Method for identifying product hole group based on point cloud segmentation
By abstracting hole features into high-dimensional points and inputting them into point cloud segmentation networks, the inefficient hole group grouping in the traditional method is realized, which solves the time-consuming and labor-consuming problem of designers and realizes fully automatic and efficient hole group identification .
Patent Information
- Application Number
- CN202411606240.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-12
- Publication Date
- 2025-02-14
AI Technical Summary
In traditional methods, designers need to time and effort to extract and compare parameter information of all hole characteristics on the part to achieve hole group grouping with low efficiency.
By abstracting hole features into high-dimensional points, combining these points into point clouds, inputting them into point cloud segmentation network, and automatically grouping of holes is achieved by classifying each point.
It realizes fully automatic hole group identification, improves efficiency, does not require designers to master complex hole grouping skills, and has strong expansion.
Smart Images
Figure CN119445121A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of product hole group recognition based on point cloud segmentation, and in particular to a method for product hole group recognition based on point cloud segmentation. Background Art
[0002] In the product development process of the mechanical design industry, hole features are a very common and critical feature. In most cases, holes are not independent of each other, but have a certain relationship with each other (that is, there is a concept of hole groups). In 2D drawing annotations, they are often divided into internal positioning annotations of hole groups and external positioning annotations of hole groups. Therefore, grouping all hole features simply, quickly and accurately is one of the keys to improving design efficiency and shortening development cycles.
[0003] The traditional grouping method is to manually group holes based on the design experience of designers, taking into account the radius, position, depth and arrangement relationship of hole features; Traditional method: Designers are required to extract and summarize the parameter information of all hole features on the part and compare each parameter one by one, which is a time-consuming and labor-intensive task.
[0004] Therefore, a method for product hole group identification based on point cloud segmentation is proposed. Summary of the invention
[0005] Technical issues solved In view of the shortcomings of the prior art, the present invention provides a method for product hole group identification based on point cloud segmentation, which abstracts the hole features into a high-dimensional point containing hole parameter information, and then merges all the abstracted points into a point cloud and inputs them into a point cloud segmentation network. By classifying each point, the holes can be grouped, etc. It solves the problem that the traditional method is time-consuming and labor-intensive, and solves the problem that designers need to extract and summarize the parameter information of all hole features on parts, and compare each parameter one by one.
[0006] Technical Solution In order to achieve the above purpose of abstracting the hole features into a high-dimensional point containing hole parameter information, then merging all the abstracted points into a point cloud and inputting it into the point cloud segmentation network, and classifying each point to achieve the purpose of grouping the holes, the present invention provides the following technical solutions: A method for identifying product hole groups based on point cloud segmentation comprises the following steps: Step 1: hole parameter information extraction; Step 2: Point cloud data construction and normalization; Step 3: Normalize the center coordinates; Step 4: Direction normalization; Step 5: Depth / radius normalization; Step 6: Train the point cloud segmentation network; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
[0007] Preferably, the hole parameters specifically include hole center coordinates (XYZ), direction (XYZ), depth, and radius, and the device for acquiring data is a three-dimensional laser scanner, a depth camera, or an industrial-grade three-dimensional scanner.
[0008] Preferably, the collected hole parameters may contain noise points, which may affect the subsequent hole group identification. A filtering algorithm may be used to remove the noise, such as Gaussian filtering, median filtering or mean filtering.
[0009] Preferably, the specific operation in step 2 is: normalizing the extracted hole parameter information respectively and merging them into an 8-dimensional vector.
[0010] Preferably, the specific operation in step three is: calculating the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translating all center points, and then dividing by the distance between the farthest point and the origin.
[0011] Preferably, the specific operation in step 4 is: each direction vector is divided by its own modulus.
[0012] Preferably, the specific operation in step five is: dividing by the maximum depth / radius respectively.
[0013] Preferably, the point cloud segmentation network in step six can group the holes that are close in distance and have the same direction by calculating the distance, angle and other relationships between the holes.
[0014] Preferably, the point cloud segmentation network in step six can further verify and identify the hole group according to the design drawings or prior knowledge of the product.
