Point cloud normal vector estimation method and system

Abstract

The invention provides a point cloud normal vector estimation method and system, and the method comprises the steps: carrying out the preliminary normal vector estimation of a neighborhood of each point in a point cloud data set containing a multi-shape plane through a principal component analysis method, and obtaining a preliminary normal vector estimation result; updating the normal vector of each point in the point cloud data set based on the preliminary normal vector estimation result; based on the direction of a preset fixed plane normal vector, the updated normal vector is adjusted, so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; and performing normalization processing on the adjusted normal vector to enable the normal vector to reach a preset convergence condition, and obtaining a point cloud normal vector estimation result. The method ensures the convergence and stability of the normal vector estimation result, and has important significance in improving the robustness and reliability of the whole estimation process.

Description

Technical Field

[0001] The present invention belongs to the technical field of point cloud data processing, and in particular to a point cloud normal vector estimation method and system. Background Art

[0002] With the continuous development of 3D scanning technology, point cloud data, as an important information carrier in the field of 3D modeling and computer vision, has been widely used in many fields, such as autonomous driving, robot navigation, building modeling, etc. Point cloud data is collected by 3D scanning instruments or depth cameras, and usually has the characteristics of high precision and high density, and can effectively describe the geometric shape of objects or scenes.

[0003] However, in practical applications, point cloud data usually has many challenges and problems, especially in the point cloud processing process containing complex geometric shapes, folded surfaces or curved areas. Traditional point cloud processing methods mostly rely on local neighborhood analysis, normal vector estimation, surface reconstruction and other technologies, but these methods are often inaccurate in the estimation of point cloud normal vectors in complex geometric structures such as folds and curved surfaces, especially in boundary areas or complex surface areas. The normal vector estimation of traditional methods is easily affected by noise and geometric morphology, resulting in reduced stability and accuracy of the estimation results. In the point cloud data processing of planar areas, it is often impossible to effectively lock the direction of the normal vector, resulting in deviations in the normal vector estimation of planar areas. Summary of the invention

[0004] In order to solve the problem that the normal vector estimation of the traditional method is easily affected by noise and geometric morphology, resulting in reduced stability and accuracy of the estimation result, the present invention provides a point cloud normal vector estimation method and system.

[0005] To achieve the above object, the present invention provides the following technical solutions: The present invention proposes a point cloud normal vector estimation method, comprising: Using principal component analysis, a preliminary normal vector estimation is performed on the neighborhood of each point in the point cloud data set containing multi-shaped planes to obtain the preliminary normal vector estimation result; Based on the preliminary normal vector estimation results, the normal vector of each point in the point cloud dataset is updated; Based on the direction of the preset fixed plane normal vector, adjusting the updated normal vector so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; The adjusted normal vector is normalized so that the normal vector reaches the preset convergence condition to obtain the point cloud normal vector estimation result.

[0006] Preferably, the point cloud data set containing multi-shaped planes comprises: Obtain an initial two-dimensional gridded point cloud; The initial two-dimensional gridded point cloud is divided into two regions, left and right, to obtain a left region point cloud and a right region point cloud; The right region point cloud is transformed, and the left region point cloud is fitted with the transformed right region to obtain the point cloud data set containing multi-shape planes.

[0007] Preferably, the transformation processing of the right area point cloud includes: The right region point cloud is rotated and folded so that the right region point cloud forms a complex geometric shape.

[0008] Preferably, the method of using the principal component analysis method to perform preliminary normal vector estimation on the neighborhood of each point in the point cloud data set containing multi-shaped planes, and obtaining the preliminary normal vector estimation result includes: Acquire point cloud data within the neighborhood of each point in the point cloud data set containing multi-shape planes, and point cloud position data corresponding to the point cloud data; Calculate a covariance matrix based on the point cloud data and the point cloud position data; Performing eigenvalue decomposition on the covariance matrix to obtain a preliminary normal vector estimation result; The preliminary normal vector estimation result includes an eigenvalue and an eigenvector corresponding to the eigenvalue.

