Matching method and system for plane point cloud, computer equipment and storage medium

Abstract

The invention discloses a matching method for plane point cloud. The matching method comprises the following steps of S1, respectively constructing a local coordinate system in target point cloud data and reference point cloud data; obtaining a relation feature between a first cloud point in the target point cloud data and a second cloud point in the reference point cloud data based on the local coordinate system; s2, performing PPF algorithm matching on the second cloud point and all the first cloud points of the target point cloud data to obtain a matching result of the second cloud point; s3, updating the weight in the matching algorithm based on the matching result of the second cloud point; and S4, repeating the steps S2-S3 until all the first cloud points are matched. The matching weight of the next second cloud point is updated by the matching weight or the matching result of the last second cloud point. The weight updating is adaptive, the matching robustness is improved, and the algorithm is more excellent in performance in different scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud matching, and in particular to a matching method, system, computer equipment and storage medium for planar point clouds. Background Art

[0002] In cutting-edge technology fields such as 3D vision, robot navigation, and virtual reality, accurate matching of planar point clouds plays a vital role. It is not only the basis for building 3D environment models and achieving accurate positioning and navigation, but also one of the key technologies for enhancing the realism of virtual reality experience. However, when faced with large-scale, complex and changeable planar point cloud data, traditional point pair features (PPF) algorithms often expose limitations such as insufficient robustness and sensitivity to noise. These problems may not only lead to a decrease in matching accuracy, but also seriously affect the quality of subsequent 3D reconstruction, path planning, and interactive experience.

[0003] Therefore, it is necessary to propose an improved PPF algorithm to improve the accuracy and robustness of planar point cloud matching. Summary of the invention

[0004] In view of the above-mentioned defects, the purpose of the present invention is to propose a matching method, system, computer device and storage medium for planar point clouds, so as to solve the problems of poor accuracy and robustness in existing planar point cloud matching.

[0005] To achieve this purpose, the present invention adopts the following technical solution: a matching method for a plane point cloud, comprising the following steps:

[0006] Step S1: constructing local coordinate systems in the target point cloud data and the reference point cloud data respectively;

[0007] Acquire the relationship characteristics between the first cloud point in the target point cloud data and the second cloud point in the reference point cloud data based on the local coordinate system;

[0008] Step S2: Use the second cloud point to match all the first cloud points of the target point cloud data using the PPF algorithm to obtain the matching result of the second cloud point;

[0009] Step S3: updating the weights in the matching algorithm based on the matching result of the second cloud point;

[0010] Step S4, repeating steps S2 to S3 until all first cloud points are matched.

[0011] Preferably, the construction process of the local coordinate system is as follows:

[0012] Select any point A and point B from the target point cloud data or the reference point cloud data to construct a point pair (A, B);

[0013] Take point A as the origin, vector AB as the positive direction of the X-axis, and select the vector perpendicular to vector AB as the Y-axis direction to construct a local coordinate system.

[0014] Preferably, the relationship features include the distance between point pairs, the rotation angle and the rotation axis.

[0015] Preferably, the method for updating the weight in step S3 is as follows:

[0016] Get the error value in this matching result;

[0017] The weighted weights are updated based on the error values ​​to obtain updated weights;

[0018] Normalize the updated weight machine.

[0019] Preferably, the method for updating the weight in step S3 is as follows:

[0020] Get the weight of the last matching result as the first weight;

[0021] Confidence information is obtained, and the first total weight is updated based on the confidence information.

[0022] Preferably, in step S2, using the second cloud point to match all the first cloud points of the target point cloud data with the PPF algorithm further comprises the following steps:

[0023] Determine whether there is a point cloud data with the same relationship features;

[0024] If it does not exist, the first cloud point with the smallest rotation matrix and translation matrix is ​​found based on the PPF algorithm as the matching result;

[0025] If it exists, when the second cloud point is matched with the first cloud point by the PPF algorithm, the matching rotation matrix and translation matrix are obtained;

[0026] Store the rotation matrix and translation matrix into a voting table respectively;

[0027] The rules for depositing a voting form are as follows:

[0028] When the rotation matrix or translation matrix appears in the first cloud point matching result for the first time, a new voting option is created in the voting table, and the number of votes for the voting option is set to 1;

[0029] When the rotation matrix or translation matrix appears repeatedly in other first cloud point matching results, the number of votes for the rotation matrix or translation matrix in the voting option is increased;

[0030] The rotation matrix and translation matrix with the most votes in the voting table are counted respectively as the first matrix and the second matrix, and the first matrix and the second matrix are used as the matching results of the second cloud point.

