A method and device for evaluating point cloud registration results

Abstract

The present application discloses a method and device for evaluating point cloud registration results. In the method, firstly, by performing point cloud registration on the source point cloud and the destination point cloud, the matching pairs and non-matching pairs of the source point cloud and the destination point cloud are determined; then, based on the distance between the source point cloud and the destination point cloud in the matching pair, the matching pairs are divided into n matching pair sets; based on the number of matching pairs in the matching pair set and the number of non-matching pairs, the weights corresponding to the non-matching pair set and the matching pair set are determined; then, based on the weight, an evaluation parameter is determined, and based on the evaluation parameter, the evaluation result of the point cloud registration is determined. The sum of the weights corresponding to each matching pair set is greater than the weight corresponding to the non-matching pair set, and the registration effect of each matching pair set is better than the registration effect of the non-matching pair set, so the evaluation parameter can assign a higher weight to the matching pairs with better registration, thereby improving the accuracy of evaluating the point cloud registration result.

Description

Technical Field

[0001] The present application relates to the field of point cloud technology, and in particular to a method and device for evaluating point cloud registration results. Background Art

[0002] Point cloud registration technology is a basic technology in the field of 3D reconstruction and simultaneous localization and mapping (SLAM). It is usually used to integrate point cloud data from different perspectives into the same coordinate system through rigid changes such as rotation and translation. In addition, after completing the point cloud registration, it is sometimes necessary to evaluate the results of the point cloud registration.

[0003] At present, the root mean square error (RMSE) is usually used as an indicator to evaluate the point cloud registration results. However, when the point cloud registration results are evaluated by this indicator, the evaluation accuracy is low. Summary of the invention

[0004] In order to solve the above technical problems, the present disclosure is proposed. The embodiments of the present disclosure provide a method and device for evaluating point cloud registration results.

[0005] According to one aspect of the present disclosure, a method for evaluating point cloud registration results is provided, comprising:

[0006] By performing point cloud registration on the source point cloud and the destination point cloud, matching pairs and non-matching pairs of the source point cloud and the destination point cloud are determined;

[0007] Based on the distance between the source point cloud and the destination point cloud in the matching pairs, the matching pairs are divided into n matching pair sets, where n is a positive integer greater than 2;

[0008] Determining weights corresponding to a non-matching pair set and each of the matching pair sets based on the number of the matching pairs in each of the matching pair sets and the number of the non-matching pairs, wherein the non-matching pair set includes the non-matching pairs;

[0009] Determining an evaluation parameter based on the weights corresponding to the non-matching pair set and each of the matching pair sets;

[0010] Based on the evaluation parameters, an evaluation result of the point cloud registration is determined.

[0011] According to another aspect of the embodiments of the present disclosure, a device for evaluating point cloud registration results is provided, comprising:

[0012] A registration module, used for determining matching pairs and non-matching pairs of the source point cloud and the destination point cloud by performing point cloud registration on the source point cloud and the destination point cloud;

[0013] a set division module, configured to divide the matching pairs into n matching pair sets based on the distance between the source point cloud and the destination point cloud in the matching pairs determined by the registration module, wherein n is a positive integer greater than 2;

[0014] a weight determination module, configured to determine weights corresponding to a non-matching pair set and each of the matching pair sets based on the number of the matching pairs in each of the matching pair sets determined by the set division module and the number of the non-matching pairs determined by the registration module, wherein the sum of the weights corresponding to each of the matching pair sets is greater than the weight corresponding to the non-matching pair set, and the non-matching pair set includes the non-matching pairs;

[0015] a parameter determination module, configured to determine an evaluation parameter based on the weights respectively corresponding to the non-matching pair set and each of the matching pair sets determined by the weight determination module;

[0016] An evaluation module is used to determine a result of the point cloud registration based on the evaluation parameters determined by the parameter determination module.

[0017] According to another aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute the method for evaluating the point cloud registration result described in any of the above embodiments of the present disclosure.

[0018] According to another aspect of the embodiments of the present disclosure, an electronic device is provided, the electronic device comprising:

[0019] processor;

[0020] a memory for storing instructions executable by the processor;

[0021] The processor is used to read the executable instructions from the memory and execute the instructions to implement the point cloud registration result evaluation method described in any of the above embodiments of the present disclosure.

[0022] When determining the evaluation parameters, the scheme of the embodiment of the present disclosure first divides the matching pairs obtained after point cloud registration of the source point cloud and the destination point cloud into at least two matching pair sets, and determines the non-matching pair set by performing point cloud registration on the source node and the destination node, and assigns corresponding weights to each matching pair set and non-matching pair set. Among them, the sum of the weights corresponding to each matching pair set is greater than the weight corresponding to the non-matching pair set. The registration effect of each matching pair set is better than the registration effect of the non-matching pair set. Therefore, the evaluation parameters determined by the scheme of the embodiment of the present disclosure can assign higher weights to matching pairs with better registration, and the evaluation of the point cloud registration results is more accurate compared with the prior art through the root mean square error. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0024] FIG. 1( a ) is a scene diagram to which the present disclosure is applicable.

[0025] FIG1( b ) is another scene diagram to which the present disclosure is applicable.

[0026] Figure 2 It is a flowchart of a method for evaluating point cloud registration results provided by an exemplary embodiment of the present disclosure.

[0027] Figure 3 It is a flowchart of a method for evaluating point cloud registration results provided by another exemplary embodiment of the present disclosure.

[0028] Figure 4 It is a flowchart of a method for evaluating point cloud registration results provided by another exemplary embodiment of the present disclosure.

[0029] Figure 5 It is a flowchart of a method for evaluating point cloud registration results provided by another exemplary embodiment of the present disclosure.

[0030] Figure 6 It is a flowchart of a method for evaluating point cloud registration results provided by another exemplary embodiment of the present disclosure.

[0031] Figure 7 It is a structural schematic diagram of a device for evaluating point cloud registration results provided by an exemplary embodiment of the present disclosure.

[0032] Figure 8It is a structural schematic diagram of a device for evaluating point cloud registration results provided by another exemplary embodiment of the present disclosure.

[0033] Fig. 9 It is a schematic diagram of the structure of an electronic device provided by an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0034] Below, the exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all the embodiments of the present disclosure, and it should be understood that the present disclosure is not limited to the exemplary embodiments described here.

