2D shape point cloud optimization method and system based on Gaussian interpolation

Abstract

The invention belongs to the technical field of computer point cloud data processing, and discloses a 2D shape point cloud optimization method based on Gaussian interpolation, comprising the following steps: (1) receiving 2D shape contour data by an input module, and obtaining the data through calibration; (2) generating initial point cloud data; (3) smoothing the initial point cloud; (4) expanding a virtual grid layer and filling a zero weight value; (5) removing abnormal points, and generating an optimized target point cloud; and (6) an output module converts the final point cloud into a standardized format, and evaluates shape fidelity and uniformity indexes. The invention discloses a 2D shape point cloud optimization system based on Gaussian interpolation. The 2D shape point cloud optimization system comprises a preprocessing unit; a self-adaptive grid generation unit; and a real-time visualization interaction unit. According to the method, the points of the non-target area are automatically hidden, the point cloud quality is improved, the calculation complexity is reduced, and the method is suitable for real-time rendering and analysis.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer point cloud data processing, and in particular relates to a 2D shape point cloud optimization method and system based on Gaussian interpolation. Background Art

[0002] During the computer drawing process, especially in 2D point cloud generation methods, the following problems are often faced:

[0003] 1. Edge detection, such as the Canny edge detection algorithm and the Sobel edge detection algorithm, directly extracts contour points, but the point cloud is not smooth and there may be discrete points.

[0004] 2. Morphological methods, such as dilation and erosion, are used to optimize the contour point cloud, but cannot automatically optimize the shape quality.

[0005] 3. Voronoi subdivision / Delaunay triangulation can be used for point cloud optimization, but the computational cost is large, affecting real-time performance.

[0006] 4. Deep learning methods, such as CNN predicting point clouds, rely on a large amount of data for training, have limited generalization ability, and are difficult to adapt to custom shapes.

[0007] In view of this, the present invention has developed a 2D shape point cloud optimization method and system based on Gaussian interpolation, which solves the defects existing in the above methods. Summary of the Invention

[0008] To solve the problems in the related art, the present application provides a 2D shape point cloud optimization method and system based on Gaussian interpolation, which solves the defects mentioned in the background art.

[0009] The technical solution is as follows:

[0010] A 2D shape point cloud optimization method based on Gaussian interpolation includes the following steps:

[0011] (1). The input module receives 2D shape contour data, which is obtained through manual calibration or automatic calibration and contains discrete boundary coordinates or parametric curve descriptions;

[0012] (2). The grid point cloud generation module rasterizes the input data into a two-dimensional grid coordinate system, dynamically adjusts the grid density according to the local curvature, and generates initial point cloud data;

[0013] (3). The Gaussian interpolation module smooths the initial point cloud, calculates the interpolation weight using the radial basis Gaussian kernel function, and adaptively adjusts the bandwidth parameter based on the local point density;

[0014] (4) The boundary processing module extends the virtual grid layer along the contour normal direction and fills zero weight values to suppress boundary overflow during the interpolation process;

[0015] (5) The filtering module removes outliers through geometric consistency detection and density clustering algorithm to generate an optimized target point cloud;

[0016] (6) The output module converts the final point cloud into a standardized format and evaluates the shape fidelity and uniformity metrics.

[0017] In a further embodiment, the grid density adjustment strategy in step (2) is specifically:

[0018] The grid spacing in the high-curvature region (curvature ≥ preset threshold) is reduced to 30%-50% of the reference value;

[0019] The grid spacing in the flat region (curvature < preset threshold) is enlarged to 150%-200% of the reference value.

[0020] In a further embodiment, in step (3):

[0021] The Gaussian kernel function is defined as:

[0022]

[0023] Where the adjustment rule of the dynamic bandwidth parameter σ is:

[0024] When the point spacing ≤ 0.5σ_initial, σ is reduced to 50%;

[0025] When the point spacing ≥ 2σ_initial, σ is enlarged to 200%.

[0026] In a further embodiment, in step (4):

[0027] The virtual grid extension includes the following steps:

[0028] (a) Extrapolate 3-5 layers of grid nodes along the contour normal direction;

[0029] (b) Set zero-weight anchor points in the extended area;

[0030] (c) Apply a distance attenuation function to constrain the interpolation weight gradient, and the attenuation function is:

[0031]

[0032] Where d_max is the maximum distance corresponding to the extended number of layers.

