Building extraction method and system based on multi-dimensional point cloud structure entropy phase change detection

Abstract

The invention relates to the technical field of laser radar three-dimensional perception, and provides a building extraction method and system based on multi-dimensional point cloud structure entropy phase change detection, and the method comprises the steps: carrying out the denoising and non-ground point extraction of original point cloud data, constructing a multi-dimensional structure entropy field fusing a geometric structure, echo distribution and elevation consistency, and carrying out the phase change detection of the multi-dimensional point cloud data; the quantification module is used for quantifying point cloud local structure orderliness; building boundary recognition is achieved through structure entropy gradient field analysis in combination with gradient flow tracking and a persistent coherence method, and candidate building areas are constructed; obtaining high-quality seed points by using multi-modal feature clustering, and forming a building point cloud aggregation area through directional growth driven by structural convergence; and carrying out region fusion and contour extraction based on a minimum description length criterion, and generating a structured building vector expression meeting a quality evaluation requirement. According to the method, the interference of high-confusion objects such as regularly arranged fences, single trees, vehicles and the like is eliminated, and high-precision, high-robustness and high-interpretability building extraction is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of laser radar three-dimensional perception technology, and particularly relates to a building extraction method and system based on multi-dimensional point cloud structure entropy phase transition detection. BACKGROUND

[0002] The airborne laser radar building extraction technology refers to a technical system in which a laser radar sensor is carried on an aerial platform to obtain information such as the geometric shape, spatial distribution and structural features of buildings in a region through three-dimensional point cloud data. Compared with the traditional building extraction method based on two-dimensional images or manual mapping, the airborne laser radar building extraction technology has the advantages of higher three-dimensional precision, stronger structural recognition ability and wider operation coverage. The airborne laser radar can quickly obtain large-scale, high-precision three-dimensional point cloud data, realize efficient and accurate automatic extraction of buildings and three-dimensional reconstruction, and provide timely and accurate data support for city geographic information systems and three-dimensional modeling.

[0003] In the prior art, the airborne laser radar building extraction mainly adopts traditional methods based on geometric features, statistical features and deep learning. Although the basic detection and extraction of buildings are realized, in actual application, there is a lack of general mathematical tools that can effectively quantify the intrinsic order of the local structure of point clouds, and the existing feature extraction methods cannot accurately depict the structural complexity and order difference of building point clouds. Therefore, it is difficult to accurately distinguish the structural differences between buildings and other ground objects in high-demand scenarios such as complex urban environments, especially under the conditions of building and vegetation adhesion and irregular building shapes, resulting in a high missing detection rate of complex structural buildings and a high misjudgment rate of mixed ground objects. SUMMARY

[0004] Therefore, the present application provides a building extraction method and system based on multi-dimensional point cloud structure entropy phase transition detection, which solves the problem that the prior art cannot accurately distinguish the structural differences between buildings and other ground objects in high-demand scenarios such as complex urban environments, especially under the conditions of building and vegetation adhesion and irregular building shapes, resulting in a high missing detection rate of complex structural buildings and a high misjudgment rate of mixed ground objects.

[0005] The technical scheme of the present application is implemented as follows: on the one hand, the present application provides a building extraction method based on multi-dimensional point cloud structure entropy phase transition detection, which comprises the following steps: The original point cloud data is denoised and non-ground points are obtained by using a progressive triangular net encryption filtering algorithm, and a structure entropy field is constructed by weighted fusion of geometric structure entropy, echo distribution entropy and elevation consistency entropy; The structure entropy field quantifies the structural order of the local neighborhood of the point cloud by fusing geometric anisotropy, echo distribution complexity and elevation fluctuation consistency in three dimensions; perform gradient field calculation on the non-ground points based on the structural entropy field, identify boundary structures based on gradient flow tracking and persistent homology, and construct a candidate building region based on a closed-loop boundary ring; construct a multi-modal feature field for the candidate building region, perform feature clustering based on a Gaussian mixture model and a density peak value to obtain a high-quality seed point set; perform seed activation based on the high-quality seed point set, make absorption decisions on candidate points based on a structural convergence dynamic evaluation method, and perform directional growth based on the structural entropy field to obtain a building point cloud aggregation region; perform optimal fusion on the building point cloud aggregation region based on a minimum description length metric method, convert the fused three-dimensional point cloud building object into a two-dimensional vector contour, and perform quality evaluation on the two-dimensional vector contour to obtain a structured building expression.

