient of feature points; Indicates the The boundary characteristic curve The feature point and The difference in matching distance between feature points; Indicates the The number of characteristic points in a boundary characteristic curve; Represents the sigmoid normalization function.

[0065] It should be noted that when a feature point is aligned with other feature points on the same boundary feature curve, and The smaller the difference in the product accumulation value, the greater the consistency of the corresponding relationship of the feature points, and the more correct their positions are; for the same pair of boundary feature curves, compare a single feature point on a boundary feature curve with other feature points on the same boundary feature curve. When the matching distance between the feature point and its matching point and the matching distance between other feature points and the corresponding matching point on the same boundary feature curve are more consistent, and the Euler angles are more consistent, then the greater the consistency of the corresponding relationship between the feature point on the boundary feature curve and the feature point on the matching curve, the more correct the position of the feature point on the boundary feature curve is.

[0066] Then, for any boundary feature curve of any reference point cloud, the number of feature points in the boundary feature curve whose relationship performance coefficient is greater than a preset third parameter is obtained, and recorded as a first number of the boundary feature curve. Based on the first number of the boundary feature curve and the degree of consistency between the changing trends of the boundary feature curve and the matching curve, an adjustment coefficient between the boundary feature curve and the matching curve is obtained, and the first number and the degree of consistency between the changing trends are positively correlated with the adjustment coefficient.

[0067] As an optional embodiment, the specific calculation method of the adjustment coefficient is: in, Indicates the The boundary characteristic curve and the corresponding The adjustment coefficient between the matching curves, Indicates the a first number of boundary characteristic curves; Indicates the The boundary characteristic curve and the corresponding The degree of consistency of the changing trends between the matching curves.

[0068] It should be noted that when the boundary feature point i in the boundary feature curve of the reference point cloud of the point cloud registration corresponds to Compared with other points g When it is the maximum, it is determined that the position of the boundary feature point i in the boundary feature curve in the reference point cloud is correct. When it is not the maximum, the degree of consistency of the boundary feature curve trend in the reference point cloud aligned with the point cloud is used. Larger and corresponding The larger curve serves as a reference for the position change of the boundary feature point i.

[0069] Finally, the maximum value of the adjustment coefficients of the boundary feature curve and all the corresponding matching curves is obtained, and the matching point in the matching curve corresponding to the maximum adjustment coefficient is used as the reference point of the feature point corresponding to the matching point in the boundary feature curve. According to the relative position relationship between the reference points in the matching curve corresponding to the boundary feature curve, the position of the feature point in the boundary feature curve is adjusted so that the relative position relationship between the feature points in the boundary feature curve is the same as the relative position relationship between the reference points in the corresponding matching curve, and the Euler angle before and after the position change of the feature point in the boundary feature curve and the distance before and after the position change are recorded, which are respectively recorded as the adjusted Euler angle and the adjusted distance.

[0070] At this point, the feature points after position adjustment, the adjusted Euler angles of the feature points, and the adjusted distances are obtained through the above method.

[0071] Step S004: Merge adjacent feature points based on the changes before and after the feature point position adjustment to obtain several areas with consistent adjustment trends. Use the overall adjustment of the feature points in the areas with consistent adjustment trends to move and adjust the data points other than the feature points in the reference point cloud. After adjusting the positions of all data points in all reference point clouds, a new reference point cloud is obtained.

[0072] It should be noted that after adjusting the position of the boundary feature curve in the reference point cloud, the positions of the points within the boundary curve should be restored. Because the distortion varies across different regions, the regions with consistent distortion are divided based on their performance during restoration. The positions of the corresponding points are adjusted based on the relationship between the position of the reference points and the position of the divided distorted regions.

[0073] Specifically, first, for the reference point cloud after the position of the feature point in the boundary feature curve is adjusted, the absolute value of the difference in the adjusted distance between adjacent feature points in the reference point cloud and the sum of the absolute value of the difference in the adjusted Euler angle are obtained, which are respectively recorded as the first difference and the second difference. According to the first difference and the second difference, the merging feasibility between adjacent feature points is obtained. The first difference and the second difference are both negatively correlated with the merging feasibility. When the merging feasibility is greater than or equal to the preset fourth parameter, the corresponding two feature points are merged. After merging all adjacent feature points whose merging feasibility is greater than or equal to the fourth parameter, several areas with consistent adjustment trends are obtained.

