Indoor point cloud wall segmentation method based on density constraint super voxel

Abstract

The invention discloses an indoor point cloud wall face segmentation method based on density constraint super voxels, and belongs to the technical field of point cloud data processing and three-dimensional space data analysis, and the method comprises the following steps: collecting point clouds, removing invalid points, dividing structural point clouds through a normal vector and a gravity direction pair included angle, and obtaining an included angle; the method comprises the following steps: obtaining a primary wall point cloud set through rasterization and combined constraint of density constraint and height constraint, optimizing super voxel segmentation by adopting an energy function, completing initialization, seed screening and adjacency relation construction, performing iterative growth and traversal screening by taking a super voxel with the minimum curvature as a seed to obtain a super voxel cluster, and obtaining an initial wall point cloud set; and finally, outputting a complete wall point cloud through binary mask expansion, label updating and point cloud elimination. According to the method, the phenomena of wrong segmentation and missing segmentation can be effectively inhibited in a complex indoor scene, and the method is suitable for point cloud wall segmentation of a multi-wall and multi-scale structure.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of point cloud segmentation, and particularly relates to an indoor point cloud wall surface segmentation method based on density constraint super voxel. BACKGROUND

[0002] In an indoor environment, point cloud data can more completely record the geometric shape and spatial distribution characteristics of building components, providing direct and reliable data basis for real scene three-dimensional modeling and indoor structure reconstruction. However, due to the influence of multiple components interweaving in three-dimensional space and light reflection in the spatial layout of the indoor scene, laser scanning is extremely susceptible to occlusion and noise interference when collecting point clouds, so that the indoor point cloud presents the characteristics of uneven density, local missing, and more noise, which brings great challenges to indoor point cloud segmentation.

[0003] Existing point cloud segmentation methods usually use model matching-based methods, region growing methods, feature clustering methods, and deep learning-based methods to extract main structural surfaces such as walls, roofs, and floors. The model fitting-based method matches and segments the point cloud by iteratively fitting a pre-set geometric model. This method works well in regular structure scenes, but its efficiency is limited in actual indoor environments; the feature clustering method divides points with similar features into the same object, but in indoor environments, uneven density often occurs, making it difficult for the algorithm to run stably; the deep learning-based method needs a large amount of labeled data set for training, and in indoor measured point clouds, the segmentation performance is significantly reduced. Compared with the above methods, the region growing method has high accuracy and is easy to implement, but it is easily affected by incomplete structures in non-Manhattan indoor scenes, making it difficult to maintain wall continuity. SUMMARY

[0004] The present application provides an indoor point cloud wall surface segmentation method based on density constraint super voxel to solve the technical problems of indoor point cloud wall surface segmentation being susceptible to point cloud noise, wall point cloud data missing, and wall two-side neighboring factors interference in the prior art.

[0005] An indoor point cloud wall surface segmentation method based on density constraint super voxel, comprising the following steps: S1. Collecting wall point clouds and removing invalid points, and dividing them into vertical structure point clouds and horizontal structure point clouds according to the included angle between the normal vector direction and the gravity direction; S2. The vertical structure point cloud is subjected to horizontal plane gridding processing, a grid density map and a grid height map are constructed, the grid density map and the grid height map are respectively subjected to smoothing processing, and density constraint and grid constraint are applied, the in-cloud points of the grid meeting the density constraint and the height constraint are back-projected to obtain a preliminary wall point cloud set; S3. Using the initial wall point cloud set as the processing object, the supervoxel segmentation method is constructed by preserving the boundary of the energy function optimization; by defining and minimizing the energy function that describes the structural characteristics of the supervoxel, a set of supervoxels with optimization boundary is generated. S4. Perform initialization, seed selection, and adjacency relationship construction on the supervoxel point cloud of the supervoxel set to obtain the supervoxel adjacency set; S5. Initialize all supervoxes in the supervox adjacency set to an unvisited state. Use the unvisited supervoxe with the smallest curvature as the growth seed. Traverse the adjacent supervoxes based on the adjacency relationship. After filtering by normal consistency geometric constraints and curvature similarity geometric constraints, merge them into the current growth cluster. Select the supervoxe with the smallest curvature from the remaining unvisited supervoxes as the new seed. Iterate the seed selection, traversal filtering and merging operations until all supervoxes have been traversed, and obtain the supervoxe cluster with corresponding independent walls and consistent geometric features. S6. Based on the wall point cloud corresponding to the supervoxel cluster, recover the supervoxel labels of each 3D point and project them onto the 2D plane. After generating the label raster, construct a binary mask and perform morphological dilation. Update the labels in the dilated binary mask and remove the uncovered point cloud to obtain the complete wall point cloud.