[0006] In the above technical solutions, preferably, the gradient field calculation on the non-ground points based on the structural entropy field, the identification of boundary structures based on gradient flow tracking and persistent homology, and the construction of a candidate building region based on a closed-loop boundary ring include: perform gradient field calculation and significance analysis on the three-dimensional structural entropy field of the non-ground points, fit the scalar entropy value field of the point neighborhood using a moving least squares method, obtain local linear fitting coefficients by solving a weighted least squares problem, calculate the entropy gradient vector and gradient amplitude at each point, and screen out a significant gradient point set that meets the gradient amplitude condition; use the significant gradient point set as seed points, perform bidirectional streamline tracking along the gradient direction, use a fourth-order Runge-Kutta method to numerically integrate the streamline equation to obtain boundary segments, identify significant boundary structures in the boundary segments based on a topological feature importance evaluation scheme of persistent homology, and construct the candidate building region based on a closed-loop boundary ring.

[0007] In the above technical solutions, preferably, the gradient field calculation and significance analysis on the three-dimensional structural entropy field of the non-ground points include: calculate the dot product of the gradient direction and the local normal vector to determine the positive and negative gradients, the positive gradient of the positive and negative gradient represents the building pointing to the environment corresponding to the building outer boundary, the negative gradient of the positive and negative gradient represents the environment pointing to the building interior corresponding to the hole or courtyard boundary, calculate the statistics of the global gradient amplitude distribution, and screen out the significant gradient point set that meets the gradient amplitude greater than the mean value plus the standard deviation multiple.

[0008] Based on the above technical solutions, preferably, the step of using the set of significant gradient points as seed points, performing bidirectional streamline tracing along the gradient direction, obtaining boundary segments by numerically integrating the streamline equation using the fourth-order Runge-Kutta method, identifying significant boundary structures in the boundary segments based on a persistent homology-based topological feature importance assessment scheme, and constructing the candidate building region based on a closed-loop boundary ring includes: For points with positive gradients, the entropy-increasing path from the building's interior to the exterior is simulated by integrating along the positive gradient direction; for points with negative gradients, the entropy-decreasing path from the exterior to the building's interior is simulated by integrating along the negative gradient direction. An ordered set of points is generated for each streamline, and spline fitting is performed to obtain a smooth, parametric curve representation, thus constructing a candidate point set. α The complex is used to calculate the birth and death times of homology group generators in each dimension, select topological features that meet the persistence threshold, and extract the corresponding boundary segments based on the streamline point set corresponding to the topological features.

[0009] Based on the above technical solutions, preferably, the optimal fusion of the building point cloud aggregation region using the minimum description length metric method, the conversion of the fused 3D point cloud building object into a 2D vector contour, and the quality evaluation of the 2D vector contour to obtain a structured building representation, includes: The optimal fusion of the building point cloud aggregation region is performed based on the minimum description length metric method between regions. The total description length is defined as the weighted sum of the geometric description length and the structural entropy description length. The fusion benefit and fusion confidence of candidate fusion pairs are calculated, and the fusion of regions that meet the conditions is performed to obtain the final set of building objects. Based on adaptive α-shape contour extraction, the final set of building objects is subjected to contour polygon extraction. The interior angle size is calculated by detecting the contour angle, and the interior angle that is close to 90 degrees is adjusted to 90 degrees to obtain the two-dimensional vector contour. The geometric fidelity and area consistency are calculated to perform the quality assessment and generate the structured building representation.

[0010] Based on the above technical solutions, preferably, the method based on minimum description length between regions is used to optimally fuse the building point cloud aggregation regions. The total description length is defined as the weighted sum of the geometric description length and the structural entropy description length. The merging benefit and merging confidence of candidate merging pairs are calculated, and region merging that meets the conditions is performed to obtain the final set of building objects, including: The geometric description length includes boundary complexity and internal structural complexity. The structural entropy description length is calculated based on the covariance of the regional point entropy value distribution and the KL divergence between the regional distribution and the global building point distribution. Based on spatial adjacency, regions with effective adjacency distances less than a preset value are retained. Regions with merging benefits less than a preset value and merging confidence greater than a preset value are merged.