[0074] As an optional embodiment, the specific calculation method of the merger feasibility is: in, Indicates the The feature point and Feasibility of merging feature points; Indicates the The feature point and The first difference between feature points; Indicates the The feature point and The second difference between feature points; Represents the sigmoid normalization function.

[0075] It should be noted that after the position of the feature points is adjusted, when the adjustment distance and adjustment angle difference of adjacent feature points in the obtained reference point cloud are small, it means that the adjustment trends of the two feature points are similar. Therefore, in the embodiment of the present invention, the feasibility of merging the two feature points is used to judge whether the adjacent feature points can be divided into the same adjustment trend area.

[0076] Then, for any area with consistent adjustment trend, obtain the feature point with the largest adjustment distance in the area with consistent adjustment trend, record it as the main feature point, use the adjustment distance of the main feature point as the adjustment length of the area with consistent adjustment trend, use the adjustment Euler angle of the main feature point as the adjustment scanning direction of the area with consistent adjustment trend, record the data points other than the feature points in the reference point cloud as non-feature points, obtain the area with consistent adjustment trend closest to any non-feature point, record it as the target area of ​​the non-feature point, for any non-feature point, obtain the midpoint of the boundary feature curve closest to the non-feature point in the target area, and pass through the midpoint and scan direction. A straight line that is the adjustment scanning direction of the target area is used as the adjustment trend line of the target area to which it belongs, and the angle between the line connecting the non-feature point and the midpoint and the adjustment trend line is obtained and recorded as the first adjustment parameter of the non-feature point. The minimum distance between the non-feature point and the adjustment trend line is obtained and recorded as the second adjustment parameter of the non-feature point. The adjustment influence of the non-feature point is obtained according to the first adjustment parameter and the second adjustment parameter. In combination with the adjustment length of the target area of ​​the non-feature point and the adjustment influence of the non-feature point, the moving distance of the non-feature point along the adjustment scanning direction of the corresponding target area is obtained.

[0077] As an optional embodiment, a specific method for calculating the moving distance of the non-feature point along the adjusted scanning direction of the corresponding target area is: in, Indicates the The moving distance of each non-feature point along the adjusted scanning direction of the corresponding target area; Indicates the The adjusted length of the target area corresponding to each non-feature point; Indicates the The adjustment influence of non-feature points.

[0078] The specific method for obtaining the adjustment influence of the non-feature point according to the first adjustment parameter and the second adjustment parameter further includes: both the first adjustment parameter and the second adjustment parameter are negatively correlated with the adjustment influence of the non-feature point.

[0079] As an optional embodiment, a specific method for calculating the adjustment influence of the non-feature point is: in, Indicates the The adjustment influence of non-feature points; Indicates the The first adjustment parameter of non-feature points; Indicates the The second adjustment parameter of non-feature points; Represents the sigmoid normalization function.

[0080] It should be noted that when the adjustment effect is greater, the moving distance of the ordinary point p should be longer. When the angle difference between the connected line and the adjustment trend at different Euler angles is The smaller the distance The smaller it is, the greater the impact of the changing trend in this area on the position of the ordinary point p.

[0081] Finally, all non-feature points in the reference point cloud are moved along the adjusted scanning direction of the corresponding target area according to the corresponding moving distance, and the feature points after position adjustment in the reference point cloud are merged according to the merging feasibility to obtain a new reference point cloud.

[0082] At this point, the new reference point cloud is obtained by adjusting the positions of all data points in the reference point cloud using the above method.

[0083] Step S005: register and stitch all new reference point clouds.

[0084] Specifically, first, all adjusted reference point clouds are registered and spliced ​​using the corresponding relationship of the boundary characteristic curves to obtain the complete point cloud data of the object. Through the above operations, the point cloud data of the 3D laser scanner is processed to obtain the adjustment method of the data points in the different point cloud data and the final spliced ​​point cloud.