[0006] Furthermore, the angle between the direction of the normal vector and the direction of gravity includes: ; In the formula, For the first The angle between the normal vector of a point cloud and the direction of gravity. For the first The normal vector of a point cloud. The direction vector of gravity; Set the vertical structure angle threshold Threshold of the angle between the horizontal structure and the horizontal structure , For vertical point clouds, This represents a horizontally structured point cloud.

[0007] Furthermore, the vertical structure point cloud is rasterized in a horizontal plane, including rasterizing the vertical structure point cloud in a horizontal plane, with the grid side length set to... Two-dimensional raster coordinates of vertical structure point cloud : ; ; In the formula, For the first The grid coordinates of a point cloud in the horizontal plane X direction. For the first The coordinates of a point cloud in the horizontal plane X direction. the minimum value of the coordinates of all vertical structure point clouds in the X direction, the grid coordinate of the first point cloud in the horizontal plane Y direction, the coordinate of the first point cloud in the horizontal plane Y direction, the minimum value of the coordinates of all vertical structure point clouds in the Y direction, the grid edge length.

[0008] Further, the grid density map and the grid height map are respectively subjected to smoothing processing, and density constraints and grid constraints are applied, including counting the number of vertical structure point clouds falling within the two-dimensional grid coordinates of each point cloud to construct a grid density map , applying two-dimensional mean filtering to reduce noise to obtain a smoothed density map , and using the following wall density constraint formula to screen the grid that meets the density requirement: ; In the formula, is the grid density screening identifier, is the number of point clouds corresponding to the grid with coordinates in the smoothed grid density map, is the density threshold value; counting the number of vertical structure point clouds falling within the two-dimensional grid coordinates of each point cloud to construct a grid height map , taking the maximum height value of all point clouds in each grid as the height value of the corresponding grid; applying maximum value filtering to the grid height map for smoothing processing to obtain a filtered height map , and using the following wall height constraint formula to screen the grid that meets the height requirement: ; In the formula, is the grid height screening identifier, is the height of the point cloud corresponding to the grid with coordinates in the smoothed grid height map, is the ceiling height, is used to filter out low structures.

[0009] Furthermore, the point cloud within the raster that meets the density and height constraints is back-projected to obtain a preliminary wall point cloud set, including wall raster determination through the joint constraint of density and height constraints. The joint constraint is calculated by the following formula: ; In the formula, For the final filtering identifier of the raster, For raster density filtering, Use grid height as the filter identifier; Will satisfy The point cloud within the raster is back-projected into the three-dimensional coordinate space to obtain a preliminary wall point cloud set.

[0010] Furthermore, the energy function includes a difference term and a constraint term, and the energy function is calculated using the following formula: ; ; ; In the formula, For point clouds With point clouds Difference value, and The first step in the preliminary wall point cloud The and the first A three-dimensional point, and for and The normal vector, Represents the distance between neighboring points. This represents the statistical value of the number of supervoxels. This is the set of results from supervoxel segmentation. To segment the relation matrix, For indicator functions, This represents the total number of 3D points in the initial wall point cloud that participate in energy function optimization. The optimization objective is to minimize the energy function. The energy function for point cloud segmentation. To preset the number of hypervoxels to be generated, These are the weighting coefficients.

[0011] Furthermore, initialization includes dividing the point cloud into several initial regions based on the color features of the supervoxels, and calculating the centroid for each supervoxel. Average normal vector curvature and the number of point clouds , the extraction of core geometric features is completed, all super voxel initial states are set to unvisited, and the region growing result set is empty; Seed screening, including ordering the super voxels in ascending order according to the curvature values, selecting the super voxel with the smallest curvature in the unvisited state as the initial seed of region growing , and if there is no super voxel satisfying the unvisited condition after traversal, the algorithm is terminated, and the super voxels satisfying the visited condition are added to the growing queue and the state of the super voxel is marked as visited at the same time; Adjacent relationship construction, including taking the centroid of the super voxel as the spatial search reference, performing k-nearest neighbor search on each super voxel to obtain the corresponding local candidate neighborhood; collecting the three-dimensional point coordinates contained in all super voxels in the candidate neighborhood, and performing a density-based noisy space clustering operation based on the point coordinates to determine the main local cluster to which the current super voxel belongs; all super voxels belonging to the same main local cluster and not being the current super voxel are regarded as the real adjacent voxels of the current super voxel, and the super voxel adjacent set is generated .