[0011] Based on the above technical solutions, preferably, the step of denoising the original point cloud data and obtaining non-ground points using a progressive triangulation filtering algorithm, and constructing a structural entropy field through weighted fusion of geometric structure entropy, echo distribution entropy, and elevation consistency entropy, includes: The original point cloud data is subjected to a statistical outlier removal method to eliminate sensor noise and outliers, and the point cloud is subjected to regular resampling using a voxel grid method to obtain a resampled point cloud. The resampled point cloud is separated using a progressive triangulation filtering algorithm to obtain the non-ground points; A multi-scale adaptive neighborhood is constructed based on the characteristic scale of the building components. By fusing the weighted geometric average of the geometric structure entropy, echo distribution entropy, and elevation consistency entropy, the credibility weight of the entropy value at each scale is dynamically determined based on the local terrain complexity, and the final representative entropy value of each point is generated to construct the structural entropy field.

[0012] Based on the above technical solutions, preferably, the step of constructing a multimodal feature field for the candidate building region and performing feature clustering based on a Gaussian mixture model and density peaks to obtain a high-quality seed point set includes: For each point within the candidate building area, calculate the local point density, normalized relative elevation, and final structural entropy. Perform robust statistical calculations and winsorization on each feature dimension to establish a mapping from the three-dimensional coordinate space to the feature space, and obtain the feature point set distribution. The optimal number of components is determined by the Bayesian information criterion and a Gaussian mixture model is fitted. The local density and minimum distance of each point are calculated in the feature space. The posterior probability of the Gaussian mixture model is fused with the density peak information. Based on Mahalanobis distance, similar clusters are merged to obtain the high-quality seed point set.

[0013] Based on the above technical solutions, preferably, the step of seed activation based on the high-quality seed point set, absorption decision-making for candidate points based on the structural convergence dynamic evaluation method, and directional growth based on the structural entropy field to obtain the building point cloud aggregation region includes: A weighted undirected graph is constructed based on the high-quality seed point set, and confidence propagation iteration is performed to obtain the activation sequence. An initial growth region is created starting from the seed point with the highest confidence. The baseline value of the region structure entropy and the dominant direction of the region normal vector are calculated. The active boundary set, candidate point buffer, growth history and dynamic convergence threshold are initialized. Starting from each point in the active boundary set, search for unlabeled candidate points. For each unlabeled candidate point, calculate the local structure compatibility, regional entropy change prediction, and topological continuity index. Construct an absorption decision state vector, combine it with a pre-trained policy network and rule thresholds to make absorption decisions, and perform batch absorption to update the growing region.

[0014] On the other hand, the present invention also provides a building extraction system based on multidimensional point cloud structural entropy phase transition detection, the system comprising: The data preprocessing and structural entropy field construction module is used to denoise the original point cloud data and obtain non-ground points using a progressive triangulation encryption filtering algorithm. The structural entropy field is constructed by weighted fusion of geometric structure entropy, echo distribution entropy and elevation consistency entropy. The gradient field calculation and boundary identification module is used to calculate the gradient field of the non-ground points based on the structural entropy field, identify the boundary structure based on gradient flow tracking and persistent cohomology, and construct candidate building regions based on closed-loop boundary rings. The multimodal feature clustering module is used to construct a multimodal feature field for the candidate building region, and perform feature clustering based on Gaussian mixture model and density peak to obtain a high-quality seed point set; The seed activation and region growth module is used to activate seeds based on the high-quality seed point set, make absorption decisions on candidate points based on the structural convergence dynamic evaluation method, and perform directional growth based on the structural entropy field to obtain the building point cloud aggregation region. The region fusion and contour extraction module is used to perform optimal fusion of the building point cloud aggregation region based on the minimum description length metric method, convert the fused 3D point cloud building objects into 2D vector contours, evaluate the quality of the 2D vector contours, and obtain a structured building representation.