[0085] Then, the obtained analysis data is transferred to the database for corresponding storage. The adjusted distance and Euler angle of feature points in different reference point clouds, as well as the adjusted distance and scanning direction of non-feature points are recorded through SQL query statements and visualized in the form of a table.

[0086] Through the above steps, the calibration of the 3D point cloud data of the 3D laser scanner is completed.

[0087] An error calibration system for a three-dimensional laser scanner includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the contents of steps S001 to S005 of the error calibration method for a three-dimensional laser scanner are implemented.

[0088] Furthermore, in an optional embodiment, the memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. The memory may also include a non-volatile random access memory. For example, the memory may also store device type information.

[0089] The memory may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory, wherein the non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available. For example, static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct memory bus random access memory (DRRAM).

[0090] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor. It is worth noting that the processor may be a processor that supports the Advanced RISC Machines (ARM) architecture.

[0091] This embodiment can achieve accurate matching of feature points by analyzing the changing trends between boundary feature curves, thereby optimizing the subsequent point cloud data registration and splicing process. In addition, by utilizing the matching relationship of boundary feature curves, a reference curve consistent with the overall trend is selected, thereby improving the selection accuracy of reference point cloud data, and being able to effectively adjust the position of feature points. By utilizing the adjustment of feature point positions and the merging of trend-consistent areas, the matching accuracy of adjacent feature points can be effectively improved, making the final reference point cloud more accurate, thereby improving the registration effect of point cloud data. Further, by adjusting the positions of data points in all reference point clouds, the global consistency of all data points is ensured, and the resulting new reference point cloud will be more realistic and accurate, thereby improving the accuracy of 3D point cloud data collected by the 3D laser scanner.

[0092] It should be noted that the The model is only used to represent negative correlation and constrain the output of the model to be in In the specific implementation, it can be replaced by other models with the same purpose. This embodiment is only based on The model is described as an example without any specific limitation. is the input to the model.

[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An error calibration method for a three-dimensional laser scanner, characterized in that: The method comprises the following steps: Scan the object with a three-dimensional laser scanner in several scanning directions to obtain three-dimensional point cloud data in each scanning direction; Constructing several 3D pooling windows of different sizes to pool any 3D point cloud data, obtaining the pooled point cloud data and several boundary feature curves composed of several feature points in the pooled point cloud data. Analyzing the consistency of the changing trends between the boundary feature curves to obtain the matching relationship between the boundary feature curves, based on the matching relationship between the boundary feature curves in the 3D point cloud data of all scanning directions, the pooled point cloud data is filtered to obtain the target point cloud data. The overall distribution level of trend consistency corresponding to the boundary feature curves with matching relationships in different target point cloud data is used to screen out reference curves. The reference point cloud is screened based on the number and length of reference curves in the target point cloud data. The position of the feature points in the reference point cloud is adjusted based on the matching relationship between the boundary feature curves. Merge adjacent feature points based on their position changes before and after adjustment to obtain several regions with consistent adjustment trends. Use the overall adjustment of feature points in these regions to move and adjust data points other than feature points in the reference point cloud. After adjusting the positions of all data points in all reference point clouds, a new reference point cloud is obtained. All new reference point clouds are registered and stitched.

2. The error calibration method for a three-dimensional laser scanner according to claim 1, characterized in that: The method of constructing a plurality of three-dimensional pooling windows of different sizes to perform pooling processing on any three-dimensional point cloud data to obtain the pooled point cloud data and a plurality of boundary feature curves composed of a plurality of feature points in the pooled point cloud data includes the following specific methods: The default side length is The three-dimensional pooling window is used to slide the three-dimensional point cloud data. The three-dimensional point cloud data is downsampled by the maximum pooling operation during the sliding traversal process. The maximum pooling operation of the three-dimensional pooling window with any side length is used as a pooling method to obtain the pooled point cloud data under any pooling method and several feature points contained in the pooled point cloud data. The side length of the three-dimensional pooling window is , and the size of the three-dimensional pooling window is in the interval Internal As the step size increases, the sliding compensation of the three-dimensional pooling window during the sliding traversal process is ,in 、 、 as well as are the preset first side length parameter, second side length parameter, first step length parameter, and second step length parameter respectively; Under any pooling mode, the distance between feature points is obtained, and the feature points whose distance is less than a preset distance threshold are connected to each other to obtain several boundary feature curves under the corresponding pooling mode.

