Precise scanning point cloud positioning method based on CAD model point cloud constraint

Abstract

The invention discloses a precise scanning point cloud positioning method based on CAD model point cloud constraint, and relates to the technical field of three-dimensional scanning measurement. The problems that noise exists in point cloud and coordinate deviation exists in the two-dimensional-to-three-dimensional mapping process of the feature matching process, and consequently the positioning accuracy is low in a traditional positioning method for determining the hole center of the porous part through the structured light scanning technology are solved. According to the method, a CAD model point cloud carries an interested subset point cloud to match a target point cloud, then the interested subset point cloud of the target point cloud is extracted according to the interested subset point cloud, a centroid coordinate is calculated according to the coordinate of each interested subset point cloud of the extracted target point cloud, and hole center positioning is completed. The method is mainly used for positioning the hole center of the porous solid part.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional scanning measurement. Background Art

[0002] When using structured light scanning technology to scan a porous part entity and generate a point cloud (referred to as a scanned point cloud), and using the scanned point cloud to accurately measure its detailed features, specifically, when positioning the centers of the holes of the porous part, it faces the problem of measurement noise interference, such as Figure 1 as shown Figure 1 Based on structured light scanning technology, the point cloud generated by scanning a physical part contains noise, which affects the measurement accuracy.

[0003] In order to accurately measure important features and achieve precise positioning of the holes of the porous part, it is necessary to extract the subset of interest to be measured from the scanned point cloud. For example, when measuring Figure 2 the distance between the centers of two circular holes in, it is necessary to use the circular feature to match the point cloud and obtain the point cloud subset located in the edge area of the two circular holes, and finally calculate the center distance of the two point cloud subsets. Two problems will arise in such a process: First, when the structured light device scans and generates a point cloud, due to material and environmental disturbances, noise will be generated, interfering with the matching and measurement, resulting in low positioning accuracy.

[0004] Second, in the process of feature matching on the target point cloud, when using the feature matching method based on object detection to mark the hole center, there is a coordinate deviation problem when mapping the marked position in the two-dimensional image to the actual three-dimensional space position. Therefore, in addition to the subset of interest to be measured, there are other parts of the point cloud subset that better conform to the matching features, resulting in feature matching failure and positioning failure. Therefore, the above problems need to be solved urgently. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem that the traditional method of using structured light scanning technology to determine the hole center positioning of porous parts has noise in the point cloud and coordinate deviation in the two-dimensional to three-dimensional mapping process during the feature matching process, resulting in low positioning accuracy. The present invention provides a precise scanned point cloud positioning method based on CAD model point cloud constraint.

[0006] The precise scanned point cloud positioning method based on CAD model point cloud constraint includes:

[0007] S1. Perform mesh sampling on the CAD model of the porous solid part to generate a CAD model point cloud, and mark each subset of interest point cloud, where each subset of interest point cloud corresponds to the area where the corresponding hole on the porous solid part is located; at the same time, scan the porous solid part to generate a noisy target point cloud;

[0008] S2. Use the CAD model point cloud to guide the denoising of the noisy target point cloud;

[0009] S3. Use the matching algorithm to transform the CAD model point cloud into the coordinate system of the denoised target point cloud, realizing the rough alignment of the denoised target point cloud and the CAD model point cloud; perform local feature matching between the point clouds of each interested subset on the CAD model point cloud and the corresponding regions on the denoised target point cloud, and extract the point clouds of the corresponding regions on the target point cloud corresponding to the point clouds of each interested subset on the CAD model point cloud, and the extracted point clouds of each corresponding region are used as an interested subset point cloud of the target point cloud;

[0010] S4. Use the coordinates of the extracted point clouds of each corresponding region to calculate the centroid coordinates of the point cloud of this corresponding region; and use the centroid coordinates of the point clouds of each corresponding region as the geometric center coordinates of the corresponding holes of the porous solid part, completing the center positioning of each hole on the porous solid part.

[0011] Preferably, perform mesh sampling on the CAD model of the porous solid part, which is implemented using the open3d random sampling function.

[0012] Preferably, in step S2, the implementation method of using the CAD model point cloud to guide the denoising of the target point cloud includes:

[0013] Use the ICP algorithm to realize the matching between the CAD model point cloud and the noisy target point cloud. After successful matching, use mean filtering to filter out the point cloud in the noisy target point cloud except for the region successfully matched with the CAD model point cloud and the point cloud within a preset distance from the successfully matched region. Repeat the above process until the number of point clouds filtered out during the filtering process is lower than the preset number of point clouds, completing the denoising.

[0014] Preferably, the preset distance is 0.01 mm.