[0012] Further, the normal consistency geometric constraint and the curvature similarity geometric constraint include: ; ; In the formula, is the absolute value of the normal vector angle between the current super voxel and the adjacent super voxel, is the average normal vector of the current super voxel, is the average normal vector of the adjacent super voxel of the current super voxel, is the absolute difference value of the curvature values of the current super voxel and the adjacent super voxel, is the curvature value of the current super voxel, is the curvature value of the adjacent super voxel of the current super voxel; Set the normal vector angle threshold and the curvature value difference threshold , the adjacent super voxels that satisfy and at the same time are marked as visited and added to the growing queue, and are incorporated into the current growing cluster at the same time.

[0013] Further, based on the wall point cloud corresponding to the super voxel cluster, the super voxel labels of each three-dimensional point are recovered and projected to a two-dimensional plane, a label grid is generated after constructing a binary mask and performing morphological dilation, including three-dimensional points The two-dimensional plane coordinates projected by the three-dimensional points are: ; In the formula, Adding clouds to the wall Three-dimensional points Two-dimensional coordinates after projection onto a two-dimensional plane. and The first Three-dimensional points The coordinate values ​​corresponding to the two coordinate axes projected onto a two-dimensional plane, and the combination of coordinate axes satisfies ; A grid is constructed based on the two-dimensional planar coordinates of the projection and the grid grid side length is used. Spatially discretize the two-dimensional projection points of each point in the wall point cloud to obtain a two-dimensional label raster. : ; In the formula, For the first The two-dimensional coordinates of a three-dimensional point projected onto a two-dimensional plane The supervoxel label of the grid location; Adding clouds to the wall A 3D point hypervoxel label; Count the number of point clouds contained in the supervoxels and set a threshold for large areas. Construct a set of large faces, which is: ; In the formula, It is a collection of large pieces of surface. For the first The number of point clouds corresponding to a large area of ​​a supervoxel. For large area thresholds; When the number of point clouds corresponding to a large area of ​​a supervoxel exceeds At that time, for large-area sets Each large facet in Constructing in a two-dimensional label grid Corresponding binary mask : ; In the formula, For the first A large piece of dough The corresponding binary mask, For two-dimensional label grids at grid positions Supervoxel tags at the location, For the first A large, flat hypervoxel label; right Perform morphological dilation operation. The expansion radius is determined by the number of iterations. Controlling the process to obtain the expanded binary mask : ; In the formula, For the first A large piece of dough The corresponding dilated binary mask, For the first A large piece of dough The corresponding binary mask, This represents the number of iterations.