3. The error calibration method for a three-dimensional laser scanner according to claim 1, characterized in that: The specific method for obtaining the degree of consistency of the change trends between the boundary characteristic curves is: For any two three-dimensional point cloud data with different scanning directions, the shortest alignment path between any two boundary feature curves in the two three-dimensional point cloud data and the correspondence between the feature points in the two boundary feature curves are obtained by the DDTW algorithm. For any one of the two boundary feature curves, it is recorded as the first target boundary feature curve, and the other boundary feature curve is recorded as the second target boundary feature curve. The number of feature points on the first target boundary feature curve that have a corresponding relationship on the second target boundary feature curve is counted and recorded as the matching parameter. The number of feature points in the two boundary feature curves whose matching parameters are greater than the preset first parameter is recorded as the overmatching parameter between the two boundary feature curves. According to the length of the shortest alignment path between the two boundary feature curves and the overmatching parameter, the degree of consistency of the change trend between the two boundary feature curves is obtained, wherein the length of the shortest alignment path and the overmatching parameter are both negatively correlated with the degree of consistency of the change trend.

4. The error calibration method for a three-dimensional laser scanner according to claim 2, characterized in that: The matching relationship between the boundary feature curves in the three-dimensional point cloud data in all scanning directions is used to filter the pooled point cloud data to obtain the target point cloud data, including the following specific methods: When the degree of consistency of the change trend is greater than a preset consistency threshold, the corresponding second target boundary feature curve is used as the matching curve of the corresponding first target boundary feature curve; under any pooling method, the number of matching curves of any boundary feature curve in the three-dimensional point cloud data in any scanning direction is obtained, and recorded as the matching amount of the boundary feature curve; and the total number of feature points in all three-dimensional point cloud data under any pooling method is counted; Calculate the pooling suitability under any pooling mode according to the side length of the corresponding pooling window, the matching amount of all boundary feature curves in the three-dimensional point cloud data in all scanning directions, and the total number of feature points in the three-dimensional point cloud data in all scanning directions; the side length of the pooling window and the matching amount of all boundary feature curves in the three-dimensional point cloud data in all scanning directions are positively correlated with the pooling suitability, and the total number of feature points in the three-dimensional point cloud data in all scanning directions is negatively correlated with the pooling suitability; The pooling mode corresponding to the maximum pooling fitness under all pooling modes is obtained, recorded as the target pooling mode, and the pooled point cloud data obtained under the target pooling mode is recorded as the target point cloud data.

5. The error calibration method for a three-dimensional laser scanner according to claim 4, characterized in that: The method of selecting a reference curve by utilizing the overall distribution level of trend consistency corresponding to the boundary feature curves having matching relationships in different target point cloud data includes the following specific methods: For target point cloud data in any scanning direction, when any boundary feature curve in the target point cloud data has a matching curve in the target point cloud data in all other scanning directions, the number of corresponding target point cloud data is obtained, and recorded as the matching inclusion amount of the boundary feature curve; the target point cloud data in any scanning direction is recorded as the first point cloud data, and any boundary feature area in the first point cloud data is recorded as the first boundary feature curve; the maximum trend consistency of all matching curves of the first boundary feature curve in any target point cloud data other than the first point cloud data is obtained, and recorded as the maximum trend consistency of the first boundary feature curve in any target point cloud data other than the first point cloud data; the average value of the maximum trend consistency of the first boundary feature curve in any target point cloud data other than the first point cloud data is obtained, and recorded as the consistency parameter of the first boundary feature curve; according to the matching inclusion amount and the consistency parameter of any boundary feature curve, the true conformity of the boundary feature curve is calculated, wherein the matching inclusion amount and the consistency parameter are both positively correlated with the true conformity; When the true conformity is greater than the preset second parameter, the corresponding boundary characteristic curve is recorded as a reference curve.