[0033] In a further embodiment, in step (5):

[0034] The geometric consistency detection includes:

[0035] Calculating local normal vectors based on principal component analysis;

[0036] Evaluating the main curvature distribution by quadratic surface fitting and removing points with curvature deviating from the mean by ±3σ;

[0037] Using the DBSCAN algorithm to filter noise points in sparse regions, and setting the neighborhood radius to 0.02 - 0.1 times the unit length of the normalized coordinate system. In a further embodiment, in step (6):

[0038] The shape fidelity evaluation metrics include:

[0039] Calculating the maximum deviation of the point cloud before and after optimization through the Hausdorff distance;

[0040] Using the Dice coefficient to quantify the coverage rate of the target shape;

[0041] Calculating the standard deviation of the point cloud spacing to evaluate the uniformity.

[0042] The present invention also discloses a 2D shape point cloud optimization system based on Gaussian interpolation, which is used in the method described in any one of claims 1 - 6, and includes:

[0043] A preprocessing unit: configured to perform point cloud normalization and missing point interpolation;

[0044] An adaptive mesh generation unit: integrating a curvature-sensitive hierarchical encryption algorithm;

[0045] A real-time visualization interaction unit: supporting users to dynamically adjust control points and real-time rendering of the optimization results.

[0046] Adopting the above technical solution, compared with the prior art, the beneficial effects of the technology of this patent are:

[0047] The present invention manually calibrates the target area in a large grid to generate an accurate point cloud, and then uses Gaussian interpolation to smooth the point cloud, avoiding discrete points and boundary overflow problems. In addition, the method of the present invention automatically hides points in non-target areas, improves the quality of the point cloud, reduces the computational complexity, and is suitable for real-time rendering and analysis. Description of the Drawings

[0048] The drawings herein are incorporated into the specification and form a part of this specification, showing embodiments consistent with the present invention, and are used together with the specification to explain the principles of the present invention.

[0049] Figure 1 is the flowchart of the present invention.

[0050] Figure 2 is the flowchart of the virtual grid expansion in the present invention.

[0051] Figure 3 is the flowchart of geometric consistency detection in the present invention.

[0052] Figure 4 is the principle block diagram of the present invention. Figure 5 is the display diagram in the picture placement grid matrix with the palm image as an example in the present invention. Figure 6 is the display diagram of calibrating the finger joint positions to the matrix grid with the palm image as an example in the present invention. Figure 7 is the display diagram after smoothing the point cloud data by using Gaussian interpolation for the picture with the palm image as an example in the present invention. Figure 8 is the display diagram after generating the optimized 2D shape point cloud for the picture with the palm image as an example in the present invention. Detailed implementation manners

[0053] The preferred embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to convey the scope of the present application fully to those skilled in the art.

[0054] The terms used in the present application are for the purpose of describing specific embodiments only and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the present application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0055] It should be understood that although the terms "first", "second", "third", etc. may be used in the present application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, "a plurality" means two or more unless otherwise specifically defined.

[0056] Embodiment

[0057] Such as Figure 1-3As shown, a 2D shape point cloud optimization method based on Gaussian interpolation includes the following steps:

[0058] (1) The input module receives 2D shape contour data, which is obtained through manual calibration or automatic calibration and contains discrete boundary coordinates or parametric curve descriptions.

[0059] The input module receives picture information, then forms a 2D shape contour data, and then places the picture in a grid matrix. For manual calibration of the data, the user calibrates the position of the 2D shape contour through an interactive interface to generate an array of (x, y) coordinate sequences. For automatic calibration of the data, the Canny edge detection algorithm is used to extract the contour and output a discrete point set. The input 2D shape contour data is mapped into the grid matrix, and the index position of the calibration position is recorded to generate an initial binary mask.

[0060] (2) The grid point cloud generation module rasterizes the input data into a two-dimensional grid coordinate system, dynamically adjusts the grid density according to the local curvature, and generates initial point cloud data.