[0015] The building extraction method and system based on multidimensional point cloud structural entropy phase transition detection of the present invention have the following advantages over the prior art: (1) By constructing a multidimensional point cloud structure entropy phase transition detection mechanism, a paradigm shift from traditional geometric feature segmentation to structural order phase transition detection has been realized. In complex urban environments, it can accurately distinguish the structural differences between buildings and vegetation and other ground objects, effectively eliminate the interference of highly confusing objects such as regularly arranged fences, single trees, and vehicles, and realize high-precision, high-robustness, and high-interpretability building extraction. (2) By integrating gradient field analysis and topological cohomology theory, the entropy field gradient is calculated and significance is screened using the moving least squares method. The positive and negative gradient directions are combined to distinguish the outer boundary of the building from the inner boundary of the courtyard. The fourth-order Runge-Kutta method is used to track the boundary segments in both directions, thus realizing high-precision automatic detection of building boundaries and construction of complete closed-loop regions. (3) By using the minimum description length metric method that integrates geometric description length and structural entropy description length, the optimal fusion of building aggregation areas is achieved by using merging benefits and confidence assessment. The contour parameters are dynamically adjusted by combining the adaptive α-shape contour extraction method, and the interior angles close to 90 degrees are normalized according to angle detection. This improves the integrity and geometric regularity of the vectorized representation of buildings. At the same time, the quantitative quality assessment indicators meet the high-precision application requirements of geographic information systems and 3D modeling. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a building extraction method based on multidimensional point cloud structural entropy phase transition detection according to the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 This invention provides a method for extracting buildings based on multidimensional point cloud structural entropy phase transition detection, comprising the following steps: The original point cloud data is denoised and non-ground points are obtained by progressive triangulation filtering algorithm. The structural entropy field is constructed by weighted fusion of geometric structure entropy, echo distribution entropy and elevation consistency entropy. The structural entropy field quantifies the structural orderliness of the local neighborhood of the point cloud by integrating three dimensions: geometric anisotropy, echo distribution complexity, and elevation undulation consistency. Gradient field calculation is performed on the non-ground points based on the structural entropy field. Boundary structure is identified based on gradient flow tracking and persistent homology. Candidate building regions are constructed based on closed-loop boundary rings. A multimodal feature field is constructed for the candidate building region, and feature clustering is performed based on Gaussian mixture model and density peak to obtain a high-quality seed point set; Seed activation is performed based on the high-quality seed point set, absorption decision is made for candidate points based on the structural convergence dynamic evaluation method, and directional growth is performed based on the structural entropy field to obtain the building point cloud aggregation region. The point cloud aggregation region of the building is optimally fused based on the minimum description length metric method. The fused 3D point cloud building object is converted into a 2D vector contour. The quality of the 2D vector contour is evaluated to obtain a structured building representation.

[0020] Specifically, this embodiment constructs a multi-dimensional point cloud structure entropy phase transition detection mechanism, realizing a paradigm shift from traditional geometric feature segmentation to structural order phase transition detection. This solves the problem that existing building extraction methods lack a general tool for quantifying structural order. In complex urban environments, it can accurately distinguish the structural differences between buildings and vegetation, effectively eliminating interference from highly confusing objects such as regularly arranged fences, single trees, and vehicles. It does not require preset geometric models, manual empirical rules, or a large amount of labeled data, achieving high-precision, high-robustness, and high-interpretability building extraction. The output result is an independent building polygon with attributes.

[0021] The process of denoising the original point cloud data and obtaining non-ground points using a progressive triangulation filtering algorithm, and constructing a structural entropy field through a weighted fusion of geometric structure entropy, echo distribution entropy, and elevation consistency entropy, includes: The original point cloud data is subjected to a statistical outlier removal method to eliminate sensor noise and outliers, and the point cloud is subjected to regular resampling using a voxel grid method to obtain a resampled point cloud. The resampled point cloud is separated using a progressive triangulation filtering algorithm to obtain the non-ground points; A multi-scale adaptive neighborhood is constructed based on the characteristic scale of the building components. By fusing the weighted geometric average of the geometric structure entropy, echo distribution entropy, and elevation consistency entropy, the credibility weight of the entropy value at each scale is dynamically determined based on the local terrain complexity, and the final representative entropy value of each point is generated to construct the structural entropy field.

[0022] In one specific embodiment, a statistical outlier removal method is used to eliminate sensor noise and abnormal points in the acquired raw point cloud dataset. Simultaneously, to avoid the impact of uneven point cloud density on local computation, a voxel grid method is used to perform regular resampling of the point cloud, providing a uniform data foundation for subsequent steps. In this embodiment, the voxel grid size is set to 0.1m.

[0023] An improved Asymptotic Triangulation Network (ATIN) filtering algorithm is used to separate ground points from non-ground points to obtain a set of non-ground points. ; Based on the characteristic scales of building components, a multi-scale adaptive domain structure entropy is constructed, the steps of which include: Step 1: Based on the typical dimensions of urban buildings, three analysis scales are preset:

[0024] Among them, the first analytical scale Suitable for detailed scales (such as window frames, railings, etc.), second analytical scale Suitable for component scales (such as walls, standard roof panels, etc.), third analysis scale Suitable for structural scales (such as complete roofs, building structures, etc.). For each scale... Establish the corresponding KD-tree spatial index;

[0025] Step 2, for each point in the non-ground point set At each analytical scale Search for the center of the ball , radius is Construct a domain point set from all points within the domain. If the neighborhood point set The number of interior points is less than (usually set) If the entropy value of that analysis scale is invalid, then the entropy value of that analysis scale is marked as invalid.