[0009] Preferably, updating the normal vector of each point in the point cloud data set based on the preliminary normal vector estimation result includes: Based on the eigenvalues, the curvature of the neighborhood points is calculated; Based on the curvature, the weight of the neighborhood point is calculated; Normalizing the weights to obtain updated weights; The normal vector of each point in the point cloud dataset is updated based on the update weight.

[0010] Preferably, the weight calculation process is:

[0011] in, is the weight, is the curvature, The constant coefficient of the effect of curvature on weight; The normalization process of the weight is as follows:

[0012] in, is the weight, To update the weights, Represents the neighborhood point set of a point in a point cloud dataset.

[0013] Preferably, the process of updating the normal vector of each point in the point cloud data set based on the update weight is:

[0014] in, To update the weights, Represents the neighborhood point set of a point in the point cloud dataset, is the normal vector of the neighborhood point, is the normal vector of the point in the updated point cloud dataset.

[0015] Preferably, the normalization process of the adjusted normal vector so that the normal vector reaches a preset convergence condition to obtain a point cloud normal vector estimation result includes: Iteratively process the adjusted normal vector to obtain an iterative normal vector; Normalizing the iterative normal vector, and determining whether the normalized iterative normal vector satisfies the convergence condition; If the normalized iterative normal vector meets the convergence condition, the iterative processing is terminated to obtain the point cloud normal vector estimation result; if the normalized iterative normal vector does not meet the convergence condition, the next iterative processing is performed.

[0016] Preferably, the convergence condition is:

[0017] in, is the iterated normal vector, is the adjusted normal vector, is the default value.

[0018] The present invention also proposes a point cloud normal vector estimation method for implementing the above method, including: The first processing unit is configured as follows: It is used to use the principal component analysis method to perform preliminary normal vector estimation on the neighborhood of each point in the point cloud data set containing multi-shaped planes, and obtain preliminary normal vector estimation results; The second processing unit is configured as follows: Used to update the normal vector of each point in the point cloud dataset based on the preliminary normal vector estimation result; The third processing unit is configured as follows: Used to adjust the updated normal vector based on the direction of the preset fixed plane normal vector so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; Used to normalize the adjusted normal vector so that the normal vector reaches the preset convergence condition and obtains the point cloud normal vector estimation result; Output unit, configured as: Used to output point cloud normal vector estimation results.

[0019] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention proposes a point cloud normal vector estimation method. In the method, a preliminary normal vector estimation is performed on the neighborhood of each point in a point cloud data set through a principal component analysis method, which can efficiently capture the local geometric features of the point cloud data and effectively improve the application scope of the normal vector estimation. The normal vector of each point in the point cloud data set is updated based on the preliminary normal vector estimation result, the error is reduced, the accuracy of the normal vector is improved, and the true normal vector direction is gradually approached, so that the estimation result is more reliable. By presetting a fixed direction of the plane normal vector and adjusting the updated normal vector to make it consistent with the fixed direction, it can be ensured that the directionality of the normal vector meets the actual needs. While maintaining the accuracy of the normal vector estimation, the practicality and flexibility of the normal vector estimation are also enhanced. The adjusted normal vector is normalized, which not only makes the length of the normal vector reach a unified standard, but also ensures the convergence and stability of the normal vector estimation result, which is of great significance to improving the robustness and reliability of the entire estimation process.

[0020] Furthermore, this method divides the initial two-dimensional gridded point cloud into left and right regions, and transforms the point cloud in the right region, so that the point cloud in the right region and the point cloud in the left region are merged to obtain a point cloud dataset containing multi-shape planes, including planar regions with fixed-direction normal vectors and folded regions with significant changes in the direction of normal vectors. This provides an ideal test dataset for evaluating the adaptability and robustness of normal vector estimation, thereby improving the estimation accuracy of normal vectors by this method.