[0035] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present disclosure unless specifically stated otherwise.

[0036] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present disclosure are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.

[0037] It should also be understood that in the embodiments of the present disclosure, “plurality” may refer to two or more than two, and “at least one” may refer to one, two, or more than two.

[0038] It should also be understood that any component, data or structure mentioned in the embodiments of the present disclosure can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.

[0039] In addition, the term "and / or" in the present disclosure is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present disclosure generally indicates that the associated objects before and after are in an "or" relationship.

[0040] It should also be understood that the description of the various embodiments in the present disclosure focuses on the differences between the various embodiments, and the same or similar aspects thereof can be referenced to each other, and for the sake of brevity, they will not be described one by one.

[0041] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0042] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0043] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered as part of the specification.

[0044] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0045] The disclosed embodiments can be applied to electronic devices such as terminal devices, computer systems, servers, etc., which can operate with many other general or special computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, servers, etc. include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network personal computers, small computer systems, large computer systems, and distributed cloud computing technology environments including any of the above systems, etc.

[0046] Electronic devices such as terminal devices, computer systems, servers, etc. can be described in the general context of computer system executable instructions (such as program modules) executed by computer systems. Generally, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in a distributed cloud computing environment, where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can be located on local or remote computing system storage media including storage devices.

[0047] Application Overview

[0048] In application scenarios such as assisted driving or autonomous driving, vehicles are sometimes required to be able to perceive the surrounding environment. In order to meet this requirement, it is often necessary to apply three-dimensional reconstruction and simultaneous localization and mapping (SLAM) technology, and point cloud registration technology is one of the basic technologies.

[0049] Point cloud registration is a technology used to integrate point cloud data at different perspectives into the same coordinate system through rigid changes such as rotation and translation. The point cloud data at different perspectives usually includes a source point cloud (i.e., source cloud) and a target point cloud (i.e., target cloud). Through point cloud registration, the source point cloud can be transformed into the same coordinate system as the target point cloud.

[0050] The source point cloud is usually a point cloud data set including multiple points, each point refers to the corresponding point cloud data. For example, if the source point cloud is determined by laser radar measurement, and the laser emitted by the laser radar is reflected after encountering a certain position of the object, then each point in the source point cloud (i.e. point cloud data) usually includes the three-dimensional coordinates of a certain position of the object and the laser reflection intensity. Correspondingly, the destination point cloud is usually also a point cloud data set including multiple points.

[0051] Among them, when performing point cloud registration, it is usually necessary to determine the corresponding points. The corresponding points refer to a point cloud in the source point cloud and a point cloud in the destination point cloud, and the distance between the two point clouds is less than the distance threshold. Then, using the determined corresponding points, the rotation matrix and translation matrix of the source point cloud and the destination point cloud are determined through repeated iterations. The rotation matrix and translation matrix can be used to move and transform the objects corresponding to the source point cloud.

[0052] The accuracy of the point cloud registration result is of great significance. For example, if the point cloud registration technology is applied in the assisted driving scenario, the accuracy of the result may sometimes affect the accuracy of the assisted driving. Therefore, after completing the point cloud registration, it is often necessary to evaluate the result of the point cloud registration.

[0053] Currently, the root mean square error (RMSE) is usually used to evaluate the results of point cloud registration. However, when evaluating the results of point cloud registration using this indicator, there is often a problem of low evaluation accuracy.

[0054] For example, when using the root mean square error as an indicator to evaluate the results of point cloud registration, it may sometimes fall into the local optimum, or, sometimes, even if the registration result is good, the root mean square error obtained may indicate that the registration result is poor, which cannot truly reflect the effect of point cloud registration.

[0055] Exemplary Systems

[0056] The embodiments of the present disclosure may be applied to application scenarios where the results of point cloud registration need to be evaluated, and the application scenarios may include application scenarios such as assisted driving or autonomous driving.

[0057] For example, in the application scenarios of assisted driving or autonomous driving, the source point cloud and the destination point cloud of the vehicle's surrounding environment can be registered, and the result of the point cloud registration can be evaluated. If the evaluation determines that the accuracy of the point cloud registration is high, a map including the surrounding environment is constructed based on the point cloud registration so that the vehicle can recognize the surrounding environment.

[0058] The source point cloud and the destination point cloud are both point cloud data sets including multiple points, and are generated by a point cloud generating device, which can be various types of sensors, such as a laser radar, a binocular stereo camera, and a depth camera.

[0059] In addition, the source point cloud and the destination point cloud may be generated by the same point cloud generating device. In this case, the source point cloud and the destination point cloud respectively include point clouds generated by the point cloud generating device at different times.

[0060] Alternatively, the source point cloud and the destination point cloud may be generated by different point cloud generating devices, in which case the source point cloud and the destination point cloud may include point clouds generated by different point cloud generating devices at the same time or at different times. For example, in one example, the source point cloud is generated by a laser radar, and the destination point cloud is generated by a depth camera.

[0061] Fig. 1(a) is a scene diagram applicable to the present disclosure. As shown in Fig. 1(a), in the solution provided by the present disclosure, the source point cloud and the destination point cloud can be generated by different point cloud generating devices 100, and the generation time of the source point cloud and the destination point cloud can be the same or different.

[0062] Fig. 1(b) is a scene diagram applicable to the present disclosure. As shown in Fig. 1(b), in the solution provided by the present disclosure, the point clouds generated by the point cloud generating device 100 at different times respectively constitute the source point cloud and the destination point cloud.

[0063] In addition, in the scenarios shown in FIG. 1( a ) and FIG. 1( b ), the point cloud generating device 100 is connected to the point cloud registering device 200 , and the point cloud registering device 200 is connected to the evaluating device 300 .

[0064] The connection between the point cloud generation device 100 and the point cloud registration device 200, and between the point cloud registration device 200 and the evaluation device 300 is not limited to a physical connection, and the two connection modes may be the same or different. For example, the two connection modes may both be network connections or both be electrical connections, or one may be a network connection and the other may be an electrical connection.

[0065] In this case, after receiving the source point cloud and the destination point cloud transmitted by the point cloud generating device 100, the point cloud registration device 200 may perform point cloud registration on the source point cloud and the destination point cloud, and output the point cloud registration result to the evaluation device 300. Then, the evaluation device 300 may evaluate the point cloud registration result based on the evaluation method of the point cloud registration result provided by the present disclosure.