[0026] Step 3, for each valid domain The geometric entropy is calculated as follows: echo distribution entropy Elevation Consistency Entropy : For geometric structure entropy Calculate the covariance matrix of points within the neighborhood to obtain the first, second, and third eigenvalues. ,definition ,in , These respectively characterize the intensity of linear, planar, and volumetric distributions of local structures. Then:

[0027]

[0028] in, The value is a very small positive number to prevent errors in logarithmic calculations. A confidence weight is also introduced. , For domain points, As an empirical threshold, this embodiment selects... The physical meaning of the geometric entropy is: when it is an ideal plane (such as a roof), it corresponds to... Random distribution (e.g., inside the tree canopy) corresponds to .

[0029] For echo distribution entropy : The distribution of echo frequency at points within a statistical domain. Let the th... The percentage of points in the secondary echo is ,but The physical meaning of this echo distribution entropy is: the surface of a rigid building mainly generates single or initial echoes, which are concentrated in a specific distribution. The vegetation canopy easily generates multiple echoes, and its distribution is uniform. High value.

[0030] Elevation Consistency Entropy Divide the range of elevation values ​​of points within the domain into B equal intervals, and calculate the probability that an elevation value falls within each interval. .but The physical meaning of this elevation uniformity entropy is: for flat or regularly sloping roofs, the elevations are concentrated. The vegetation canopy exhibits dramatic elevation variations and is sparsely distributed. High value.

[0031] The three sub-entropies are fused using a weighted geometric mean:

[0032] Recording point Multiscale entropy values ​​at three analytical scales: First analytical scale entropy value Second analytical scale entropy value The third analytical scale entropy value .

[0033] Based on local terrain features, the multi-scale entropy values ​​are fused to generate a final representative entropy value for each point. The main steps include: Step 4, for each point In the second analytical scale ( ) Calculate terrain complexity index within the domain :

[0034] in, The first, second, and third eigenvalues ​​of the neighborhood point coordinate covariance matrix ( ), The variance of the elevation of the neighborhood points. Normalized. arrive Interval, as a terrain complexity coefficient .

[0035] Step 5, based on the terrain complexity coefficient Dynamically determine the credibility weights of entropy values ​​at each scale:

[0036] in, The attenuation coefficient is 0.3 in this embodiment. As the first credibility weight, As the second credibility weight, This is the third confidence weight. At this point, when the point is in a simple platform region, Large-scale information is more reliable. Large; in complex and delicate areas, Small-scale information is more reliable. big.

[0037] Step 6: Generate points using a robust fusion strategy. The final structural entropy value :

[0038] in, This represents the final structural entropy value. As the first credibility weight, As the second credibility weight, As the third credibility weight; , , These are the entropy values ​​of the first analytical scale. Second analytical scale entropy value The third analytical scale is the entropy value.

[0039] The steps to construct a pyramid representation of the structural entropy field include: Step 7, set the original resolution as... Generated by stepwise downsampling First low-resolution layer, second low-resolution layer, third low-resolution layer. For each layer... Perform voxelization (voxel size is For each downsampled point, the entropy distribution statistical characteristics (mean, median, minimum) of its original neighborhood (corresponding to multiple points in the original point cloud) are transferred to that point.

[0040] Step 8, for the original resolution level For each point, check its position via spatial coordinate mapping. Entropy statistics for the corresponding region. If a point exhibits significantly low entropy across multiple resolution levels (i.e., below the global threshold for each sub-level), then the confidence level of that point as a building point is increased by multiplying its final entropy value by a decay factor. Attenuation factor In this embodiment, 0.9 is used; if the point only exhibits low entropy at the finest level but high entropy at the coarse level, it may be an object that is ordered at a small scale but disordered at a large scale, such as a single vehicle, and the final entropy value of the point can be appropriately increased.

[0041] The process of calculating the gradient field for the non-ground points based on the structural entropy field, identifying boundary structures based on gradient flow tracing and persistent cohomology, and constructing candidate building regions based on closed-loop boundary rings includes: The gradient field of the three-dimensional structural entropy field of the non-ground points is calculated and its significance is analyzed. The scalar entropy field of the neighborhood of the point is fitted by the moving least squares method. The local linear fitting coefficient is obtained by solving the weighted least squares problem. The entropy gradient vector and gradient magnitude at each point are calculated, and the set of significant gradient points that meet the gradient magnitude condition is selected. Using the set of significant gradient points as seed points, bidirectional streamline tracing is performed along the gradient direction. The fourth-order Runge-Kutta method is used to numerically integrate the streamline equation to obtain the boundary segments. The significant boundary structures in the boundary segments are identified based on the persistent homology topological feature importance evaluation scheme, and the candidate building regions are constructed based on the closed-loop boundary ring.