[0015] Preferably, in step S3, the point clouds of the corresponding regions on the extracted target point cloud include: the regions corresponding to the point clouds of each interested subset on the CAD model point cloud and the regions within 0.005 mm from the corresponding regions.

[0016] The precise scanning point cloud positioning device based on the CAD model point cloud constraint includes a storage device, a processor, and a computer program stored in the storage device and executable on the processor. The processor executes the computer program to implement the precise scanning point cloud positioning method based on the CAD model point cloud constraint as described.

[0017] A computer-readable storage device stores a computer program, and when the computer program is executed, it implements the precise scanning point cloud positioning method based on the CAD model point cloud constraint as described.

[0018] A computer program product includes a computer program which, when executed by a processor, implements the described precise scanning point cloud positioning method based on CAD model point cloud constraints.

[0019] Advantages of the present invention:

[0020] The precise scanning point cloud positioning method based on CAD model point cloud constraints proposed by the present invention can remove noise efficiently and with high quality, and guide the feature matching process of the point cloud, so as to correctly and accurately extract the point cloud features for feature measurement, and improve the positioning accuracy of each hole of the porous solid part. The specific advantages are as follows:

[0021] (1) Utilize the CAD model of the porous solid part to guide the denoising of the target point cloud (physical point cloud), and improve the accuracy of the target point cloud model;

[0022] (2) Can ensure the correctness of the feature extraction area. The CAD model point cloud carries the point cloud of the interested subset to match the target point cloud (physical point cloud), and then extracts the point cloud of the interested subset of the target point cloud according to the point cloud of the interested subset. Ensure the correctness of the extraction position of the point cloud of the interested subset of the target point cloud, and improve the positioning accuracy;

[0023] (3) Due to the higher accuracy of the target point cloud model and the correctness of the feature extraction area, the target point cloud can better fit the feature to be matched (the point cloud of the interested subset of the CAD model point cloud) in the feature matching stage, and can improve the accuracy of the target point cloud subset extraction.

[0024] (4) The present invention has high versatility and can be easily promoted. It is not only suitable for the spherical porous part in the example. As long as the target point cloud (physical point cloud) of any shaped porous part is obtained and the point cloud of each interested subset on the CAD model point cloud is marked, the corresponding interested area on the target point cloud of the porous part can be accurately extracted. Description of the drawings

[0025] Figure 1 is a schematic diagram of the point cloud generated by scanning a physical porous part based on the structured light scanning technology in the prior art;

[0026] Figure 2 is for Figure 1 the schematic diagram of marking the point cloud of the physical porous part in

[0027] Figure 3 is the flowchart of the precise scanning point cloud positioning method based on CAD model point cloud constraints described in the present invention;

[0028] Figure 4 is the CAD model of the porous solid part;

[0029] Figure 5 isFigure 4 Schematic diagram of the point cloud of the CAD model;

[0030] Figure 6 is Figure 5 The point cloud of the CAD model after marking;

[0031] Figure 7 is the point cloud of the marked subset of interest;

[0032] Figure 8 is the target point cloud after denoising;

[0033] Figure 9 is the effect diagram of the rough alignment between the target point cloud after denoising and the point cloud of the CAD model;

[0034] Figure 10 is the schematic diagram of the local feature matching between each subset of interest point cloud on the CAD model and the corresponding area on the target point cloud after denoising;

[0035] Figure 11 is the schematic diagram of the point cloud of the corresponding area on the target point cloud extracted corresponding to each subset of interest point cloud;

[0036] Figure 12 is the three-dimensional structure schematic diagram of the spotlight experiment target. Detailed implementation manners

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0038] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0039] In order to solve the problems existing in the traditional method of using structured light scanning technology to determine the hole center positioning of porous parts, such as the presence of noise in the point cloud and coordinate deviation in the two-dimensional to three-dimensional mapping process during the feature matching process, resulting in low positioning accuracy, a precise scanning point cloud positioning method based on CAD model point cloud constraint is proposed, specifically as follows:

[0040] Detailed implementation manner 1. Refer to Figure 3 To illustrate this implementation manner, the precise scanning point cloud positioning method based on CAD model point cloud constraint described in this implementation manner includes:

[0041] S1. Perform mesh sampling on the CAD model of the porous solid part to generate the CAD model point cloud, and mark the point clouds of each interested subset. Among them, the point clouds of each interested subset correspond to the regions where the corresponding holes on the porous solid part are located; at the same time, scan the porous solid part to generate the noisy target point cloud;

[0042] During application, perform mesh sampling on the CAD model of the porous solid part, which is implemented using the open3d random sampling function;