[0066] Furthermore, in the solution provided by the present disclosure, the evaluation device 300 and the point cloud registration device 200 can be set in different electronic devices, or can also be set in the same electronic device. For example, the evaluation device 300 and the point cloud registration device 200 can both be set in a server in a vehicle (such as a vehicle-mounted smart terminal).

[0067] Exemplary Methods

[0068] Figure 2 is a flow chart of a method for evaluating point cloud registration results provided by an exemplary embodiment of the present disclosure. This embodiment can be applied to electronic devices, such as Figure 2 As shown, the following steps are included:

[0069] Step S201: performing point cloud registration on a source point cloud and a destination point cloud to determine matching pairs and non-matching pairs of the source point cloud and the destination point cloud.

[0070] Wherein, each matching pair includes a point in the source point cloud and a point in the destination point cloud, and the distance between the two points is less than or equal to the distance threshold. In addition, each matching pair includes a point in the source point cloud and a point in the destination point cloud, and the distance between the two points is greater than the distance threshold.

[0071] Step S202: based on the distance between the source point cloud and the destination point cloud in the matching pair, divide the matching pair into n matching pair sets, where n is a positive integer greater than 2.

[0072] In the solution provided by the embodiment of the present disclosure, the distance between the source point cloud and the target point cloud in the matching pair can reflect the effect of point cloud registration. The smaller the distance, the more accurate the effect of point cloud registration. In this case, different matching pair sets have different point cloud registration effects.

[0073] In a feasible example, n is 3, that is, the matching pairs are divided into 3 matching pair sets, and the distances between the source point cloud and the target point cloud are different in different matching pair sets, and the smaller the distance, the better the registration effect of the matching pair set.

[0074] Step S203: Based on the number of matching pairs in each matching pair set and the number of non-matching pairs, determine the weights corresponding to the non-matching pair set and each matching pair set.

[0075] The non-matching pair set includes the non-matching pair.

[0076] Step S204: Determine evaluation parameters based on the weights corresponding to the non-matching pair set and each matching pair set.

[0077] Step S205: Determine the evaluation result of the point cloud registration based on the evaluation parameters.

[0078] When determining the evaluation parameters through the scheme of the embodiment of the present disclosure, the matching pairs obtained after the point cloud registration of the source point cloud and the destination point cloud are divided into at least two matching pair sets, and the non-matching pair set is determined by performing point cloud registration on the source node and the destination node, and corresponding weights are assigned to each matching pair set and non-matching pair set, and then the evaluation parameters are determined based on the weights, and the evaluation results of the point cloud registration are determined through the evaluation parameters. Therefore, the evaluation parameters determined by the scheme of the embodiment of the present disclosure comprehensively consider the weights corresponding to the non-matching pair set and each matching pair set, and compared with the prior art scheme of evaluating the results of point cloud registration by the root mean square error, the evaluation parameters used in the embodiment of the present disclosure are more accurate in evaluating the results of point cloud registration.

[0079] In one embodiment of the present disclosure, the sum of the weights corresponding to the respective matching pair sets determined in step S203 is greater than the weight corresponding to the non-matching pair set.

[0080] For example, in one example, there are three matching pair sets, namely MA 1, MA 2 and MA 3, and the weights corresponding to the three matching pair sets are respectively: a1, a2 and a3, while the weight corresponding to the non-matching pair set is a4. In this case, a1+a2+a3>a4.

[0081] Since the sum of the weights corresponding to each matching pair set is greater than the weight corresponding to the non-matching pair set, the solution provided by the embodiment of the present disclosure can assign higher weights to matching pairs with better registration.

[0082] Compared with non-matching pairs, matching pairs have better registration effects. In the solution provided by the embodiment of the present disclosure, if the sum of the weights corresponding to each matching pair set is greater than the weight corresponding to the non-matching pair set, it will help the evaluation parameter to be close to the weight corresponding to each matching pair set. Accordingly, the result of point cloud registration is evaluated by the evaluation parameter, which will help to further improve the accuracy of the evaluation.

[0083] In step S202 of the present disclosure, based on the distance between the source point cloud and the destination point cloud in the matching pair, the matching pair is divided into at least two matching pair sets. In a feasible implementation, Figure 3 As shown in the above Figure 2 Based on the illustrated embodiment, step S202 may include the following steps:

[0084] Step S2021: Calculate the ratio between the target distance and the distance threshold of each matching pair respectively.

[0085] The target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair. The ratio between the target distance of each matching pair and the distance threshold is set as ratio, the distance between the source point cloud and the destination point cloud in the matching pair is d, and the distance threshold is max_range, then ratio = d / max_range.

[0086] In addition, in the point cloud registration process, matching pairs and non-matching pairs are determined according to the distance between the source point cloud and the destination point cloud. The distance between the source point cloud and the destination point cloud in each matching pair is not greater than the distance threshold, so the ratio calculated in step S2021 is a positive number not greater than 1.

[0087] Step S2022: Divide the matching pairs into n matching pair sets based on the ratio and the thresholds corresponding to each matching pair set.

[0088] In the solution provided by the embodiment of the present disclosure, corresponding thresholds may be set in advance for different sets of matching pairs, and the matching pairs may be divided based on the thresholds.

[0089] In one example, n is 2, that is, each matching pair is divided into two matching pair sets, and the threshold corresponding to the ratio is set to y. In this case, it can be determined that each matching pair with a ratio less than y is divided into one matching pair set, and each matching pair with a ratio not less than y is divided into another matching pair set.

[0090] Furthermore, since the ratio between the target distance and the distance threshold of the matching pairs in different matching pair sets is different, the point cloud registration effects of different matching pair sets are different. Generally, the smaller the ratio between the target distance and the distance threshold of the matching pairs in a matching pair set is, the smaller the distance between the source node and the destination node in the matching pair set is, and accordingly, the better the registration effect of the matching pair set is.

[0091] In another example, n is set to 3, that is, each matching pair is divided into three matching pair sets, and the thresholds corresponding to the ratios are 1 / 3 and 2 / 3 respectively. In this case, the matching pairs with a ratio less than 1 / 3 can be divided into a matching pair set (i.e., set 1), the matching pairs with a ratio not less than 1 / 3 and not greater than 2 / 3 can be divided into a matching pair set (i.e., set 2), and the matching pairs with a ratio greater than 1 / 3 can be divided into a matching pair set (i.e., set 3).