[0042] The gradient field calculation and significance analysis of the three-dimensional structural entropy field for the non-ground points are performed. The moving least squares method is used to fit the scalar entropy field of the point's neighborhood. Local linear fitting coefficients are obtained by solving a weighted least squares problem. The entropy gradient vector and gradient magnitude at each point are calculated, and a set of significant gradient points satisfying the gradient magnitude condition is selected, including: The positive and negative gradients are determined by calculating the dot product of the gradient direction and the local normal vector. The positive gradient represents the direction from the building to the outer boundary of the corresponding building in the environment, and the negative gradient represents the direction from the environment to the boundary of the corresponding hole or courtyard inside the building. The statistics of the global gradient magnitude distribution are calculated, and the significant gradient point set that satisfies the requirement that the gradient magnitude is greater than a multiple of the mean plus the standard deviation is selected.

[0043] The process involves using the set of significant gradient points as seed points, performing bidirectional streamline tracing along the gradient direction, numerically integrating the streamline equations using the fourth-order Runge-Kutta method to obtain boundary segments, identifying significant boundary structures within these segments based on a persistent homology-based topological feature importance assessment scheme, and constructing the candidate building regions based on closed-loop boundary rings. For points with positive gradients, the entropy-increasing path from the building's interior to the exterior is simulated by integrating along the positive gradient direction; for points with negative gradients, the entropy-decreasing path from the exterior to the building's interior is simulated by integrating along the negative gradient direction. An ordered set of points is generated for each streamline, and spline fitting is performed to obtain a smooth, parametric curve representation, thus constructing a candidate point set. α The complex is used to calculate the birth and death times of homology group generators in each dimension, select topological features that meet the persistence threshold, and extract the corresponding boundary segments based on the streamline point set corresponding to the topological features.

[0044] In one specific embodiment, the gradient field calculation and significance analysis of the three-dimensional structural entropy field are performed on the non-ground points. The specific steps include: Step 9, based on non-ground points , The structural entropy value is Build a KD-tree spatial index for each point. Search its radius k nearest neighbors , Typically, a distance of 1.0 to 1.5 times the average point spacing is used. The moving least squares method is employed to fit the points. Scalar entropy field of the domain :

[0045] ; in, basis functions Gaussian weighting function , By solving this weighted least squares problem, the local linear fitting coefficients are obtained. ,point The entropy gradient vector at point is:

[0046] ; Gradient magnitude is .

[0047] Step 10: Calculate the gradient direction and local normal vector using PCA. dot product: ; Positive gradient ( ): This indicates the direction of increasing entropy outward along the normal, that is, from the building (low entropy) to the environment (high entropy), corresponding to the outer boundary of the building; negative gradient ( ): This indicates the direction inward along the normal direction, where the entropy value decreases, i.e., from the environment towards the interior of the building, corresponding to the boundary of a hole or inner courtyard.

[0048] Step 11, calculate the statistics of the global gradient magnitude distribution: mean. and standard deviation Select the set of significant gradient points that meet the following conditions. : ,in, The sensitivity parameter is set to 1.0 in this embodiment. The location of each significant gradient point is also recorded. gradient vector Gradient magnitude Positive and negative .

[0049] The candidate boundary curve segment is extracted by using gradient flow tracing to extract significant gradient points. The steps include: Step 12, set the significant gradient points As a seed point, perform bidirectional streamline tracing along the gradient direction (considering positive and negative values): Forward tracing of positive gradient points: along Directional integrals are used to simulate the entropy-increasing path from the inside of a building to the outside.

[0050] Backtracking of negative gradient points: along Directional integrals are used to simulate the entropy-reducing path from the outside into the building's interior.

[0051] Numerical integration of the streamline equations using the fourth-order Runge-Kutta method: , to the step length Adaptive adjustments are made. The tracking stops under the following conditions: 1. Reaching the point cloud boundary; 2. The gradient magnitude falls below a threshold. 3. Tracking length exceeds the maximum allowed length. .