[0043] S2. Use the CAD model point cloud to guide the denoising of the noisy target point cloud. Specifically: Implement the matching of the CAD model point cloud and the noisy target point cloud using the ICP algorithm. After successful matching, use mean filtering to filter out the point cloud in the noisy target point cloud except for the region that matches the CAD model point cloud and the point cloud within a preset distance from the matching successful region. Repeat the above process until the number of point clouds filtered during the filtering process is lower than the preset number of point clouds to complete the denoising; During application, the preset distance is best at 0.01 mm;

[0044] S3. Use the matching algorithm to transform the CAD model point cloud into the coordinate system where the denoised target point cloud is located to achieve rough alignment between the denoised target point cloud and the CAD model point cloud;

[0045] Perform local feature matching between the point clouds of each interested subset on the CAD model point cloud and the corresponding regions on the denoised target point cloud, and extract the point cloud of the corresponding region on the target point cloud corresponding to the point clouds of each interested subset on the CAD model point cloud. And the extracted point clouds of each corresponding region are used as an interested subset point cloud of the target point cloud; During specific application, the extracted point cloud of the corresponding region on the target point cloud includes: the region corresponding to the point clouds of each interested subset on the CAD model point cloud and the region within 0.005 mm from the corresponding region.

[0046] S4. Use the coordinates of the point clouds of each corresponding region to calculate the centroid coordinates of the point cloud of the corresponding region; and use the centroid coordinates of the point clouds of each corresponding region as the geometric center coordinates of the corresponding holes on the porous solid part to complete the center positioning of each hole on the porous solid part.

[0047] This embodiment proposes an accurate processing method based on CAD digital model constraints, which can remove noise efficiently and with high quality, and guide the feature matching process of the point cloud, so as to correctly and accurately extract the point cloud features for feature measurement.

[0048] Verification test:

[0049] During a micro-optical atomic experiment, a micro light energy absorption substance is placed at the center of the condensing experiment target, and multiple high-energy light beams are guided to precisely enter the centers of multiple holes of the condensing experiment target, so that the multiple high-energy light beams are concentrated on the substance at the center of the part, enabling the atoms of the central substance to absorb energy and thus undergo structural changes, thereby completing the micro-optical atomic experiment. During the micro-optical atomic experiment, it is necessary to accurately locate the central position coordinates of each hole of the condensing experiment target to ensure that the multiple high-energy light beams are precisely focused on the central substance and the micro-optical atomic experiment is successfully completed. Figure 12 In [reference], the porous part is used as the condensing experiment target. Specifically, taking the porous part as the condensing experiment target as an example, the principle and effect are described as follows:

[0050] 1) Generating point cloud by CAD digital model sampling

[0051] Using a CAD digital model as shown in Figure 4 to perform grid sampling to generate the CAD model point cloud. The sampling uses the random sampling function of open3d, and its internal algorithm conforms to the Poisson distribution, as shown in Figure 5 . And mark the subset point cloud of interest on the CAD model point cloud as the area to be feature-matched, as shown in Figure 6 and Figure 7 .

[0052] 2) Denoising the noisy target point cloud

[0053] The ICP algorithm is used to realize the matching between the CAD model point cloud and the measured point cloud of the noisy physical part as shown in Figure 5 . After matching, mean filtering is used to retain the points in the noisy target point cloud except for the area that successfully matches the CAD model point cloud and the target point cloud at a distance of 0.01 mm from the edge point cloud of the successfully matched area. This process is looped multiple times to achieve denoising, as shown in Figure 8 .

[0054] 3) Matching the subset point cloud of interest of the CAD model point cloud with the target point cloud

[0055] Using the transformation matrix obtained by matching the CAD subset point cloud of interest with the scanned target point cloud by the ICP algorithm to roughly align the CAD subset point cloud of interest with the target point cloud. During this process, the relative positions of all CAD subset point clouds of interest and the CAD model point cloud do not change. The rough alignment result is shown in Figure 9 , and all CAD interested area subsets are constrained around the 5 subset point cloud areas of interest corresponding to the target point cloud.

[0056] 4) Extracting the subset point cloud of interest on the target point cloud.

[0057] Perform local feature matching between each CAD interested subset point cloud and the target point cloud respectively. The ICP algorithm or the nearest Mahalanobis distance algorithm can be selected as the matching algorithm. In the matching results, all CAD interested subset point clouds are basically coincident with the target point cloud. As Figure 10 shown, the two indication colors overlap.

[0058] After matching, extract all the corresponding area point clouds on the target point cloud that correspond to the CAD interested point cloud subset on the target point cloud. The corresponding area point cloud can be the corresponding area on the target point cloud that is less than 0.005 mm away from the CAD interested point cloud subset. They can be used as the interested subset point cloud of the target point cloud. See Figure 11 .