[0021] Furthermore, this method gradually improves the accuracy of normal vector estimation for each point through multiple optimizations. The normal vector estimation is effectively smoothed and stabilized, avoiding the error accumulation that may be caused by a single method. In addition, the normalization process ensures the consistency of the normal vector, making the subsequent point cloud analysis and processing more accurate. After multiple rounds of iterations and convergence, the result of the normal vector estimation will achieve high stability and accuracy, providing a reliable foundation for subsequent point cloud processing tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of a flow chart of a point cloud normal vector estimation method proposed by the present invention; Figure 2 A front schematic diagram of obtaining an initial two-dimensional gridded point cloud in a point cloud normal vector estimation method proposed by the present invention; Figure 3 A side schematic diagram of obtaining an initial two-dimensional gridded point cloud in a point cloud normal vector estimation method proposed by the present invention; Figure 4This is a schematic diagram of the results of using the PCA normal vector estimation algorithm; Figure 5 This is a schematic diagram of the results of using the RANSAC normal vector estimation algorithm; Figure 6 A schematic diagram of a normal vector estimation result using a point cloud normal vector estimation method proposed by the present invention; DETAILED DESCRIPTION In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0023] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0024] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0025] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or a communication; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0026] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0027] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0028] This paper proposes a point cloud normal vector estimation method, which aims to improve the normal vector estimation accuracy of point clouds with complex geometric shapes (such as folded planes, curved surfaces, etc.), while taking into account the stability and consistency of the normal vectors in the plane area, such as Figure 1 As shown, the following steps are included: S101, using the principal component analysis method, perform preliminary normal vector estimation on the neighborhood of each point in the point cloud data set containing multi-shaped planes, and obtain preliminary normal vector estimation results; exemplary, obtain an initial two-dimensional grid point cloud, and obtain the initial two-dimensional grid point cloud atlas, such as Figure 2 and Figure 3 As shown in the figure; the initial two-dimensional gridded point cloud is divided into two regions, left and right, to obtain the left region point cloud and the right region point cloud; the left region point cloud is a flat plane, and the right region point cloud is transformed, that is, the regional point cloud is rotated and folded, so that the right region point cloud forms a complex geometric shape, and the left region point cloud is merged with the transformed right region to obtain a point cloud dataset containing multi-shaped planes. The point cloud dataset has both simple planes and complex folded surfaces, and contains obvious geometric feature differences. The point cloud dataset provides an ideal test dataset for evaluating the adaptability and robustness of normal vector estimation; a set of points in the neighborhood of each point in the point cloud dataset containing multi-shaped planes is obtained. Point cloud data, and point cloud position data corresponding to the point cloud data; the covariance matrix is ​​calculated through the point cloud data and the point cloud position data, and the covariance matrix reflects the spatial structural characteristics of the neighborhood points; the covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and eigenvectors corresponding to the eigenvalues, the eigenvalues ​​are compared, and the eigenvector corresponding to the smallest eigenvalue is extracted. The direction of the eigenvector reflects the direction in which the point cloud changes the least in the local area, that is, the surface normal vector, and a preliminary normal vector is obtained. The preliminary normal vector can make a preliminary estimate of the normal vector of each point, which provides a basis for subsequent processing and can effectively provide fast normal vector estimation for large-scale point cloud data.

[0029] S102, based on the preliminary normal vector estimation result, update the normal vector of each point in the point cloud data set; illustratively, obtain the eigenvalue corresponding to the eigenvector corresponding to each point in the point cloud data set, and calculate the curvature of the neighborhood point through the eigenvalue. The smaller the principal component variance corresponding to the eigenvector is, the larger the curvature is, indicating that the point has a larger curvature feature in the local geometry; calculate the weight of the neighborhood point through the curvature, and the calculated weight is adaptively adjusted through the Gaussian function. The larger the curvature is, the smaller the weight is, indicating that the geometric characteristics of the point have less influence on the normal vector estimation; conversely, the point with smaller curvature (i.e., closer to the plane) has a greater influence on the normal vector; The weight calculation process is:

[0030] in, is the weight, is the curvature, The constant coefficient of the effect of curvature on weight; Normalize the weights adjusted by the Gaussian function to obtain the updated weights, so that the sum of the updated weights of all neighborhood points is 1; The normalization process of weights is:

[0031] in, is the weight, To update the weights, Represents the neighborhood point set of a point in a point cloud dataset.