[0092] Among the three matching pair sets divided in this example, the ratio between the target distance and the distance threshold of each matching pair in set 1 is the smallest, and accordingly, the registration effect of set 1 is the best, while the ratio between the target distance and the distance threshold of each matching pair in set 3 is the largest, and accordingly, the registration effect of set 3 is relatively the worst.

[0093] Through the solution provided by the above embodiments of the present disclosure, the matching pairs can be divided into corresponding matching pair sets based on the distance between the source point cloud and the destination point cloud in the matching pairs, so as to subsequently assign corresponding weights to each matching pair set.

[0094] After determining n matching pair sets, the weights corresponding to the matching pair sets may be determined based on the number of matching pairs and the number of non-matching pairs in each matching pair set. Figure 4 As shown in the above Figure 2 Based on the illustrated embodiment, the operation may include the following steps:

[0095] Step S2031: Determine first ratios corresponding to each matching pair set based on the number of matching pairs included in each matching pair set and the ratio of the number of matching pairs.

[0096] In this step, the first ratios corresponding to the respective matching pair sets are set to ra, wherein ra1 is the first ratio corresponding to the first matching pair set among the n matching pair sets, ra2 is the first ratio corresponding to the second matching pair set among the n matching pair sets, ran is the first ratio corresponding to the nth matching pair set, N1 is the number of matching pairs included in the first matching pair set, N2 is the number of matching pairs included in the second matching pair set, and Nn is the number of matching pairs included in the nth matching pair set. Then, the following formula can be obtained:

[0097]

[0098]

[0099] …

[0100]

[0101] Step S2032: Determine a second ratio corresponding to the set of non-matching pairs based on the ratio of the number of non-matching pairs to the first value, wherein the first value is the sum of the number of all matching pairs and the number of non-matching pairs.

[0102] In this step, the second ratio corresponding to the non-matching pair set is set to RA, and Nr is the number of non-matching pairs, then the following formula can be obtained:

[0103]

[0104] Step S2033: Determine the weights corresponding to each of the matching pair sets based on the ratios of the first ratios and the second values ​​corresponding to each of the matching pair sets.

[0105] The second value is the sum of the first ratio and the second ratio corresponding to each matching pair set.

[0106] In this step, a11 is set as the weight corresponding to the first matching pair set among the n matching pair sets, a21 is set as the weight corresponding to the second matching pair set among the n matching pair sets, an1 is set as the weight corresponding to the nth matching pair set among the n matching pair sets, and sum is the second value (i.e., the sum of the first ratio and the second ratio corresponding to each matching pair set respectively), and the following formula can be obtained:

[0107] sum=ra1+ra2+……+ran+RA formula (5)

[0108]

[0109]

[0110] …

[0111]

[0112] Based on the solution provided by the above-mentioned embodiment of the present disclosure, a corresponding weight can be assigned to each matching pair set, so that the evaluation parameter can be subsequently determined based on the weight.

[0113] Furthermore, in the process of assigning weights to each matching pair set, the solution of the above embodiment utilizes the number of matching pairs in each matching pair set. The more matching pairs there are in a matching pair set, the greater the weight of the matching pair set. Accordingly, the obtained evaluation parameters will be more inclined to the matching pair set, so that the evaluation parameters can better evaluate the result of point cloud registration.

[0114] In one example, each matching pair is divided into three matching pair sets, which are set 1, set 2 and set 3. Among them, the registration effect of set 1 is the best, and the registration effect of set 3 is relatively poor. In this case, if the point cloud registration result is good, the number of matching pairs contained in set 1 is more than the number of matching pairs contained in sets 2 and 3. Accordingly, the weight corresponding to set 1 will be greater than the weights corresponding to sets 2 and 3 respectively, and the obtained evaluation parameters tend to set 1, so that the evaluation parameters can reflect the effect of point cloud registration.

[0115] In addition, in the solution provided by the present disclosure, it is also necessary to determine the weight corresponding to the non-matching pair set in step S203. Figure 2 Based on the illustrated embodiment, step S203 may include the following steps:

[0116] The weight corresponding to the non-matching pair set is determined based on the ratio of the second ratio corresponding to the non-matching pair set to the second value, wherein the second value is the sum of the first ratio and the second ratio corresponding to each matching pair set.

[0117] In this step, sum can be set as the second value (i.e., the sum of the first ratio and the second ratio corresponding to each matching pair set), and a2 can be set as the weight corresponding to the non-matching pair set, and the following formula can be obtained:

[0118]

[0119] Based on the above formula, the weight corresponding to the non-matching pair set can be determined.

[0120] By using formula (1) to formula (9), the weight corresponding to the non-matching pair set and the weight corresponding to each matching pair set can be determined respectively.

[0121] In addition, according to the above formula, it can be seen that the sum of the weights corresponding to each matching pair set can be expressed by the following formula:

[0122]

[0123] According to formula (1) to formula (3), the following formula can be obtained:

[0124] ra1+ra2+…+ran=1 formula (11).

[0125] According to formula (9) and formula (10), the following formula can be obtained:

[0126]

[0127] According to formula (4) for calculating RA, RA is a positive number less than 1. Therefore, according to formula (8) and formula (12) for calculating the weight a2 corresponding to the non-matching pair set, it can be seen that the sum of the weights corresponding to each matching pair set is greater than the weight corresponding to the non-matching degree set. The weights corresponding to each matching pair set and the weights corresponding to the non-matching pair set calculated based on the schemes of formula (1) to formula (9) meet the requirements of the embodiments of the present disclosure.

[0128] After determining the weights corresponding to the assignments of the matching pair sets and the weights corresponding to the non-matching pair sets, the evaluation parameters may be determined based on the weights corresponding to the assignments of the matching pair sets and the weights corresponding to the non-matching pair sets. Figure 2 Based on the illustrated embodiment, the operation may include the following steps:

[0129] The evaluation parameters are determined by the following formula:

[0130]

[0131] Among them, RWMSE is the evaluation parameter, a11 is the weight corresponding to the first matching pair set in the n matching pair sets, N1 is the number of matching pairs contained in the first matching pair set, d is the target distance, di is the target distance of the i-th matching pair in the first matching pair set, i is a positive integer not greater than N1, an1 is the weight corresponding to the n-th matching pair set in the n matching pair sets, Nn is the number of matching pairs contained in the n-th matching pair set, ds is the target distance of the s-th matching pair in the n-th matching pair set, s is a positive integer not greater than Nn, a2 is the weight corresponding to the non-matching pair set, Nr is the number of non-matching pairs, dt is the target distance of the t-th non-matching pair, t is a positive integer not greater than Nr, the target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair, and the target distance of the non-matching pair is the distance between the source point cloud and the destination point cloud in the non-matching pair.