[0031] A matching system for a plane point cloud, using the matching method for a plane point cloud, characterized in that it comprises a coordinate construction module, a matching module and an update module;

[0032] The coordinate construction module is used to construct a local coordinate system in the target point cloud data and the reference point cloud data respectively;

[0033] Acquire the relationship characteristics between the first cloud point in the target point cloud data and the second cloud point in the reference point cloud data based on the local coordinate system;

[0034] The matching module is used to use the second cloud point to perform PPF algorithm matching with all the first cloud points of the target point cloud data to obtain a matching result of the second cloud point;

[0035] The updating module is used to update the weights in the matching algorithm based on the matching result of the second cloud point.

[0036] A computer device comprising:

[0037] memory and at least one processor;

[0038] The memory is used to store at least one program;

[0039] When the program is executed by the processor, the processor implements the matching method for a planar point cloud as described above.

[0040] One of the above technical solutions has the following advantages or beneficial effects: In the present invention, the weight in the matching algorithm is also updated based on the matching result of the second cloud point, thereby improving the sensitivity of a certain relationship feature in the matching algorithm. In the matching process of the present invention, the second cloud points are matched one by one, not multiple second cloud points are matched simultaneously. Although this will affect the matching efficiency, the matching weight of the next second cloud point is updated by the matching weight or matching result of the previous second cloud point. This makes the weight update adaptive, improves the robustness of the matching, and makes the algorithm perform better in different scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart of an embodiment of the method of the present invention.

[0042] Figure 2 It is a schematic diagram of the structure of an embodiment of the system of the present invention. DETAILED DESCRIPTION

[0043] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0044] In the description of the embodiments of the present invention, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0045] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "plurality" is two or more. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0046] like Figures 1-2 As shown, a matching method for a plane point cloud includes the following steps:

[0047] Step S1: constructing local coordinate systems in the target point cloud data and the reference point cloud data respectively;

[0048] Acquire the relationship characteristics between the first cloud point in the target point cloud data and the second cloud point in the reference point cloud data based on the local coordinate system;

[0049] Step S2: Use the second cloud point to match all the first cloud points of the target point cloud data using the PPF algorithm to obtain the matching result of the second cloud point;

[0050] Step S3: updating the weights in the matching algorithm based on the matching result of the second cloud point;

[0051] Step S4, repeating steps S2 to S3 until all first cloud points are matched.

[0052] The reference point cloud data is the point cloud data previously constructed by the model, and the acquired target point cloud data is matched with the reference point cloud data. In the present invention, by constructing local coordinate systems in the target point cloud data and the reference point cloud data, the relationship features of different point clouds are better described in the coordinate system by constructing the local coordinate system, and then the first cloud point and the second cloud point are matched by the existing PPF algorithm. Since some relationship features have poor sensitivity in the matching process, if the default weight is used to continue the subsequent matching, a certain relationship feature will be ignored in the matching process, so that the matching result is inaccurate. Affecting the accuracy and robustness of the plane point cloud during matching. For this reason, in the present invention, the weight in the matching algorithm is also updated based on the matching result of the second cloud point, so as to improve the sensitivity of a certain relationship feature in the matching algorithm. In the matching process of the present invention, the second cloud points are matched one by one, not multiple second cloud points are matched simultaneously. Although it will affect the matching efficiency, the matching weight of the next second cloud point is updated by the matching weight or matching result of the last second cloud point. This makes the weight update adaptive, improves the robustness of the matching, and makes the algorithm perform better in different scenarios.

[0053] Preferably, the construction process of the local coordinate system is as follows:

[0054] Select any point A and point B from the target point cloud data or the reference point cloud data to construct a point pair (A, B);

[0055] Take point A as the origin, vector AB as the positive direction of the X-axis, and select the vector perpendicular to vector AB as the Y-axis direction to construct a local coordinate system.

[0056] Preferably, the relationship features include the distance between point pairs, the rotation angle and the rotation axis.

[0057] Assume 1 that two points p A and p B , and their corresponding normal vectors n A and n B , the calculation process of PPF descriptor is as follows:

[0058] The distance between point pairs: d = || p B -p A ||;

[0059] The rotation angle between the normal vector at point A and the vector AB:

[0060] The rotation angle between the normal vector at point B and the vector AB:

[0061] The rotation axis of the normal vector at point A and the normal vector at point B:

[0062] These geometric eigenvalues ​​constitute the relationship characteristics of a point pair:

[0063] PPF(p A ,p B )=(d,α,β,γ).