[0032] The normal vector of each point in the point cloud dataset is updated by updating the weights. In the normal vector updating process, the weighted average method is used to smooth the normal vector of each point. The normal vector of each point is updated as the weighted average of the normal vectors of its neighborhood. The update weight is determined by the curvature. For points with smaller curvature, they are more dependent on the normal vectors of the neighborhood; for points with larger curvature, smaller update weights are used to adjust their normal vectors.

[0033] The process of updating the weights to update the normal vector of each point in the point cloud dataset is:

[0034] in, To update the weights, Represents the neighborhood point set of a point in the point cloud dataset, is the normal vector of the neighborhood point, is the normal vector of the point in the updated point cloud dataset.

[0035] S103, based on the preset fixed direction of the plane normal vector, adjusting the updated normal vector so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; illustratively, the updated normal vector is identified by a simple height difference method, the plane area is identified, and the point cloud normal vector of the updated plane area is obtained. The direction of the fixed plane normal vector is preset, and based on the direction of the plane normal vector, the point cloud normal vector of the updated plane area is adjusted so that the direction of the point cloud normal vector of the updated plane area is consistent with the direction of the fixed plane normal vector, ensuring that the normal vectors of these areas are consistent with the global plane direction, avoiding error accumulation in the plane area, illustratively, the direction of the preset fixed plane normal vector is: , the adjustment process is:

[0036] in, ∈[0,1] is the constraint weight, which is used to adjust the balance between smoothness and constraint; It is the normal vector of the point in the updated point cloud dataset obtained by adaptive smoothing in the previous step.

[0037] S104, normalize the adjusted normal vector so that the normal vector reaches the preset convergence condition and obtains the point cloud normal vector estimation result. Exemplarily, the adjusted normal vector is iteratively processed to obtain the iterative normal vector; the iterative normal vector is normalized to determine whether the normalized iterative normal vector meets the convergence condition; if the normalized iterative normal vector meets the convergence condition, the iterative processing is terminated to obtain the point cloud normal vector estimation result; if the normalized iterative normal vector does not meet the convergence condition, the next iterative processing is performed until the convergence condition is met. The normal vector is gradually updated through multiple iterations. In each iteration, the update of the normal vector combines local geometric features with global plane information to ensure that the normal vector in the point cloud smoothly transitions between the plane area and the non-plane area. After each iteration, the normal vector will be normalized to maintain its unit vector properties.

[0038] The iterative formula process is:

[0039] in, Represents the neighborhood point set of a point in the point cloud dataset, is the updated normal vector; The normalization formula is:

[0040] in, is the iterated normal vector.

[0041] The convergence conditions are:

[0042] in, is the iterated normal vector, is the adjusted normal vector, is the default value.

[0043] The method is further explained by an example below; the meshgrid function in MATLAB is used to generate a two-dimensional grid coordinate, and x and y are defined as the grid horizontal and vertical coordinates, ranging from [0,5], where , by partitioning the grid points, the first half is defined as the left area point cloud (x<2.5), and the second half is the right area point cloud (x≥2.5). The right point cloud is rotated 60 degrees around the y axis, and the right area point cloud and the transformed right area point cloud are merged to form a folded "V" shaped point cloud, and a point cloud dataset containing multi-shaped planes is obtained. A set of point cloud data in the neighborhood of each point in a point cloud dataset containing multi-shaped planes is obtained, and the principal component analysis (PCA) method is used to calculate the eigenvalues ​​and the eigenvectors corresponding to the eigenvalues. The neighborhood points whose distances to the points corresponding to the eigenvectors are less than 0.15 are selected to obtain the eigenvalues ​​of the neighborhood points. The curvature of the neighborhood points is calculated by the eigenvalues. The weights of the neighborhood points are calculated by the curvature, and the weights are normalized by the weights to obtain the updated weights. The normal vectors of each point in the point cloud dataset are updated by updating the weights. The fixed plane normal vector is preset to [0,0,1]. The normal vectors will gradually move toward the fixed plane normal vector [0,0,1]. After 100 iterations, the normal vectors are continuously optimized and finally converge to a stable result to obtain the point cloud normal vector estimation result.