[0132] Based on formula (13), the evaluation parameters for evaluating the point cloud registration results can be obtained. In addition, the evaluation parameters determined by formula (13) are adjusted based on the root mean square error RMSE, and corresponding weights are set for each matching pair set and non-matching pair set. In this case, the evaluation parameter can also be called the weighted root mean square error (rootweight mean square error, RWMSE).

[0133] If the point cloud registration result is good, then usually, the number of matching pairs included in the matching pair set that represents the better point cloud registration result is large, and since the weight corresponding to the matching pair set is related to the number of matching pairs included in the matching pair set, the weight corresponding to the matching pair set is high. In addition, in the matching pair set with a good point cloud registration result, the target distance of each matching pair is small. Therefore, if the point cloud registration result is evaluated by the evaluation parameter determined by formula (13), when the source node and the destination node are the same, the smaller the evaluation parameter, the more accurate the point cloud registration result.

[0134] In actual application scenarios, each point in the source point cloud may correspond to different semantics, and correspondingly, each point in the destination node may also correspond to different semantics. Figure 5 As shown in the above Figure 2 Based on the illustrated embodiment, step S201 may include the following steps:

[0135] Step S2011, determine the semantics of each point in the source point cloud and each point in the destination point cloud.

[0136] In this step, the semantics of each point in the source point cloud and each point in the destination point cloud can be determined by performing semantic segmentation on the source point cloud and the destination point cloud respectively.

[0137] Step S2012: Based on semantics, the source point cloud and the destination point cloud are classified respectively, and the semantics of each point in the same type are the same.

[0138] That is, in this step, based on the semantics of each point in the source node, each point in the source node is classified, and each point with the same semantics is classified into the same class. Correspondingly, in this step, based on the semantics of each point in the destination node, each point in the destination node is classified, and each point with the same semantics is classified into the same class.

[0139] Step S2013: perform point cloud registration on the source point cloud and the destination point cloud of the same type, and obtain matching pairs and non-matching pairs of the source point cloud and the destination point cloud of the same type.

[0140] Generally, when the semantics of two points are the same, the possibility of the two points matching is high. Therefore, in step S2013, the source point cloud and the target point cloud of the same type are registered to improve the accuracy of point cloud registration.

[0141] For example, the semantics of each point in the source point cloud and the destination point cloud include: trees, road edges and buildings. When performing point cloud registration, the points in the source point cloud with the semantics of trees can be registered with the points in the destination point cloud with the semantics of trees, the points in the source point cloud with the semantics of road edges can be registered with the points in the destination point cloud with the semantics of road edges, and the points in the source point cloud with the semantics of buildings can be registered with the points in the destination point cloud with the semantics of buildings. In this way, matching pairs and non-matching pairs corresponding to the type of trees can be obtained, as well as matching pairs and non-matching pairs corresponding to the type of road edges and matching pairs and non-matching pairs corresponding to the type of buildings can be obtained.

[0142] Through the above steps of the embodiment of the present disclosure, it is possible to classify each point in the source point cloud and the destination point cloud based on semantics, and perform point cloud registration on the source point cloud and the destination point cloud of the same type, thereby improving the accuracy of point cloud registration.

[0143] In addition, since the matching pairs and non-matching pairs corresponding to each type are obtained through step S2013, the evaluation parameters determined based on the matching pairs and non-matching pairs corresponding to a certain type can only be used to evaluate the point cloud registration results of that type. Therefore, step S25 (i.e., the operation of determining the evaluation result of the point cloud registration based on the evaluation parameters) provided in the embodiment of the present disclosure generally includes the following operations:

[0144] Based on the evaluation parameters corresponding to the matching pairs and the non-matching pairs of the source point cloud and the destination point cloud of the same type, an evaluation result of the point cloud registration of the source point cloud and the destination point cloud of the same type is determined.

[0145] That is to say, if the source point cloud and the destination point cloud are divided into multiple types by classifying them, the evaluation parameters corresponding to each type can be determined respectively through the operations of steps S201 to S205, and based on the evaluation parameters, the results of point cloud registration of each type can be evaluated respectively.

[0146] In view of this situation, the embodiment of the present disclosure further includes the following steps:

[0147] After the source point cloud and the destination point cloud are classified based on semantics, weights corresponding to the source point cloud or the destination point cloud of each type are determined based on the semantics.

[0148] In this step, the correspondence between the type and the weight can be preset, and then based on the correspondence, the weights corresponding to the source point cloud or the destination point cloud of each type can be determined. Generally, when setting the correspondence, a higher weight can be set for the source point cloud or the destination point cloud of a more concerned type.

[0149] In one example, the source point cloud and the destination point cloud are point clouds obtained for the vehicle's surrounding environment and are used to assist the vehicle's assisted driving. In this case, pedestrians on the road may interfere with the vehicle's driving behavior. Therefore, if the type of the source point cloud or the destination point cloud includes pedestrians, a higher weight can be set for the source point cloud or destination point cloud corresponding to the pedestrians.

[0150] Correspondingly, such as Figure 6 As shown in the above Figure 2 Based on the embodiment shown, the following steps may also be included:

[0151] Step S206: After determining the evaluation result of the point cloud registration based on the evaluation parameters, the evaluation indicators of different types of source point clouds and destination point clouds are determined based on the weights corresponding to each type of source point cloud or destination point cloud, and the evaluation parameters. The evaluation indicators are used to evaluate the registration results of different types of the source point cloud and the destination point cloud.

[0152] Step S207: Determine the evaluation results of point cloud registration of different types of source point clouds and target point clouds based on the evaluation indicators.

[0153] In the solution provided by the embodiment of the present disclosure, the evaluation parameters corresponding to a certain type of point cloud can be used to evaluate the point cloud registration results of the source point cloud of that type and the destination point cloud. In this case, the evaluation index can be determined based on the weights corresponding to each type of source point cloud or destination point cloud, and the evaluation parameters corresponding to each type, and the evaluation index can be used to evaluate the point cloud registration results of each type of source point cloud and destination point cloud.