[0015] Compared with the prior art, the present invention provides a method for product hole group identification based on point cloud segmentation, which has the following beneficial effects: 1. This method of product hole group identification based on point cloud segmentation does not require designers to master hole grouping skills to perform relevant annotations; Fully automatic grouping is more efficient than manual grouping; Through AI learning, it has strong scalability and can be expanded quickly when new hole group types appear.
[0016] 2. This method of product hole group identification based on point cloud segmentation abstracts the hole features into a high-dimensional point containing hole parameter information, and then merges all the abstracted points into a point cloud and inputs it into the point cloud segmentation network. These high-dimensional vector sets can be directly trained based on the point cloud segmentation network, and the holes can be grouped by classifying each point. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 It is a structural flow chart of the present invention; Figure 2 A schematic diagram of an embodiment of the present invention; Figure 3 This is a schematic diagram of the network structure of the present invention. DETAILED DESCRIPTION
[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0019] See also Figure 1-Figure 3 ; Embodiment 1: A method for identifying product hole groups based on point cloud segmentation comprises the following steps: Step 1: hole parameter information extraction; The hole parameters specifically include the hole center coordinates (XYZ), direction (XYZ), depth, and radius. The device used to obtain the data is an industrial-grade 3D scanner. The collected hole parameters may contain noise points, which will affect the subsequent hole group identification. Filtering algorithms can be used to remove the noise, such as median filtering; Step 2: Point cloud data construction and normalization; The extracted hole parameter information is normalized separately and then merged into an 8-dimensional vector; Step 3: Normalize the center coordinates; Calculate the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translate all center points, and then divide it by the distance between the farthest point and the origin; Step 4: Direction normalization; Each direction vector is divided by its own magnitude; Step 5: Depth / radius normalization; Divide by the maximum depth / radius respectively; Step 6: Train the point cloud segmentation network; The cloud segmentation network can further verify and identify the hole group based on the product’s design drawings or prior knowledge; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
[0020] Embodiment 2: A method for identifying product hole groups based on point cloud segmentation comprises the following steps: Step 1: hole parameter information extraction; The hole parameters specifically include the hole center coordinates (XYZ), direction (XYZ), depth, and radius. The device for acquiring data is a 3D laser scanner; The collected hole parameters may contain noise points, which will affect the subsequent hole group identification. You can use filtering algorithms to remove the noise, such as Gaussian filtering; Step 2: Point cloud data construction and normalization; The extracted hole parameter information is normalized separately and then merged into an 8-dimensional vector; Step 3: Normalize the center coordinates; Calculate the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translate all center points, and then divide it by the distance between the farthest point and the origin; Step 4: Direction normalization; Each direction vector is divided by its own magnitude; Step 5: Depth / radius normalization; Divide by the maximum depth / radius respectively; Step 6: Train the point cloud segmentation network; The cloud segmentation network can further verify and identify the hole group based on the product’s design drawings or prior knowledge; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
[0021] Embodiment three: A method for identifying product hole groups based on point cloud segmentation comprises the following steps: Step 1: hole parameter information extraction; The hole parameters specifically include the hole center coordinates (XYZ), direction (XYZ), depth, and radius. The device that acquires the data is a depth camera. The collected hole parameters may contain noise points, which will affect the subsequent hole group identification. You can use filtering algorithms to remove the noise, such as mean filtering; Step 2: Point cloud data construction and normalization; The extracted hole parameter information is normalized separately and then merged into an 8-dimensional vector; Step 3: Normalize the center coordinates; Calculate the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translate all center points, and then divide it by the distance between the farthest point and the origin; Step 4: Direction normalization; Each direction vector is divided by its own magnitude; Step 5: Depth / radius normalization; Divide by the maximum depth / radius respectively; Step 6: Train the point cloud segmentation network; The cloud segmentation network can further verify and identify the hole group based on the product’s design drawings or prior knowledge; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