[0014] Furthermore, the labels within the expanded binary mask are updated, and the uncovered point cloud is removed to obtain the complete wall point cloud, including the first... A large piece of dough Corresponding dilated binary mask The covered area is considered as the outer extension of the wall structure in two-dimensional space, and any three-dimensional point in the wall point cloud is considered as such. Get , Falling into A large piece of dough Corresponding dilated binary mask Within the covered area, three-dimensional points The hypervoxel label has been updated to: ; In the formula, Adding clouds to the wall Three-dimensional points Updated hypervoxel labels; 3D points in wall point cloud of Not by any If covered, then three-dimensional points are considered. 3D points not belonging to the wall structure are directly removed. After updating the labels and removing the uncovered point clouds from the wall point cloud, a complete wall point cloud is obtained.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This invention employs hypervoxels as the core processing unit, combining density constraints and morphological dilation optimization to effectively suppress point cloud noise interference. Simultaneously, it fills in missing areas of wall data and distinguishes adjacent wall boundaries, avoiding segmentation interruptions and region mismerging issues caused by local anomalies or structural proximity in existing technologies. This makes it suitable for segmentation needs of complex interior structures. Through energy function optimization, this invention achieves precise definition of hypervoxel boundaries. Combined with a region growth strategy constrained by normal consistency and curvature similarity, it can accurately divide independent wall regions, reducing segmentation fragments and ensuring clear wall boundaries and complete morphology. Compared to existing technologies, it significantly reduces segmentation errors and improves the accuracy and completeness of wall segmentation. This invention forms a complete closed-loop process of "point cloud screening - hypervoxel construction - cluster growth - refined repair," requiring minimal manual parameter adjustment. It can automatically complete invalid point removal, hypervoxel clustering, and wall repair, adapting to wall point cloud data in different interior scenarios. It is convenient to operate, highly efficient, and easier to apply to real-world engineering scenarios. It has been specifically optimized for typical indoor point cloud problems such as adjacent walls on both sides, missing data, and noise interference. It can not only handle conventional indoor wall segmentation, but also adapt to indoor scenes with complex structures and many interference factors, making it more applicable to a wider range of scenarios than traditional methods. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the technical process of the present invention. Figure 2 This is a flowchart of a raster filtering process based on raster density and raster height constraints. Figure 3 This is a flowchart of the region growth process based on boundary-preserving supervoxels. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this disclosure.

[0018] like Figure 1 As shown, an indoor point cloud wall segmentation method based on density-constrained hypervoxels includes the following steps: S1. Collect wall point clouds. For the effective wall point clouds obtained after removing invalid points, calculate the normal vector direction of each effective point using the principal component analysis method. Divide the effective point clouds into vertical structural point clouds and horizontal structural point clouds according to the angle between the normal vector direction and the gravity direction. The angle between the direction of the normal vector and the direction of gravity is calculated using the following formula: ; In the formula, For the first The angle between the normal vector of a point cloud and the direction of gravity. For the first The normal vector of a point cloud. The direction vector of gravity; Set the vertical structure angle threshold Threshold of the angle between the horizontal structure and the horizontal structure , This is a vertical point cloud. This results in a horizontally structured point cloud.

[0019] S2. For example Figure 2 As shown, the filtered vertical post-node point cloud is rasterized on a horizontal plane, with the grid side length set to 1. Coordinates of a 2D raster of a vertically structured point cloud : ; ; In the formula, For the first The grid coordinates of a point cloud in the horizontal plane X direction. For the first The coordinates of a point cloud in the horizontal plane X direction. The minimum coordinate in the X direction for all vertical structure point clouds. For the first The grid coordinates of a point cloud in the Y direction of the horizontal plane. For the first The coordinates of a point cloud in the Y direction of the horizontal plane. The minimum value of the Y-axis coordinates of all vertical structure point clouds. This represents the side length of the grid.

[0020] Based on the two-dimensional raster coordinates of each point cloud Statistical analysis of two-dimensional grid coordinates The number of inset vertical structure point clouds is used to construct a raster density map. ,right Applying a two-dimensional mean filter to reduce noise yields a smoothed density map. The following wall density constraint formula is used to select grids that meet the density requirements: ; In the formula, For raster density filtering, In the smoothed raster density map, the coordinates are... The number of point clouds corresponding to the raster. Density threshold; Based on the two-dimensional raster coordinates of each point cloud Statistical analysis of two-dimensional grid coordinates The number of vertically structured point clouds falling in is used to construct a raster height map. The maximum height value of all point clouds within each grid cell is used as the height value of the corresponding grid cell; for the grid height map Applying maximum value filtering to smooth the image yields the filtered height map. The following wall height constraint formula is used to filter grids that meet the height requirements: ; In the formula, Filtering identifier for grid height, In the smoothed grid height map, the coordinates are... The point cloud height corresponding to the raster. Ceiling height Used to filter out low-profile structures; The wall grid is then determined by combining density and height constraints, and calculated using the following formula: ; In the formula, For the final filtering identifier of the raster, For raster density filtering, Use grid height as the filter identifier; Will satisfy The point cloud within the raster is back-projected into the three-dimensional coordinate space to obtain a preliminary wall point cloud set.