[0154] In one example, the types of the source point cloud and the destination point cloud include: trees, road edges and buildings respectively. The source point cloud of the tree type can be point cloud registered with the destination point cloud to obtain corresponding evaluation parameters (i.e., first evaluation parameters), and the source point cloud of the road edge type can be point cloud registered with the destination point cloud to obtain corresponding evaluation parameters (i.e., second evaluation parameters), and the source point cloud of the building type can be point cloud registered with the destination point cloud to obtain corresponding evaluation parameters (i.e., third evaluation parameters). Then, based on the weights corresponding to each type of source point cloud or destination point cloud, as well as the first evaluation parameter, the second evaluation parameter and the third evaluation parameter, an evaluation index is determined, which can evaluate the point cloud registration results of the three types of source point clouds and destination point clouds.

[0155] In a feasible implementation of the present disclosure, determining evaluation indicators of different types of source point clouds and destination point clouds may include the following steps:

[0156] The evaluation indexes of the source point cloud and the destination point cloud of different types are determined by the following formula:

[0157]

[0158] Among them, W is the evaluation index, m is the number of types of the source point cloud or the destination point cloud, m is a positive integer, A is the weight corresponding to the source point cloud or the destination point cloud of each type, Ak is the weight corresponding to the k-th type of the source point cloud or the destination point cloud, k is a positive integer not greater than m, R is the evaluation parameter, and Rk is the evaluation parameter corresponding to the k-th type of the source point cloud or the destination point cloud.

[0159] In this case, after determining the evaluation parameters corresponding to each type of source point cloud and destination point cloud, as well as the weights corresponding to each type of source point cloud or destination point cloud, the corresponding evaluation index can be obtained based on formula (14).

[0160] Exemplary Devices

[0161] Figure 71 is a structural diagram of a point cloud registration result evaluation device provided by an exemplary embodiment of the present disclosure. The point cloud registration result evaluation device can be set in an electronic device such as a terminal device, a server, or an object such as a vehicle to execute the point cloud registration result evaluation method of any of the above embodiments of the present disclosure. Figure 7 As shown, the device for evaluating the point cloud registration result of this embodiment includes: a registration module 201 , a set division module 202 , a weight determination module 203 , a parameter determination module 204 and an evaluation module 205 .

[0162] The registration module 201 is used to determine matching pairs and non-matching pairs of the source point cloud and the destination point cloud by performing point cloud registration on the source point cloud and the destination point cloud;

[0163] The set division module 202 is used to divide the matching pairs into n matching pair sets based on the distance between the source point cloud and the destination point cloud in the matching pairs determined by the registration module 201, where n is a positive integer greater than 2;

[0164] A weight determination module 203, configured to determine weights corresponding to a non-matching pair set and each of the matching pair sets, respectively, based on the number of the matching pairs in each of the matching pair sets determined by the set division module 202 and the number of the non-matching pairs determined by the registration module 201, wherein the non-matching pair set includes the non-matching pairs;

[0165] A parameter determination module 204, configured to determine an evaluation parameter based on the weights corresponding to the non-matching pair set and each of the matching pair sets determined by the weight determination module 203;

[0166] The evaluation module 205 is used to determine the result of the point cloud registration based on the evaluation parameters determined by the parameter determination module 204.

[0167] When determining the evaluation parameters, the device provided by the embodiment of the present disclosure first divides the matching pairs obtained after point cloud registration of the source point cloud and the destination point cloud into at least two matching pair sets, and determines the non-matching pair set by performing point cloud registration on the source node and the destination node, and assigns corresponding weights to each matching pair set and non-matching pair set. Among them, the sum of the weights corresponding to each matching pair set is greater than the weight corresponding to the non-matching pair set. The registration effect of each matching pair set is better than the registration effect of the non-matching pair set. Therefore, the evaluation parameters determined by the embodiment of the present disclosure can assign higher weights to matching pairs with better registration, and the evaluation of the point cloud registration results is more accurate compared with the prior art through the root mean square error.

[0168] In a feasible example, the sum of the weights corresponding to the respective matching pair sets determined by the weight determination module 203 is greater than the weight corresponding to the non-matching pair set determined by the weight determination module 203 .

[0169] For further information, see Figure 8 As shown in the structural schematic diagram, in a feasible example, the set partitioning module 202 includes: a computing unit 2021 and a partitioning unit 2022.

[0170] The calculation unit 2021 is used to respectively calculate the ratio between the target distance of each matching pair and the distance threshold, wherein the target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair;

[0171] The division unit 2022 is used to divide the matching pairs into the n matching pair sets based on the ratio calculated by the calculation unit 2021 and the thresholds corresponding to the respective matching pair sets.

[0172] In another feasible example, the weight determination module 203 includes: a first ratio determination unit 2031 , a second ratio determination unit 2032 , and a first weight determination unit 2033 .

[0173] The first ratio determination unit 2031 is used to determine the first ratios corresponding to the respective matching pair sets based on the ratios of the number of the matching pairs included in the respective matching pair sets to the number of all matching pairs;

[0174] The second ratio determination unit 2032 is used to determine a second ratio corresponding to the non-matching pair set based on a ratio of the number of the non-matching pairs to a first value, where the first value is the sum of the number of all matching pairs and the number of the non-matching pairs;

[0175] The weight determination unit 2033 is used to determine the weights corresponding to each of the matching pair sets based on the ratio of the first ratio corresponding to each of the matching pair sets to the second value, wherein the second value is the sum of the first ratio and the second ratio corresponding to each of the matching pair sets.

[0176] In another feasible example, the weight determination module 203 includes: a second weight determination unit 2034 .

[0177] The second weight determination unit 2034 is used to determine the weight corresponding to the non-matching pair set based on the ratio of the second ratio corresponding to the non-matching pair set to the second value, wherein the second value is the sum of the first ratio and the second ratio corresponding to each of the matching pair sets.