[0059] Combined with Figures 4 to 11 , the precise scanning point cloud positioning method based on CAD model point cloud constraint described in the present invention can be obtained. Use the CAD model to guide the denoising of the target point cloud (physical point cloud) to improve the accuracy of the target point cloud model; it can ensure the correctness of the feature extraction area. The CAD model point cloud carries the interested subset point cloud to match the target point cloud (physical point cloud), and then extracts the interested subset point cloud of the target point cloud according to the interested subset point cloud. Ensure the correctness of the extraction position of the interested subset point cloud of the target point cloud. Due to the higher accuracy of the target point cloud model and the correctness of the feature extraction area, the target point cloud can better fit the feature to be matched (the interested subset point cloud of the CAD model point cloud) in the feature matching stage, and can improve the accuracy of the target point cloud subset extraction.

[0060] The present invention has high versatility and can be easily promoted. It is not only suitable for the spherical porous part in the example. As long as the target point cloud (physical point cloud) of a porous part with any shape is obtained and the interested subset point clouds on the CAD model point cloud are marked, the corresponding interested area on the target point cloud of the porous part can be accurately extracted.

[0061] Although the present invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed, as long as they do not deviate from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein can be combined in a manner different from that described in the original claims. It should also be understood that the features described in connection with a single embodiment can be used in other described embodiments.

Claims

1. A precise scanning point cloud positioning method based on CAD model point cloud constraints, characterized in that The method includes: S1. Perform mesh sampling on the CAD model of the porous solid part to generate the CAD model point cloud, and mark each point cloud of the interested subset. Among them, each point cloud of the interested subset corresponds to the area where the corresponding hole on the porous solid part is located; at the same time, scan the porous solid part to generate the noisy target point cloud; S2. Use the CAD model point cloud to guide the denoising of the noisy target point cloud; S3. Use the matching algorithm to transform the CAD model point cloud into the coordinate system where the denoised target point cloud is located, so as to achieve the rough alignment of the denoised target point cloud and the CAD model point cloud; perform local feature matching on each point cloud of the interested subset on the CAD model point cloud and the corresponding area on the denoised target point cloud, and extract the point cloud of the corresponding area on the target point cloud corresponding to each point cloud of the interested subset on the CAD model point cloud, and each extracted point cloud of the corresponding area is used as a point cloud of an interested subset of the target point cloud; S4. Use the coordinates of each extracted point cloud of the corresponding area to calculate the centroid coordinates of the point cloud of the corresponding area; and use the centroid coordinates of each point cloud of the corresponding area as the geometric center coordinates of the corresponding hole of the porous solid part to complete the center positioning of each hole on the porous solid part.

2. The precise scanning point cloud positioning method based on CAD model point cloud constraint according to claim 1, wherein The mesh sampling of the CAD model of the porous solid part is implemented using the open3d random sampling function.

3. The precise scanning point cloud positioning method based on the CAD model point cloud constraint according to claim 1, wherein In step S2, the implementation method of using the CAD model point cloud to guide the denoising of the target point cloud includes: The ICP algorithm is used to implement the matching between the CAD model point cloud and the noisy target point cloud. After successful matching, mean filtering is used to filter out the point cloud in the noisy target point cloud except for the area successfully matched with the CAD model point cloud and the point cloud within a preset distance from the successfully matched area. Repeat the above process until the number of point clouds filtered out during the filtering process is lower than the preset number of point clouds to complete the denoising.

4. The precise scanning point cloud positioning method based on CAD model point cloud constraints according to claim 3, characterized in that The preset distance is 0.01 mm.

5. The precise scanning point cloud positioning method based on CAD model point cloud constraint according to claim 1, characterized in that In step S3, the point cloud of the corresponding area on the extracted target point cloud includes: the area corresponding to each point cloud of the interested subset on the CAD model point cloud and the area within 0.005 mm from the corresponding area.

6. A precise scanning point cloud positioning device based on CAD model point cloud constraints, comprising a storage device, a processor, and a computer program stored in the storage device and operable on the processor, characterized in that, The processor executes the computer program to implement the precise scanning point cloud positioning method based on the CAD model point cloud constraint as described in any one of claims 1 to 5.

7. A computer-readable storage device storing a computer program, characterized in that, When the computer program is executed, it implements the precise scanning point cloud positioning method based on the CAD model point cloud constraint as described in any one of claims 1 to 5.

8. A computer program product comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the precise scanning point cloud positioning method based on the CAD model point cloud constraint as described in claims 1 to 5.

Images