[0061] Generate a two-dimensional grid coordinate system, and generate a point map with the same length, width, and shape as the grid matrix within the two-dimensional grid coordinate system. Perform dynamic grid division in the point map, dividing it into high-curvature regions and flat regions. The grid spacing in the high-curvature region (curvature ≥ preset threshold) is reduced to 30% - 50% of the reference value. The grid spacing in the flat region (curvature < preset threshold) is expanded to 150% - 200% of the reference value.

[0062] (3) The Gaussian interpolation module smooths the initial point cloud, calculates the interpolation weights using the radial basis Gaussian kernel function, and adaptively adjusts the bandwidth parameter based on the local point density.

[0063] The Gaussian kernel function is defined as:

[0064]

[0065] Among them, the adjustment rule of the dynamic bandwidth parameter σ is:

[0066] When the point spacing ≤ 0.5σ_initial, σ is reduced to 50%;

[0067] When the point spacing ≥ 2σ_initial, σ is expanded to 200%.

[0068] The calculation of the interpolation weights includes:

[0069] Calculate the weighted average value for each grid node to generate a smoothed weight field W_smoothed;

[0070] Thresholding (threshold = 0.5) generates a continuous point cloud.

[0071] (4) The boundary processing module extends the virtual grid layer along the contour normal direction and fills it with zero weight values to suppress boundary overflow during the interpolation process.

[0072] The virtual grid extension includes the following steps:

[0073] (a) Extrapolate 3 - 5 layers of grid nodes along the contour normal direction;

[0074] (b) Set zero - weight anchor points in the extended area;

[0075] (c) Apply a distance attenuation function to constrain the interpolation weight gradient, and the attenuation function is:

[0076]

[0077] where d_max is the maximum distance corresponding to the number of extended layers.

[0078] (5) The filtering module eliminates abnormal points through geometric consistency detection and density clustering algorithm to generate an optimized target point cloud; the geometric consistency detection includes:

[0079] Calculation of local normal vectors based on principal component analysis;

[0080] Quadratic surface fitting to evaluate the principal curvature distribution, and eliminate points with curvature deviating from the mean by ±3σ;

[0081] Use the DBSCAN algorithm to filter noise points in sparse areas, and set the neighborhood radius to 0.02 - 0.1 times the unit length of the normalized coordinate system. (6) The output module converts the final point cloud into a standardized format and evaluates the shape fidelity and uniformity metrics.

[0082] The shape fidelity evaluation metrics include:

[0083] Calculate the maximum deviation of the point cloud before and after optimization through the Hausdorff distance;

[0084] Use the Dice coefficient to quantify the coverage rate of the target shape;

[0085] Calculate the standard deviation of the point cloud spacing to evaluate uniformity.

[0086] Taking the palm image as an example

[0087] Step 1, Input module: Receive 2D shape contour data;

[0088] 1) Place the picture in the grid matrix, as Figure 5 shown;

[0089] 2) Calibrate the finger joint positions on the matrix grid, as Figure 6 shown;

[0090] 3), Record the index position of the calibration black dots.

[0091] Step 2, Grid point cloud generation module: Initialize the point cloud data in the grid.

[0092] Generate a dot map with the same length, width and shape as the grid matrix

[0093] Step 3, Gaussian interpolation module: Smooth the point cloud data to avoid discrete points.

[0094] As Figure 7 shown, fill in the corresponding calibration index into the index of the dot map, and use Gaussian interpolation to smooth the point cloud data. Step 4, Boundary processing module: Optimize the boundary points to avoid Gaussian overflow problems.

[0095] Add a border of two columns of 0s to the edge to optimize the boundary points and avoid Gaussian overflow problems.

[0096] Step 5, Filtering module: Hide the points that do not conform to the target shape to improve the point cloud quality.

[0097] Hide the red area and show the shape

[0098] Step 6, Output module: Generate the optimized 2D shape point cloud, as Figure 8 shown.

[0099] Embodiment 2

[0100] As Figure 4 shown, a 2D shape point cloud optimization system based on Gaussian interpolation, using the method in Embodiment 1, includes:

[0101] Preprocessing unit: Configured to perform point cloud normalization and missing point interpolation;

[0102] Adaptive grid generation unit: Integrate a curvature-sensitive hierarchical encryption algorithm;

[0103] Real-time visualization and interaction unit: Support users to dynamically adjust the control points and real-time render the optimization results.