[0178] In addition, in a feasible example, the parameter determination module 204 is used to determine the evaluation parameter by the following formula:

[0179]

[0180] Among them, RWMSE is the evaluation parameter, a11 is the weight corresponding to the first matching pair set in the n matching pair sets, N1 is the number of matching pairs contained in the first matching pair set, d is the target distance, di is the target distance of the i-th matching pair in the first matching pair set, i is a positive integer not greater than N1, an1 is the weight corresponding to the n-th matching pair set in the n matching pair sets, Nn is the number of matching pairs contained in the n-th matching pair set, ds is the target distance of the s-th matching pair in the n-th matching pair set, s is a positive integer not greater than Nn, a2 is the weight corresponding to the non-matching pair set, Nr is the number of non-matching pairs, dt is the target distance of the t-th non-matching pair, t is a positive integer not greater than Nr, the target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair, and the target distance of the non-matching pair is the distance between the source point cloud and the destination point cloud in the non-matching pair.

[0181] The source point cloud and the target point cloud that need to be registered often include points with multiple semantics. In view of this situation, in another feasible example of the present disclosure, the registration module 201 includes: a semantic determination unit 2011, a classification unit 2012 and a registration unit 2013.

[0182] The semantic determination unit 2011 is used to determine the semantics of each point in the source point cloud and each point in the destination point cloud;

[0183] The classification unit 2012 is used to classify the source point cloud and the destination point cloud respectively based on the semantics determined by the semantics determination unit 2011, and the semantics of each point in the same type are the same;

[0184] The registration unit 2013 is used to perform point cloud registration on the source point cloud and the destination point cloud of the same type classified by the classification unit 2012, and obtain matching pairs and non-matching pairs of the source point cloud and the destination point cloud of the same type.

[0185] Correspondingly, the point cloud registration result evaluation device provided in the embodiment of the present disclosure also includes: a type weight determination module.

[0186] The type weight determination module is used to determine the weights corresponding to the source point cloud or the destination point cloud of each type based on the semantics determined by the semantic determination unit 2011.

[0187] In this case, the point cloud registration result evaluation device provided in the embodiment of the present disclosure further includes: an evaluation index determination module.

[0188] Among them, the evaluation index determination module is used to, after determining the evaluation result of the point cloud registration based on the evaluation parameters, determine the evaluation indexes of different types of the source point cloud and the destination point cloud based on the weights corresponding to the respective types of the source point cloud or the destination point cloud determined by the type weight determination module and the evaluation parameters determined by the evaluation parameter determination module, and the evaluation indexes are used to evaluate the registration results of different types of the source point cloud and the destination point cloud.

[0189] In this case, the evaluation module 205 may also be used to determine the evaluation results of point cloud registration between different types of source point clouds and target point clouds based on the evaluation index.

[0190] In addition, in a feasible example, the evaluation index determination module may determine the evaluation indexes of the different types of the source point cloud and the target point cloud by the following formula:

[0191]

[0192] Among them, W is the evaluation index, m is the number of types of the source point cloud or the destination point cloud, m is a positive integer, A is the weight corresponding to the source point cloud or the destination point cloud of each type, Ak is the weight corresponding to the k-th type of the source point cloud or the destination point cloud, k is a positive integer not greater than m, R is the evaluation parameter, and Rk is the evaluation parameter corresponding to the k-th type of the source point cloud or the destination point cloud.

[0193] Exemplary Electronic Devices

[0194] Below, reference Fig. 9 To describe an electronic device according to an embodiment of the present application. In an exemplary embodiment of the present disclosure, the electronic device may be an electronic device including the evaluation device 300 shown in FIG. 1(a) or FIG. 1(b), or, in another exemplary embodiment of the present disclosure, the electronic device may be an electronic device including the point cloud generation device 100, the point cloud registration device 200 and the evaluation device 300 shown in FIG. 1(a) or FIG. 1(b). Of course, the electronic device may also be in other forms, which is not limited by the present disclosure.

[0195] Fig. 9 A block diagram of an electronic device according to an embodiment of the present application is illustrated.

[0196] like Fig. 9 As shown, the electronic device 11 includes one or more processors 111 and a memory 112 .

[0197] The processor 111 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 11 to perform desired functions.

[0198] The memory 112 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 111 may execute the program instructions to implement the evaluation of the point cloud registration results of the various embodiments of the present application described above and / or other desired functions. Various contents such as source point cloud, destination point cloud, evaluation parameters, etc. may also be stored in the computer-readable storage medium.

[0199] In one example, the electronic device 11 may further include: an input device 113 and an output device 114 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0200] In addition, the input device 113 may also include, for example, a keyboard, a mouse, and the like.

[0201] The output device 114 can output various information to the outside, including the determined evaluation parameters, evaluation indicators, etc. The output device 114 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.

[0202] Of course, to simplify, Fig. 9 Only some of the components related to the present application in the electronic device 11 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 11 may also include any other appropriate components.

[0203] Exemplary computer program products and computer-readable storage media

[0204] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method for evaluating point cloud registration results according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.

[0205] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0206] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, causes the processor to execute the steps of the method for evaluating point cloud registration results according to various embodiments of the present application described in the above “Exemplary Method” section of this specification.

[0207] The computer readable storage medium may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0208] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0209] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0210] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0211] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0212] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for evaluating point cloud registration results, wherein: include: By performing point cloud registration on the source point cloud and the destination point cloud, matching pairs and non-matching pairs of the source point cloud and the destination point cloud are determined; Based on the distance between the source point cloud and the destination point cloud in the matching pairs, the matching pairs are divided into n matching pair sets, where n is a positive integer greater than 2; Determining weights corresponding to a non-matching pair set and each of the matching pair sets based on the number of the matching pairs in each of the matching pair sets and the number of the non-matching pairs, wherein the non-matching pair set includes the non-matching pairs; Determining an evaluation parameter based on the weights corresponding to the non-matching pair set and each of the matching pair sets; Based on the evaluation parameters, determining an evaluation result of the point cloud registration; The step of determining the evaluation parameters based on the weights corresponding to the respective matching pair sets and the non-matching pair sets includes: The evaluation parameters are determined by the following formula: ； Among them, RWMSE is the evaluation parameter, a11 is the weight corresponding to the first matching pair set in the n matching pair sets, N1 is the number of matching pairs contained in the first matching pair set, d is the target distance, di is the target distance of the i-th matching pair in the first matching pair set, i is a positive integer not greater than N1, an1 is the weight corresponding to the n-th matching pair set in the n matching pair sets, Nn is the number of matching pairs contained in the n-th matching pair set, ds is the target distance of the s-th matching pair in the n-th matching pair set, s is a positive integer not greater than Nn, a2 is the weight corresponding to the non-matching pair set, Nr is the number of non-matching pairs, dt is the target distance of the t-th non-matching pair, t is a positive integer not greater than Nr, the target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair, and the target distance of the non-matching pair is the distance between the source point cloud and the destination point cloud in the non-matching pair.