[0021] S3. Taking the initial wall point cloud set as the processing object, the structural characteristics of the super voxels are described by an explicitly defined energy function, and the energy function is minimized by selecting a subset of the point cloud, thereby generating super voxels with optimized boundary definitions. The energy function includes a difference term and a constraint term, and is calculated using the following formula: ; ; ; In the formula, For point clouds With point clouds Difference value, and The first step in the preliminary wall point cloud The and the first A three-dimensional point, and for and The normal vector, Represents the distance between neighboring points. This represents the statistical value of the number of supervoxels. This is the set of results from supervoxel segmentation. To segment the relation matrix, For indicator functions, This represents the total number of 3D points in the initial wall point cloud that participate in energy function optimization. The optimization objective is to minimize the energy function. The energy function for point cloud segmentation. To preset the number of hypervoxels to be generated, These are the weighting coefficients.

[0022] S4. For example Figure 3 As shown, the input is a supervoxel point cloud segmented using the Boundary Preserving Supervoxel Segmentation (BPSS) method. First, the point cloud is divided into several regions based on the supervoxel color. Then, features, including the centroid, are calculated for each supervoxel. Average normal vector curvature and the number of point clouds Set all supervoxels to an unvisited initial state and set the region growth results together. Set to empty; then perform seed filtering, which involves sorting hypervoxels in ascending order of curvature value and selecting the hypervoxel that is currently unvisited and has the smallest curvature. This serves as the initial seed for region growth; if no unvisited supervoxel is found after traversal, the algorithm terminates, and supervoxels that meet the visit criteria are added to the growth queue. Simultaneously, the state of the supervoxel is marked as visited, and then adjacency relationships are constructed. The adjacency relationship construction includes the centroid of the supervoxel. As a spatial search benchmark, k-nearest neighbor search is performed on each supervoxel to obtain the corresponding local candidate neighborhood. The 3D point coordinates of all supervoxels within the candidate neighborhoods are collected. Based on these point coordinates, density-based noisy spatial clustering (DBSCAN) is performed to determine the principal local cluster to which the current supervoxel belongs. All supervoxels belonging to the same principal local cluster but not the current supervoxel are considered as the true neighboring voxels of the current supervoxel, generating a supervoxel adjacency set. Obtain the adjacency set It can effectively avoid the erroneous connectivity between parallel walls caused by the traditional k-nearest neighbor method.

[0023] S5. Initialize all supervoxels in the supervoxel adjacency set to an unvisited state. Select the supervoxel with the smallest curvature as the growth seed and add it to the growth queue. Sequentially retrieve the current supervoxel from the growth queue, traverse the adjacent supervoxels, and make judgments based on the geometric constraints of normal consistency and curvature similarity between the current supervoxel and its adjacent supervoxels. Normal consistency geometric constraints and curvature similarity geometric constraints include: ; ; In the formula, This represents the absolute value of the angle between the normal vectors of the current supervoxel and its neighboring supervoxels. The average normal vector of the current supervoxel. Let be the average normal vector of the neighboring supervoxels of the current supervoxel. This represents the absolute difference in curvature between the current supervoxel and its neighboring supervoxels. The curvature value of the current supervoxel. The curvature value of the adjacent supervoxels of the current supervoxel; Set the threshold for the angle between normal vectors Threshold for difference between curvature value and , will simultaneously satisfy and The adjacent supervoxels are marked as visited and added to the growth queue, and simultaneously merged into the current growth cluster. The current growth cluster is added to the global result set. From the remaining unvisited supervoxels, the unvisited supervoxel with the smallest curvature is selected as the new growth seed. The seed selection, traversal filtering and merging operations are iteratively executed until all supervoxels have been traversed, and finally a wall structure region with consistent geometric features is obtained.