[0044] Figure 4 This is a schematic diagram of the results of the PCA normal vector estimation method. Figure 4 It can be seen that the PCA normal vector estimation method performs poorly in the left half of the data and in the corners, which reflects that this method has higher accuracy and robustness; Figure 5 This is a schematic diagram of the results of using the RANSAC normal vector estimation method. Figure 5 It can be seen that the results of the RANSAC normal vector estimation method are relatively messy and the estimation effect is poor, which reflects that this method has a high accuracy rate. Figure 6 This is a schematic diagram of the normal vector estimation method after using this method. Figure 6 It can be seen that the method in this paper overcomes the shortcomings of PCA and RANSAC normal vector estimation methods, which further reflects that this method has high accuracy and high robustness.

[0045] The present invention also proposes a point cloud normal vector estimation method for implementing the above method, including: The first processing unit is configured as follows: It is used to use the principal component analysis method to perform preliminary normal vector estimation on the neighborhood of each point in the point cloud data set containing multi-shaped planes, and obtain preliminary normal vector estimation results; It is further configured as follows: obtaining an initial two-dimensional gridded point cloud, dividing the initial two-dimensional gridded point cloud into a left area and a right area to obtain a left area point cloud and a right area point cloud, performing transformation processing on the right area point cloud, that is, rotating and folding the area point cloud, merging the left area point cloud with the transformed right area to obtain a point cloud data set containing multi-shape planes, obtaining a set of point cloud data in the neighborhood of each point in the point cloud data set containing multi-shape planes, and point cloud position data corresponding to the point cloud data; calculating a covariance matrix through the point cloud data and the point cloud position data, performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors corresponding to the eigenvalues, comparing the eigenvalues, and extracting the eigenvector corresponding to the smallest eigenvalue. The second processing unit is configured as follows: Used to update the normal vector of each point in the point cloud dataset based on the preliminary normal vector estimation result; It is further configured to: obtain the eigenvalue corresponding to the eigenvector corresponding to each point in the point cloud data set, calculate the curvature of the neighborhood point through the eigenvalue, calculate the weight of the neighborhood point through the curvature, normalize the weight to obtain the update weight, so that the sum of the update weights of all neighborhood points is 1, and update the normal vector of each point in the point cloud data set through the update weight.

[0046] The third processing unit is configured as follows: Used to adjust the updated normal vector based on the direction of the preset fixed plane normal vector so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; It is further configured to: identify the updated normal vector through a simple height difference method, identify the plane area, obtain the point cloud normal vector of the updated plane area, preset the direction of the fixed plane normal vector, and adjust the point cloud normal vector of the updated plane area based on the direction of the plane normal vector so that the direction of the point cloud normal vector of the updated plane area is consistent with the direction of the fixed plane normal vector.

[0047] The fourth processing unit is configured to: It is used to normalize the adjusted normal vector so that the normal vector reaches the preset convergence condition and obtains the point cloud normal vector estimation result.

[0048] It is further configured to: iteratively process the adjusted normal vector to obtain an iterative normal vector; normalize the iterative normal vector to determine whether the normalized iterative normal vector meets the convergence condition; if the normalized iterative normal vector meets the convergence condition, terminate the iterative process to obtain the point cloud normal vector estimation result; if the normalized iterative normal vector does not meet the convergence condition, perform the next iterative process until the convergence condition is met.

[0049] It is used to normalize the adjusted normal vector so that the normal vector reaches the preset convergence condition and obtains the point cloud normal vector estimation result.

[0050] Output unit, configured as: Used to output point cloud normal vector estimation results.

[0051] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0052] In addition, it should be understood that although this specification is described in accordance with the implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A point cloud normal vector estimation method, characterized in that: include: Using principal component analysis, a preliminary normal vector estimation is performed on the neighborhood of each point in the point cloud data set containing multi-shaped planes to obtain the preliminary normal vector estimation result; Based on the preliminary normal vector estimation results, the normal vector of each point in the point cloud dataset is updated; Based on the direction of the preset fixed plane normal vector, adjusting the updated normal vector so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; The adjusted normal vector is normalized so that the normal vector reaches the preset convergence condition to obtain the point cloud normal vector estimation result.