2. The method according to claim 1, wherein: The sum of the weights corresponding to each of the matching pair sets is greater than the weight corresponding to the non-matching pair set.

3. The method according to claim 1, wherein: The dividing the matching pairs into n matching pair sets based on the distance between the source point cloud and the destination point cloud in the matching pairs comprises: Calculating the ratio between the target distance of each matching pair and the distance threshold respectively, wherein the target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair; The matching pairs are divided into the n matching pair sets based on the ratio and the thresholds corresponding to the respective matching pair sets.

4. The method according to claim 1, wherein: The determining, based on the number of the matching pairs in each of the matching pair sets and the number of the non-matching pairs, the weights corresponding to each of the matching pair sets respectively include: Determine first ratios corresponding to each of the matching pair sets based on the ratios of the number of the matching pairs included in each of the matching pair sets to the number of all matching pairs; Determining a second ratio corresponding to the set of non-matching pairs based on a ratio of the number of non-matching pairs to a first value, wherein the first value is the sum of the number of all matching pairs and the number of non-matching pairs; Based on the ratio of the first ratio and the second value respectively corresponding to each of the matching pair sets, the weights respectively corresponding to each of the matching pair sets are determined, wherein the second value is the sum of the first ratio and the second ratio respectively corresponding to each of the matching pair sets.

5. The method according to claim 4, wherein: The determining, based on the number of the matching pairs in each of the matching pair sets and the number of the non-matching pairs, the weight corresponding to the non-matching pair set includes: The weight corresponding to the non-matching pair set is determined based on the ratio of the second ratio corresponding to the non-matching pair set to the second value, wherein the second value is the sum of the first ratio and the second ratio respectively corresponding to each of the matching pair sets.

6. The method according to any one of claims 1 to 5, wherein: The step of performing point cloud registration on the source point cloud and the destination point cloud to determine matching pairs and non-matching pairs of the source point cloud and the destination point cloud includes: Determining the semantics of each point in the source point cloud and each point in the destination point cloud; Based on the semantics, the source point cloud and the destination point cloud are classified respectively, and the semantics of each point in the same type are the same; Performing point cloud registration on the source point cloud and the destination point cloud of the same type to obtain matching pairs and non-matching pairs of the source point cloud and the destination point cloud of the same type; The step of determining an evaluation result of the point cloud registration based on the evaluation parameter comprises: Based on the evaluation parameters corresponding to the matching pairs and the non-matching pairs of the source point cloud and the destination point cloud of the same type, an evaluation result of the point cloud registration of the source point cloud and the destination point cloud of the same type is determined.

7. The method according to claim 6, wherein: After classifying the source point cloud and the destination point cloud respectively based on the semantics, the method further includes: Based on the semantics, determining weights corresponding to the source point cloud or the destination point cloud of each type; After determining the evaluation result of the point cloud registration based on the evaluation parameters, the method further includes: Based on the weights corresponding to the source point clouds or the destination point clouds of the various types, and the evaluation parameters, determining evaluation indicators of different types of the source point clouds and the destination point clouds, wherein the evaluation indicators are used to evaluate the registration results of different types of the source point clouds and the destination point clouds; Based on the evaluation index, an evaluation result of the point cloud registration of the source point cloud and the destination point cloud of the different types is determined.

8. The method according to claim 7, wherein: The determining of evaluation indicators of different types of the source point cloud and the destination point cloud includes: The evaluation indexes of the source point cloud and the destination point cloud of different types are determined by the following formula: ； Among them, W is the evaluation index, m is the number of types of the source point cloud or the destination point cloud, m is a positive integer, A is the weight corresponding to the source point cloud or the destination point cloud of each type, Ak is the weight corresponding to the k-th type of the source point cloud or the destination point cloud, k is a positive integer not greater than m, R is the evaluation parameter, and Rk is the evaluation parameter corresponding to the k-th type of the source point cloud or the destination point cloud.

9. A device for evaluating point cloud registration results, wherein: include: A registration module, used for determining matching pairs and non-matching pairs of the source point cloud and the destination point cloud by performing point cloud registration on the source point cloud and the destination point cloud; a set division module, configured to divide the matching pairs into n matching pair sets based on the distance between the source point cloud and the destination point cloud in the matching pairs determined by the registration module, wherein n is a positive integer greater than 2; a weight determination module, configured to determine weights corresponding to a non-matching pair set and each of the matching pair sets, respectively, based on the number of the matching pairs in each of the matching pair sets determined by the set division module and the number of the non-matching pairs determined by the registration module, wherein the non-matching pair set includes the non-matching pairs; a parameter determination module, configured to determine an evaluation parameter based on the weights respectively corresponding to the non-matching pair set and each of the matching pair sets determined by the weight determination module; An evaluation module, configured to determine a result of the point cloud registration based on the evaluation parameters determined by the parameter determination module; The parameter determination module is specifically used to determine the evaluation parameter by the following formula: ； Among them, RWMSE is the evaluation parameter, a11 is the weight corresponding to the first matching pair set in the n matching pair sets, N1 is the number of matching pairs contained in the first matching pair set, d is the target distance, di is the target distance of the i-th matching pair in the first matching pair set, i is a positive integer not greater than N1, an1 is the weight corresponding to the n-th matching pair set in the n matching pair sets, Nn is the number of matching pairs contained in the n-th matching pair set, ds is the target distance of the s-th matching pair in the n-th matching pair set, s is a positive integer not greater than Nn, a2 is the weight corresponding to the non-matching pair set, Nr is the number of non-matching pairs, dt is the target distance of the t-th non-matching pair, t is a positive integer not greater than Nr, the target distance of the matching pair is the distance between the source point cloud and the destination point cloud in the matching pair, and the target distance of the non-matching pair is the distance between the source point cloud and the destination point cloud in the non-matching pair.

10. A computer-readable storage medium storing a computer program, wherein the computer program is used to execute the point cloud registration result evaluation method described in any one of claims 1 to 8.

11. An electronic device, comprising: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the point cloud registration result evaluation method described in any one of claims 1-8 above.

Images