[0024] S6. After the supervoxel region growth and clustering in S5, wall structure regions with consistent geometric features are obtained. However, the wall edges and corner structures are prone to fluctuations due to local geometric features. Affected by the dual geometric constraints of normal consistency and curvature similarity, the structural protrusions are prone to forming fragmented small voxels that cannot be merged, which destroys the overall continuity of the wall and affects the integrity of the wall. Therefore, the supervoxel point cloud after growth and clustering is subjected to fine repair and integrity optimization processing: the supervoxel labels of each three-dimensional point are restored according to the RGB encoding of the input point cloud, and the point cloud is projected onto a two-dimensional plane. Let three-dimensional points In the two-dimensional plane coordinates of the projection for: ; In the formula, Adding clouds to the wall Three-dimensional points Two-dimensional coordinates after projection onto a two-dimensional plane. and The first Three-dimensional points The coordinate values ​​corresponding to the two coordinate axes projected onto a two-dimensional plane, and the combination of coordinate axes satisfies ; A grid is constructed based on the two-dimensional planar coordinates of the projection and the grid grid side length is used. Spatially discretize the two-dimensional projection points of each point in the wall point cloud to obtain a two-dimensional label raster. : ; In the formula, For the first The two-dimensional coordinates of a three-dimensional point projected onto a two-dimensional plane The supervoxel label of the grid location; Adding clouds to the wall A 3D point hypervoxel label; To avoid fragmented hypervoxels dominating subsequent processing, the number of points for each hypervoxel is counted, and a set of large patches is constructed: ; In the formula, It is a collection of large pieces of surface. For the first The number of point clouds corresponding to a large area of ​​a supervoxel. For large area thresholds; When the number of point clouds corresponding to a large area of ​​a supervoxel exceeds At that time, it was assumed that the region corresponding to the supervoxel had sufficient geometric stability in the wall structure. Since the wall point cloud would exhibit local fractures in three-dimensional space, connectivity constraints needed to be applied to large patches in two-dimensional raster space for each large patch. Constructing in a two-dimensional label grid Corresponding binary mask : ; In the formula, For the first A large piece of dough The corresponding binary mask, For two-dimensional label grids at grid positions Supervoxel tags at the location, For the first A large, flat hypervoxel label; right Perform morphological dilation operation. The expansion radius is determined by the number of iterations. Controlling the process to obtain the expanded binary mask The dilated binary mask : ; In the formula, For the first A large piece of dough The corresponding dilated binary mask, For the first A large piece of dough The corresponding binary mask, This represents the number of iterations. The first A large piece of dough Corresponding dilated binary mask The covered area is considered as the outer extension of the wall structure in two-dimensional space, and any three-dimensional point in the wall point cloud is considered as such. Get , Falling into A large piece of dough Corresponding dilated binary mask Within the covered area, three-dimensional points The hypervoxel label has been updated to: ; In the formula, Adding clouds to the wall Three-dimensional points Updated hypervoxel labels; 3D points in wall point cloud of Not by any If covered, then three-dimensional points are considered. 3D points not belonging to the wall structure are directly removed. After updating the labels and removing the uncovered point clouds from the wall point cloud, a complete wall point cloud is obtained.

[0025] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for indoor point cloud wall segmentation based on density-constrained super voxel, characterized in that, The method comprises the following steps: S1. Collecting wall point cloud and removing invalid points, and dividing into vertical structure point cloud and horizontal structure point cloud according to the included angle between the normal vector direction and the gravity direction; S2. The vertical structure point cloud is subjected to horizontal plane gridding processing, a grid density map and a grid height map are constructed, the grid density map and the grid height map are respectively subjected to smoothing processing, and density constraint and grid constraint are applied, the point cloud in the grid meeting the density constraint and the height constraint is back projected to obtain a preliminary wall point cloud set; S3. Taking the preliminary wall point cloud set as a processing object, constructing a super voxel by using a boundary of an energy function optimization and a super voxel segmentation method; A super voxel set with an optimized boundary is generated by defining an energy function describing the structure characteristics of the super voxel and minimizing the energy function; S4. The super voxel point cloud of the super voxel set is subjected to initialization, seed screening and adjacent relationship construction to obtain a super voxel adjacency set; S5. All super voxels in the super voxel adjacency set are initialized as unvisited states, the unvisited super voxel with the minimum curvature is taken as a growth seed, the adjacent super voxels are traversed based on the adjacency relationship, and after screening by the normal consistency geometric constraint and the curvature similarity geometric constraint, the super voxels are merged into the current growth cluster, the super voxel with the minimum curvature is selected from the remaining unvisited super voxels as a new seed, and the seed selection, traversal screening and merging operations are iteratively performed until all super voxels are traversed to obtain a super voxel cluster corresponding to an independent wall surface and having consistent geometric characteristics; S6. Based on the wall point cloud corresponding to the super voxel cluster, the super voxel labels of the three-dimensional points are recovered and projected to a two-dimensional plane, a label grid is generated, a binary mask is constructed and morphological dilation is performed, the labels in the dilated binary mask are updated, the uncovered point cloud is removed, and a complete wall surface point cloud is obtained.