[0052] Step 13: Generate an ordered set of points for each streamline. For each curve Perform spline fitting to obtain a smooth parametric curve representation. By resampling at equal intervals from each smooth curve, a discrete sequence of boundary points is obtained. This ensures that the distance between adjacent points is approximately the average distance between points. For each boundary point, record its associated curve ID, parameter position, and gradient direction at that point as an estimate of the boundary normal.

[0053] Step 14: Based on the gradient sign and curve shape of the tracking starting point, the boundary segments obtained in Step 13 are divided into three categories: Outer boundary: An open or closed curve generated by a positive gradient seed point, enclosing a low-entropy region; Inner boundary: A closed curve generated by the negative gradient seed point, corresponding to spaces such as courtyards and atriums inside the building; Ambiguous boundaries: Short curves that fail to form a clear closure trend may be noise or overly complex regions.

[0054] A topological feature importance assessment scheme based on persistent cohomology identifies topologically significant boundary structures that persist across multiple scales from a large number of boundary fragments, including the following steps: Step 15: Obtain the boundary point set obtained in Step 13. Construct the boundary point set With the radius parameter From 0 to Calculate the birth time of homology group generators in each dimension. and time of death : 0-dimensional features: the occurrence and merging of connected components; 1D features: Generation and filling of ring structures; 2D features: formation and filling of voids.

[0055] For boundary detection, this embodiment focuses on 0-dimensional features (boundary endpoints) and 1-dimensional features (boundary loops). The persistence of each topological feature is calculated. : .

[0056] Step 16, Set the persistence threshold ,in, and The mean and standard deviation of the persistence of all 1-dimensional features. Typically, a value of 1.0 is used. Filtering is performed to find those that meet the criteria. The significant one-dimensional topological features, namely the "long-lived" ring structures, are mapped back to the original boundary fragments, i.e., those that formed earlier in the filtering process. Smaller, disappeared later ( (Large) boundary curve. For each boundary segment... Assign a topological importance score :

[0057] ; in, It is the persistence of associated topological features. This represents the maximum durability. It is the curve length. This represents the maximum curve length.

[0058] Based on the construction of a closed-loop boundary ring, candidate building regions are divided. This includes the following steps:

[0059] Step 17, Establish the boundary segment connection graph: If two boundary segments and The endpoint distance is less than the connection threshold , Typically, twice the average point spacing is used; in this embodiment, 0.3m is used, and the gradient direction angle at the endpoints is less than... (like If any of these connections are connected, then they are considered to be connectable. Based on the constructed connection graph, a depth-first search is used to find closed loops.

[0060] 0-preferred connections are made to segments with high topological importance scores; ② Geometric bridging is allowed at the gap: if the distance between the two endpoints is within If there are no significant gradient points in the intermediate region, connect them with straight line segments. Output a set of closed boundary loops. Each ring is represented by an ordered sequence of boundary points.

[0061] Step 18, for each closed loop Calculate its geometric and topological verification metrics: ① Area The area of ​​the ring projected onto the horizontal plane; ② Circularity : ,in Perimeter; ③Internal average entropy The average structural entropy of all points within the ring. ④ Boundary gradient consistency : Average cosine similarity of gradient directions of adjacent boundary points on the ring.

[0062] Set a threshold to filter out non-building loops, with the following retention criteria: ; in, To minimize the area, The average entropy of all non-ground points. To achieve the minimum roundness, in this embodiment... , .

[0063] Step 19, each verified boundary ring A candidate building area was defined. At the same time, it is necessary to handle the inclusion and intersection relationships between rings:

[0064] Inclusion relation handling: If a cycle Completely contained in another ring Internally, a parent-child hierarchical relationship is established. Marked as inner boundary, Marked as outer boundary; Intersection handling: If two rings intersect, calculate the intersection angle and area overlap ratio. If the overlap area exceeds 30% of the smaller ring's area and the intersection angle is close to perpendicular, it may be a connected building structure, and the two rings will be merged into a complex boundary.

[0065] Output the final candidate building region geometry. Each region has the following attributes: boundary ring, area, average entropy, and hierarchical relationship (i.e., a list of parent and child regions).

[0066] The process involves constructing a multimodal feature field for the candidate building region, performing feature clustering based on a Gaussian mixture model and density peaks to obtain a high-quality seed point set, including: For each point within the candidate building area, calculate the local point density, normalized relative elevation, and final structural entropy. Perform robust statistical calculations and winsorization on each feature dimension to establish a mapping from the three-dimensional coordinate space to the feature space, and obtain the feature point set distribution. The optimal number of components is determined by the Bayesian information criterion and a Gaussian mixture model is fitted. The local density and minimum distance of each point are calculated in the feature space. The posterior probability of the Gaussian mixture model is fused with the density peak information. Based on Mahalanobis distance, similar clusters are merged to obtain the high-quality seed point set.