[0064] Preferably, the method for updating the weight in step S3 is as follows:

[0065] Get the error value in this matching result;

[0066] The weighted weights are updated based on the error values ​​to obtain updated weights;

[0067] Normalize the updated weight machine.

[0068] Specifically, when the matching algorithm is first performed, since there is no prior error value, an initial weight can be set Where N is the total number of matching pairs, (1) represents the weight of the first version; and the error value in the matching process can be geometric distance, color difference or other features.

[0069] The specific formula for updating the weighted weights by the error value is as follows:

[0070] in represents the error value in the t-th match, ε is a positive number used to avoid division by zero errors, Indicates the weight value of the t+1th matching of the point pair.

[0071] After the weights are updated by the error value, the weight values ​​may tend to focus on the relationship features that were previously ignored. In order to ensure that the contribution of other relationship features in the model is fair, the updated weights are also normalized in the invention to avoid the situation where certain relationship features dominate the model due to excessive weights.

[0072] The specific formula for normalization is as follows:

[0073]

[0074] in Represents the weight value of the point pair t+1 matching, Represents the weight of the point pair before the t+1th matching update.

[0075] Preferably, the method for updating the weight in step S3 is as follows:

[0076] Get the weight of the last matching result as the first weight;

[0077] Confidence information is obtained, and the first total weight is updated based on the confidence information.

[0078] Confidence information is an important indicator for measuring the reliability of association rules or model prediction results. By introducing confidence information into the weight update process, the present invention can ensure that the model pays more attention to those relationship features when updating the weights, thereby improving the reliability of the overall model.

[0079] Preferably, in step S2, using the second cloud point to match all the first cloud points of the target point cloud data with the PPF algorithm further comprises the following steps:

[0080] Determine whether there is a point cloud data with the same relationship features;

[0081] If it does not exist, the first cloud point with the smallest rotation matrix and translation matrix is ​​found based on the PPF algorithm as the matching result;

[0082] If it exists, when the second cloud point is matched with the first cloud point by the PPF algorithm, the matching rotation matrix and translation matrix are obtained;

[0083] Store the rotation matrix and translation matrix into a voting table respectively;

[0084] The rules for depositing a voting form are as follows:

[0085] When the rotation matrix or translation matrix appears in the first cloud point matching result for the first time, a new voting option is created in the voting table, and the number of votes for the voting option is set to 1;

[0086] When the rotation matrix or translation matrix appears repeatedly in other first cloud point matching results, the number of votes for the rotation matrix or translation matrix in the voting option is increased;

[0087] The rotation matrix and translation matrix with the most votes in the voting table are counted respectively as the first matrix and the second matrix, and the first matrix and the second matrix are used as the matching results of the second cloud point.

[0088] Since the matching is performed through the characteristic relationship between cloud points in the present invention, and the characteristic relationship is the relative position relationship of cloud points in the local coordinate system, there may be first cloud points with the same relationship characteristics. If the same matching results are obtained when matching is performed, the matching cannot be successful. In order to solve this problem, a bidding mechanism is added to the present invention. When it is found that there are first cloud points with the same relationship characteristics, when the second cloud point is matched with the first cloud point, the obtained rotation matrix and translation matrix will be stored in a voting table respectively, and the rotation matrices and translation matrices of other first cloud points that are subsequently obtained will continue to be stored in the voting table. If there are identical rotation matrices or translation matrices, their votes will be increased. When the second cloud point is matched with all the first cloud points, the rotation matrix and translation matrix with the highest number of votes are selected as the matching result of the second cloud point.

[0089] For example, three pairs of points are matched between the target point cloud data and the reference point cloud data:

[0090] (p1,p2) corresponds to (p′1,p′2), vote R 12 , T 12 ;

[0091] (p2,p3) corresponds to (p′2,p′3), vote R 23 , T 23 ;

[0092] (p1,p3) corresponds to (p′1,p′3), vote R 13 , T 13 ;

[0093] Assume that R 12 , T 12 Get the highest number of votes among multiple pairs of points. 12 , T 12 As a matching result.