2. The indoor point cloud wall surface segmentation method based on density constraint super voxel according to claim 1, wherein, The included angle between the normal vector direction and the gravity direction comprises: ； wherein is the angle between the normal vector direction of the point cloud and the direction of gravity, is the normal vector of the point cloud, is the gravity direction vector; Setting a vertical structure included angle threshold and a horizontal structure included angle threshold , is a vertical structure point cloud, is a horizontal structure point cloud.

3. The indoor point cloud wall segmentation method based on density constraint super voxel according to claim 1, characterized in that, The vertical structure point cloud is horizontally gridded, including gridding the vertical structure point cloud on a horizontal plane, setting the grid mesh length as , and the two-dimensional grid coordinates of the vertical structure point cloud : ； ； wherein, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction, is the horizontal coordinate of the i-th point cloud in the horizontal plane X direction.

4. The indoor point cloud wall surface segmentation method based on density constraint super voxel according to claim 1, characterized in that, Smoothing is applied to both the raster density map and the raster height map, and density and raster constraints are imposed, including based on the two-dimensional raster coordinates of each point cloud. Statistical analysis of two-dimensional grid coordinates The number of inset vertical structure point clouds is used to construct a raster density map. ,right Applying a two-dimensional mean filter to reduce noise yields a smoothed density map. The following wall density constraint formula is used to select grids that meet the density requirements: ； In the formula, is a grid density screening mark, is a point cloud corresponding to a grid with coordinates in the smoothed grid density map, is a density threshold value; According to the two-dimensional grid coordinates of each point cloud , the number of vertical structure point clouds falling into the two-dimensional grid coordinates is counted, and a grid height map is constructed , taking the maximum height value of all point clouds in each grid as the height value of the corresponding grid; the grid height map is smoothed by applying maximum value filtering to obtain a filtered height map , and the following wall height constraint formula is used to screen grids that meet the height requirements: ； In the formula, is a grid height filter mark, is a point cloud height corresponding to a grid with coordinates in the smoothed grid height map, is a ceiling height, is a low structure filter.

5. The method of claim 1, wherein the method is based on density-constrained super voxel. The preliminary wall point cloud set is obtained by back projecting the point cloud in the grid meeting the density constraint and the height constraint, which comprises judging the wall grid by the joint constraint of the density constraint and the height constraint, and the joint constraint is calculated by the following formula: ； In the formula, is a grid final screening identifier, is a grid density screening identifier, is a grid height screening identifier; The point cloud within the grid satisfying is back-projected to the three-dimensional coordinate space to obtain a preliminary wall point cloud set.

6. The method of claim 1, wherein the method is based on density-constrained super voxel. The energy function comprises a difference degree term and a constraint term, and the energy function is calculated by the following formula: ； ； ； In the formula, is a point cloud is a difference value of a point cloud is a first three-dimensional point in a preliminary wall point cloud is a first three-dimensional point in a preliminary wall point cloud is a first three-dimensional point in a preliminary wall point cloud is a first three-dimensional point in a preliminary wall point cloud is a normal vector of is a normal vector of represents a distance of a neighborhood point is a statistical value of a super voxel quantity, and Z is a result set of super voxel segmentation is a segmentation relationship matrix is an indicator function is a total number of three-dimensional points in a preliminary wall point cloud participating in energy function optimization is a minimum optimization target of an energy function is an energy function of point cloud segmentation is a preset generated super voxel quantity is a weight coefficient​​​ 7. The method of claim 1, wherein the method is based on density-constrained super voxel. Initialization includes dividing the point cloud into several initial regions based on the color features of the supervoxels, and calculating the centroid for each supervoxel. Average normal vector curvature and the number of point clouds Complete the extraction of core geometric features, set all supervoxels to an unvisited initial state, and aggregate the region growing results. Leave blank; Seed selection, including ordering the super-voxels in ascending order according to the curvature value, and selecting the super-voxel with the minimum curvature value which is currently in the unvisited state As the initial seed of region growing; if there is no super-voxel satisfying the unvisited condition after traversal, the algorithm terminates, and the super-voxel satisfying the visited condition is added to the growing queue And the state of the super-voxel is marked as visited simultaneously The adjacency relationship construction comprises taking the center of the super voxel For the spatial search reference, a k-nearest neighbor search is performed on each super voxel to obtain a corresponding local candidate neighborhood; all three-dimensional point coordinates contained in all super voxels in the candidate neighborhood are collected, and a density-based noisy space clustering operation is performed based on the point coordinates to determine the main local cluster to which the current super voxel belongs; all super voxels belonging to the same main local cluster and not being the current super voxel are regarded as the real adjacent voxels of the current super voxel, and a super voxel adjacency set is generated .