2. A point cloud normal vector estimation method according to claim 1, characterized in that: The point cloud data set containing multi-shape planes comprises: Obtain an initial two-dimensional gridded point cloud; The initial two-dimensional gridded point cloud is divided into two regions, left and right, to obtain a left region point cloud and a right region point cloud; The right region point cloud is transformed, and the left region point cloud is fitted with the transformed right region to obtain the point cloud data set containing multi-shape planes.

3. A point cloud normal vector estimation method according to claim 2, characterized in that: The transformation processing of the right area point cloud includes: The right region point cloud is rotated and folded so that the right region point cloud forms a complex geometric shape.

4. A point cloud normal vector estimation method according to claim 1, characterized in that: The principal component analysis method is used to perform preliminary normal vector estimation on the neighborhood of each point in the point cloud data set containing multi-shaped planes, and the preliminary normal vector estimation result includes: Acquire point cloud data within the neighborhood of each point in the point cloud data set containing multi-shape planes, and point cloud position data corresponding to the point cloud data; Calculate a covariance matrix based on the point cloud data and the point cloud position data; Performing eigenvalue decomposition on the covariance matrix to obtain a preliminary normal vector estimation result; The preliminary normal vector estimation result includes an eigenvalue and an eigenvector corresponding to the eigenvalue.

5. A point cloud normal vector estimation method according to claim 4, characterized in that: The updating of the normal vector of each point in the point cloud data set based on the preliminary normal vector estimation result includes: Based on the eigenvalues, the curvature of the neighborhood points is calculated; Based on the curvature, the weight of the neighborhood point is calculated; Normalizing the weights to obtain updated weights; The normal vector of each point in the point cloud dataset is updated based on the update weight.

6. A point cloud normal vector estimation method according to claim 5, characterized in that: The weight calculation process is: in, is the weight, is the curvature, The constant coefficient of the effect of curvature on weight; The normalization process of the weight is as follows: in, is the weight, To update the weights, Represents the neighborhood point set of a point in a point cloud dataset.

7. A point cloud normal vector estimation method according to claim 6, characterized in that: The process of updating the normal vector of each point in the point cloud dataset based on the update weight is: in, To update the weights, Represents the neighborhood point set of a point in the point cloud dataset, is the normal vector of the neighborhood point, is the normal vector of the point in the updated point cloud dataset.

8. A point cloud normal vector estimation method according to claim 1, characterized in that: The normalization process of the adjusted normal vector is performed so that the normal vector reaches a preset convergence condition to obtain a point cloud normal vector estimation result, including: Iteratively process the adjusted normal vector to obtain an iterative normal vector; Normalizing the iterative normal vector, and determining whether the normalized iterative normal vector satisfies the convergence condition; If the normalized iterative normal vector meets the convergence condition, the iterative processing is terminated to obtain the point cloud normal vector estimation result; if the normalized iterative normal vector does not meet the convergence condition, the next iterative processing is performed.

9. A point cloud normal vector estimation method according to claim 8, characterized in that: The convergence condition is: in, is the iterated normal vector, is the adjusted normal vector, is the default value.

10. A point cloud normal vector estimation method for implementing the method according to any one of claims 1 to 9, characterized in that: include: The first processing unit is configured as follows: It is used to use the principal component analysis method to perform preliminary normal vector estimation on the neighborhood of each point in the point cloud data set containing multi-shaped planes, and obtain preliminary normal vector estimation results; The second processing unit is configured as follows: Used to update the normal vector of each point in the point cloud dataset based on the preliminary normal vector estimation result; The third processing unit is configured as follows: Used to adjust the updated normal vector based on the direction of the preset fixed plane normal vector so that the direction of the updated normal vector is consistent with the direction of the fixed plane normal vector; The fourth processing unit is configured to: Used to normalize the adjusted normal vector so that the normal vector reaches the preset convergence condition and obtain the point cloud normal vector estimation result; Output unit, configured as: Used to output point cloud normal vector estimation results.

Images