[0094] A matching system for a plane point cloud, using the matching method for a plane point cloud, characterized in that it comprises a coordinate construction module, a matching module and an update module;

[0095] The coordinate construction module is used to construct a local coordinate system in the target point cloud data and the reference point cloud data respectively;

[0096] Acquire the relationship characteristics between the first cloud point in the target point cloud data and the second cloud point in the reference point cloud data based on the local coordinate system;

[0097] The matching module is used to use the second cloud point to perform PPF algorithm matching with all the first cloud points of the target point cloud data to obtain a matching result of the second cloud point;

[0098] The updating module is used to update the weights in the matching algorithm based on the matching result of the second cloud point.

[0099] A computer device comprising:

[0100] memory and at least one processor;

[0101] The memory is used to store at least one program;

[0102] When the program is executed by the processor, the processor implements the matching method for a planar point cloud as described above.

[0103] A storage medium containing computer executable instructions, which are used to perform a matching method for a planar point cloud when executed by a computer processor.

[0104] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0105] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A matching method for a plane point cloud, characterized in that: The steps include: Step S1: constructing local coordinate systems in the target point cloud data and the reference point cloud data respectively; Acquire the relationship characteristics between the first cloud point in the target point cloud data and the second cloud point in the reference point cloud data based on the local coordinate system; Step S2: Use the second cloud point to match all the first cloud points of the target point cloud data using the PPF algorithm to obtain the matching result of the second cloud point; Step S3: updating the weights in the matching algorithm based on the matching result of the second cloud point; Step S4, repeating steps S2 to S3 until all first cloud points are matched.

2. A matching method for a planar point cloud according to claim 1, characterized in that: The construction process of the local coordinate system is as follows: Select any point A and point B from the target point cloud data or the reference point cloud data to construct a point pair (A, B); Take point A as the origin, vector AB as the positive direction of the X-axis, and select the vector perpendicular to vector AB as the Y-axis direction to construct a local coordinate system.

3. A matching method for a planar point cloud according to claim 2, characterized in that: The relationship features include the distance between point pairs, the rotation angle, and the rotation axis.

4. The matching method for a planar point cloud according to claim 2, characterized in that: The method for updating the weight in step S3 is as follows: Get the error value in this matching result; The weighted weights are updated based on the error values ​​to obtain updated weights; Normalize the updated weight machine.

5. The matching method for planar point cloud according to claim 2, characterized in that: The method for updating the weight in step S3 is as follows: Get the weight of the last matching result as the first weight; Confidence information is obtained, and the first total weight is updated based on the confidence information.

6. The matching method for planar point cloud according to claim 1, characterized in that: In step S2, using the second cloud point to match all the first cloud points of the target point cloud data with the PPF algorithm also includes the following steps: Determine whether there is a point cloud data with the same relationship features; If it does not exist, the first cloud point with the smallest rotation matrix and translation matrix is ​​found based on the PPF algorithm as the matching result; If it exists, when the second cloud point is matched with the first cloud point by the PPF algorithm, the matching rotation matrix and translation matrix are obtained; Store the rotation matrix and translation matrix into a voting table respectively; The rules for depositing a voting form are as follows: When the rotation matrix or translation matrix appears in the first cloud point matching result for the first time, a new voting option is created in the voting table, and the number of votes for the voting option is set to 1; When the rotation matrix or translation matrix appears repeatedly in other first cloud point matching results, the number of votes for the rotation matrix or translation matrix in the voting option is increased; The rotation matrix and translation matrix with the most votes in the voting table are counted respectively as the first matrix and the second matrix, and the first matrix and the second matrix are used as the matching results of the second cloud point.

7. A matching system for a plane point cloud, using a matching method for a plane point cloud as claimed in any one of claims 1 to 6, characterized in that: Includes coordinate construction module, matching module and update module; The coordinate construction module is used to construct a local coordinate system in the target point cloud data and the reference point cloud data respectively; Acquire the relationship characteristics between the first cloud point in the target point cloud data and the second cloud point in the reference point cloud data based on the local coordinate system; The matching module is used to use the second cloud point to perform PPF algorithm matching with all the first cloud points of the target point cloud data to obtain a matching result of the second cloud point; The updating module is used to update the weights in the matching algorithm based on the matching result of the second cloud point.

8. A computer device, characterized in that: include: memory and at least one processor; The memory is used to store at least one program; When the program is executed by the processor, the processor implements a matching method for a planar point cloud as described in any one of claims 1 to 6.

9. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions are used to execute a matching method for a planar point cloud as described in any one of claims 1 to 6 when executed by a computer processor.

Images