6. The error calibration method for a three-dimensional laser scanner according to claim 1, characterized in that: The specific method of filtering out the reference point cloud by the number and length of the reference curves in the target point cloud data is as follows: Obtain the number of reference curves contained in the target point cloud data of any scanning direction, and the number of feature points contained in all the reference curves in the target point cloud data; calculate the reference coefficient of the target point cloud data in the scanning direction based on the number of reference curves in the target point cloud data of the scanning direction and the proportion of the feature points contained in all the reference curves in the target point cloud data in all the target point cloud data, wherein the number of reference curves and the proportion of the feature points are positively correlated with the reference coefficient; and use the target point cloud data with the largest reference coefficient in the target point cloud data of all the scanning directions as the reference point cloud.

7. The error calibration method for a three-dimensional laser scanner according to claim 4, characterized in that: The specific method of adjusting the position of the feature points in the reference point cloud based on the matching relationship between the boundary feature curves includes: Obtain the maximum value of the adjustment coefficients of the boundary feature curve and all the corresponding matching curves, use the matching point in the matching curve corresponding to the maximum adjustment coefficient as the reference point of the feature point corresponding to the matching point in the boundary feature curve, and adjust the position of the feature point in the boundary feature curve according to the relative position relationship between the reference points in the matching curve corresponding to the boundary feature curve, so that the relative position relationship between the feature points in the boundary feature curve is the same as the relative position relationship between the reference points in the corresponding matching curve, and record the Euler angle before and after the position change of the feature point in the boundary feature curve and the distance before and after the position change, which are respectively recorded as the adjusted Euler angle and the adjusted distance.

8. The error calibration method for a three-dimensional laser scanner according to claim 7, characterized in that: The method of merging adjacent feature points based on the changes before and after the feature point positions are adjusted to obtain several regions with consistent adjustment trends includes: For the reference point cloud after the position of the feature points in the boundary feature curve is adjusted, the absolute value of the difference in the adjusted distance between adjacent feature points in the reference point cloud and the sum of the absolute value of the difference in the adjusted Euler angle are obtained, which are respectively recorded as the first difference and the second difference. According to the first difference and the second difference, the merging feasibility between adjacent feature points is obtained. The first difference and the second difference are both negatively correlated with the merging feasibility. When the merging feasibility is greater than or equal to the preset fourth parameter, the corresponding two feature points are merged. After merging all adjacent feature points whose merging feasibility is greater than or equal to the fourth parameter, several areas with consistent adjustment trends are obtained.

9. The error calibration method for a three-dimensional laser scanner according to claim 7, characterized in that: The method of using the overall adjustment of the feature points in the region with the same adjustment trend to move and adjust the data points other than the feature points in the reference point cloud includes the following specific methods: For any consistent adjustment trend region, a feature point with the largest adjustment distance in the consistent adjustment trend region is obtained and recorded as a main feature point. The adjustment distance of the main feature point is used as the adjustment length of the consistent adjustment trend region to which it belongs. The adjusted Euler angle of the main feature point is used as the adjustment scanning direction of the consistent adjustment trend region to which it belongs. Data points other than feature points in the reference point cloud are recorded as non-feature points. The consistent adjustment trend region closest to any non-feature point is obtained and recorded as the target region of the non-feature point. For any non-feature point, the midpoint of the boundary feature curve closest to the non-feature point in the target region is obtained. A straight line passing through the midpoint and having a scanning direction in the adjusted scanning direction of the target region is used as the adjustment trend line of the target region to which it belongs. The angle between the line connecting the non-feature point and the midpoint and the adjustment trend line is obtained and recorded as a first adjustment parameter of the non-feature point. The minimum distance between the non-feature point and the adjustment trend line is obtained and recorded as a second adjustment parameter of the non-feature point. The adjustment influence of the non-feature point is obtained based on the first adjustment parameter and the second adjustment parameter. The movement distance of the non-feature point along the adjusted scanning direction of the corresponding target region is obtained by combining the adjustment length of the target region of the non-feature point and the adjustment influence of the non-feature point.

10. An error calibration system for a three-dimensional laser scanner, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the error calibration method for a three-dimensional laser scanner as described in any one of claims 1 to 9 are implemented.

Images