8. The indoor point cloud wall segmentation method based on density constraint super voxel according to claim 1, characterized in that, The normal consistency geometric constraint and the curvature similarity geometric constraint comprise: ； ； In the formula, is the absolute value of the angle between the normal vector of the current super voxel and the normal vector of the adjacent super voxel, is the average normal vector of the current super voxel, is the average normal vector of the adjacent super voxel of the current super voxel, is the absolute difference value of the curvature value of the current super voxel and the curvature value of the adjacent super voxel, is the curvature value of the current super voxel, is the curvature value of the adjacent super voxel of the current super voxel; Setting a threshold for the angle between normal vectors and a threshold for the difference in curvature values will simultaneously satisfy and marking the adjoining super-voxel as visited and adding it to the growing queue while incorporating it into the current growing cluster.

9. The indoor point cloud wall segmentation method based on density-constrained super voxel according to claim 1, characterized in that, Based on the wall point cloud corresponding to the super voxel cluster, the super voxel label of each three-dimensional point is recovered and projected to a two-dimensional plane to generate a label grid, then a binary mask is constructed and morphological dilation is performed, including three-dimensional points In the projected two-dimensional plane coordinates is: ； In the formula, is the first three-dimensional point in the wall point cloud is the second three-dimensional point in the wall point cloud is the two-dimensional coordinate of the first three-dimensional point after projection onto a two-dimensional plane, and are the two-dimensional coordinates of the second three-dimensional point after projection onto a two-dimensional plane, is the first three-dimensional point in the wall point cloud is the second three-dimensional point in the wall point cloud ; A grid is constructed according to the two-dimensional plane coordinates of the projection and the grid mesh length The two-dimensional projection points of each point in the wall point cloud are spatially discretized to obtain a two-dimensional label grid : ； In the formula, is the second dimensional coordinate of the projection of the third dimensional point to the two-dimensional plane is the voxel label of the wall point cloud is the voxel label of the third Counting the number of point clouds contained in the super voxel, setting a large facet threshold , constructing a large facet set, the large facet set being: ； In the formula, is a set of large patches, is the number of point clouds of large patches corresponding to the th super voxel, is a large patch threshold value; When the number of point clouds corresponding to a super voxel exceeds , for each large facet in the large facet set , a corresponding binary mask is constructed in a two-dimensional label grid : ； In the formula, For the first A large piece of dough The corresponding binary mask, For two-dimensional label grids at grid positions Supervoxel tags at the location, For the first A large, flat hypervoxel label; To perform a morphological dilation operation, the dilation radius is controlled by the number of iterations to obtain a dilated binary mask : ； wherein is the th large facet corresponding inflated binary mask, is the th large facet corresponding binary mask, is the iteration number.

10. The indoor point cloud wall segmentation method based on density-constrained super voxel according to claim 1, wherein, The labels within the expanded binary mask are updated, and the uncovered point cloud is removed to obtain the complete wall point cloud, including the first label. A large piece of dough Corresponding dilated binary mask The covered area is considered as the outer extension of the wall structure in two-dimensional space, and any three-dimensional point in the wall point cloud is considered as such. Get , Falling into A large piece of dough Corresponding dilated binary mask Within the covered area, three-dimensional points The hypervoxel label has been updated to: ； In the formula, is the first three-dimensional point in the wall point cloud is the first three-dimensional point in the wall point cloud is the first three-dimensional point in the wall point cloud Three-dimensional points in a wall point cloud of If any one of the three-dimensional points is not covered, the three-dimensional point is considered not to belong to the wall structure and is directly removed. The three-dimensional point After the wall point cloud is labeled and the uncovered point cloud is removed, a complete wall surface point cloud is obtained.

Images