[0022] Embodiment 4: A method for identifying product hole groups based on point cloud segmentation comprises the following steps: Step 1: hole parameter information extraction; The hole parameters specifically include the hole center coordinates (XYZ), direction (XYZ), depth, and radius. The device used to obtain the data is an industrial-grade 3D scanner. The collected hole parameters may contain noise points, which will affect the subsequent hole group identification. You can use filtering algorithms to remove the noise, such as mean filtering; Step 2: Point cloud data construction and normalization; The extracted hole parameter information is normalized separately and then merged into an 8-dimensional vector; Step 3: Normalize the center coordinates; Calculate the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translate all center points, and then divide it by the distance between the farthest point and the origin; Step 4: Direction normalization; Each direction vector is divided by its own magnitude; Step 5: Depth / radius normalization; Divide by the maximum depth / radius respectively; Step 6: Train the point cloud segmentation network; The cloud segmentation network can further verify and identify the hole group based on the product’s design drawings or prior knowledge; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
[0023] Embodiment five: A method for identifying product hole groups based on point cloud segmentation comprises the following steps: Step 1: hole parameter information extraction; The hole parameters specifically include the hole center coordinates (XYZ), direction (XYZ), depth, and radius. The device for acquiring data is a 3D laser scanner; The collected hole parameters may contain noise points, which will affect the subsequent hole group identification. You can use filtering algorithms to remove the noise, such as Gaussian filtering; Step 2: Point cloud data construction and normalization; The extracted hole parameter information is normalized separately and then merged into an 8-dimensional vector; Step 3: Normalize the center coordinates; Calculate the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translate all center points, and then divide it by the distance between the farthest point and the origin; Step 4: Direction normalization; Each direction vector is divided by its own magnitude; Step 5: Depth / radius normalization; Divide by the maximum depth / radius respectively; Step 6: Train the point cloud segmentation network; The cloud segmentation network can group holes that are close in distance and have the same direction by calculating the distance, angle and other relationships between holes; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
[0024] The beneficial effects of the present invention are: This method of product hole group identification based on point cloud segmentation does not require designers to master hole grouping skills to perform relevant annotations; fully automatic grouping is more efficient than manual grouping; through AI learning, it has strong scalability and can be quickly expanded when new hole group types appear; This method for product hole group identification based on point cloud segmentation abstracts the hole features into a high-dimensional point containing hole parameter information, and then merges all the abstracted points into a point cloud and inputs it into the point cloud segmentation network. These high-dimensional vector sets can be directly trained based on the point cloud segmentation network, and the holes can be grouped by classifying each point.
[0025] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for identifying product hole groups based on point cloud segmentation, characterized in that: The following steps are involved: Step 1: hole parameter information extraction; Step 2: Point cloud data construction and normalization; Step 3: Normalize the center coordinates; Step 4: Direction normalization; Step 5: Depth / radius normalization; Step 6: Train the point cloud segmentation network; Step 7: According to the segmentation results, the holes corresponding to the points classified into the same category are divided into the same group to obtain the merged hole group information.
2. The method for product hole group identification based on point cloud segmentation according to claim 1, characterized in that: The hole parameters specifically include hole center coordinates (XYZ), direction (XYZ), depth, and radius. The device for acquiring data is a 3D laser scanner, a depth camera, or an industrial-grade 3D scanner.
3. The method for product hole group identification based on point cloud segmentation according to claim 2, characterized in that: The collected hole parameters may contain noise points, which may affect the subsequent hole group identification. A filtering algorithm may be used to remove the noise, such as Gaussian filtering, median filtering or mean filtering.
4. The method for product hole group identification based on point cloud segmentation according to claim 1, characterized in that: The specific operation in step 2 is: normalizing the extracted hole parameter information respectively and merging them into an 8-dimensional vector.
5. The method for product hole group identification based on point cloud segmentation according to claim 1, characterized in that: The specific operation in step three is: calculate the translation matrix from the smallest coordinate point among all center point coordinates to the origin, translate all center points, and then divide by the distance between the farthest point and the origin.
6. The method for identifying product hole groups based on point cloud segmentation according to claim 1, characterized in that: The specific operation in step 4 is: each direction vector is divided by its own modulus.
7. The method for identifying product hole groups based on point cloud segmentation according to claim 1, characterized in that: The specific operation in step five is: dividing by the maximum depth / radius respectively.
8. The method for identifying product hole groups based on point cloud segmentation according to claim 1, characterized in that: The point cloud segmentation network in step six can group holes that are close in distance and have the same direction by calculating the distance, angle and other relationships between the holes.
9. The method for identifying product hole groups based on point cloud segmentation according to claim 1, characterized in that: The point cloud segmentation network in step six can further verify and identify the hole group according to the design drawings or prior knowledge of the product.