[0104] In the present invention, the target area is manually calibrated in a large grid to generate an accurate point cloud. Next, Gaussian interpolation is used to smooth the point cloud to avoid discrete points and boundary overflow problems. In addition, the method of the present invention automatically hides the points in the non-target area, improves the point cloud quality, reduces the computational complexity, and is suitable for real-time rendering and analysis.

[0105] Other embodiments of the present invention will be readily apparent to those skilled in the art upon consideration of the specification and practice of the invention herein. This application is intended to cover any variations, uses, or adaptations of the invention following the general principles of the invention and including known or customary techniques in the art not invented by the present invention. The specification and examples are only to be considered exemplary, and the true scope and spirit of the invention are indicated by the appended claims.

[0106] It should be understood that the present invention is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A 2D shape point cloud optimization method based on Gaussian interpolation, characterized in that It includes the following steps: (1). The input module receives 2D shape contour data, which is obtained through manual calibration or automatic calibration and contains discrete boundary coordinates or parametric curve descriptions; (2). The grid point cloud generation module rasterizes the input data into a two-dimensional grid coordinate system, dynamically adjusts the grid density according to the local curvature, and generates initial point cloud data; (3). The Gaussian interpolation module smooths the initial point cloud, calculates the interpolation weights using a radial basis Gaussian kernel function, and adaptively adjusts the bandwidth parameter based on the local point density; (4). The boundary processing module extends the virtual grid layer along the contour normal direction and fills it with zero weight values to suppress boundary overflow during the interpolation process; (5). The filtering module removes abnormal points through geometric consistency detection and density clustering algorithms to generate optimized target point cloud; (6). The output module converts the final point cloud into a standardized format and evaluates the shape fidelity and uniformity metrics.

2. A 2D shape point cloud optimization method based on Gaussian interpolation according to claim 1, characterized in that: The grid density adjustment strategy in step (2) is specifically: The grid spacing in the high curvature region (curvature ≥ preset threshold) is reduced to 30%-50% of the reference value; The grid spacing in the flat region (curvature < preset threshold) is expanded to 150%-200% of the reference value.

3. A 2D shape point cloud optimization method based on Gaussian interpolation according to claim 1, characterized in that In step (3): The Gaussian kernel function is defined as: where the adjustment rule of the dynamic bandwidth parameter σ is: When the point spacing ≤ 0.5σ_initial, σ is reduced to 50%; When the point spacing ≥ 2σ_initial, σ is expanded to 200%.

4. A 2D shape point cloud optimization method based on Gaussian interpolation according to claim 1, characterized in that In step (4): The virtual grid expansion includes the following steps: (a) Extrapolate 3-5 layers of grid nodes along the contour normal direction; (b) Set zero weight anchor points in the expanded area; (c) Apply a distance attenuation function to constrain the interpolation weight gradient, and the attenuation function is: where d_max is the maximum distance corresponding to the expanded number of layers.

5. A 2D shape point cloud optimization method based on Gaussian interpolation according to claim 1, characterized in that In step (5): The geometric consistency detection includes: Calculation of local normal vectors based on principal component analysis; Quadratic surface fitting to evaluate the principal curvature distribution, and removing points with curvature deviating from the mean by ±3σ; Using the DBSCAN algorithm to filter noise points in sparse regions, and setting the neighborhood radius to 0.02-0.1 times the unit length of the normalized coordinate system.

6. A 2D shape point cloud optimization method based on Gaussian interpolation according to claim 1, characterized in that In step (6): The shape fidelity evaluation metrics include: Calculating the maximum deviation of the point cloud before and after optimization through the Hausdorff distance; Quantifying the target shape coverage rate using the Dice coefficient; Calculating the standard deviation of the point cloud spacing to evaluate the uniformity.

7. A 2D shape point cloud optimization system based on Gaussian interpolation for implementing the method according to any one of claims 1-6, characterized in that, It includes: A preprocessing unit: configured to perform point cloud normalization and missing point interpolation; An adaptive grid generation unit: integrating a curvature-sensitive hierarchical encryption algorithm; A real-time visualization and interaction unit: supporting users to dynamically adjust control points and real-time rendering of the optimization results.

Images