[0067] In one specific embodiment, the construction and normalization of a multimodal feature field for the candidate building region includes the following steps: Step 20, for candidate building areas Each point within Calculate three core features: Local point density : Among them, the search radius Set to 0.8m; Normalized relative elevation ,in, and Candidate building areas The lowest and highest elevations of all points within the area are calculated. For areas including the inner boundary, a watershed algorithm is used to identify the main body of the roof, and the elevation range is calculated using only the roof points.

[0068] Final structural entropy The result calculated in step 6 is sufficient.

[0069] Step 21: For each feature dimension calculated in Step 20, calculate the candidate building region. Robust statistics within the range include: median First quartile Third and quartiles ; Interquartile range ; Apply quantile-based winsorization processing: ; Perform processing on the eigenvalues Interval normalization: ; We obtained three normalized features: .

[0070] Step 22, for candidate building areas For all points within the feature space, establish a mapping from the three-dimensional coordinate space to the feature space: ; ; Record candidate building areas Distribution of point sets in feature space : .

[0071] Feature clustering based on Gaussian mixture model and density peaks in the feature space includes the following steps: Step 23, for the feature point set The optimal score is determined using the Bayesian information criterion. :

[0072] in, Let K be the likelihood value of the GMM for the K component. For points, Let be the total number of parameters. The expectation-maximization algorithm is used for fitting. The component-based GMM is used, where each component corresponds to a latent semantic category, such as main roof, auxiliary structures, residual vegetation, noise, etc. The posterior probability of each point belonging to each component is calculated to obtain the soft assignment matrix.

[0073] Step 24: Calculate the local density of each point in the feature space. and minimum distance : ; ; in, Take 5% of the feature space diameter. Calculate the decision value. Local maxima are selected as candidate cluster centers, and the posterior probability of the GMM is fused with the density peak information to correct the cluster center position.

[0074] Step 25, calculate the Mahalanobis distance between each cluster:

[0075] in, Let be the global covariance matrix of the feature space. And based on Mahalanobis distance, clusters of similar elements are merged, i.e., when... If the clusters are merged into the same category, the final cluster set is output. Each cluster Including the center of feature space covariance matrix A set of points in three-dimensional space .

[0076] The process of selectively identifying core seed points for buildings from clusters to extract a high-quality seed point set for the main body of the building includes the following steps: Step 26, for each cluster Calculate the following distinguishing features: Building Core Index : ,in, These represent the mean and weight of the clusters across the three feature dimensions, respectively. .

[0077] Space Compactness : ,in, For the 3D convex hull volume of the clustered point cloud, For surface area, high spatial compactness indicates a compact volumetric structure.

[0078] Planar dominance : , is the eigenvalue of the covariance matrix of the clustered point cloud, and a value close to 1 indicates a strong planar property.

[0079] Step 27: Set constraints to filter out non-building clusters. The retention criteria are as follows: ; Establish a soft rating system: ; in, The entropy represents the three-dimensional spatial distribution of cluster points. According to... Sort by size in descending order and select the first few. One as a candidate cluster for buildings .

[0080] Step 28: Check the spatial relationship between candidate clusters. If the distance between two clusters in three-dimensional space is less than [missing information], then [missing information]. (e.g., 2m), and the elevation difference is less than If the cluster size is 1m, then consider merging the two clusters and calculating the combined score of the merged clusters to obtain the final cluster size. .

[0081] Constructing a multi-level seed point structure for a building seed point set includes the following steps: Step 29, for each cluster finally determined in step 28 Calculate the local building confidence score for each point in the cluster. : ,in Represents Mahalanobis distance, For normalized feature vectors, For clusters The center of the feature space. The top N points with the highest confidence are selected as the core seed point set. . It is an exponential function.

[0082] Step 30, around each core seed point set In three-dimensional space, with each core point as the center and a radius of... Within a sphere (2m in this embodiment), points satisfying the following conditions are selected as seed points for expansion: ; in, The standard deviation of the cluster entropy value. and These represent the mean and standard deviation of the three-dimensional locations of the clusters. Merging all expanded points yields the expanded seed point set. Calculate the growth priority of the expansion point. :

[0083]

[0084] in, The distance to the nearest core point. The maximum distance to the nearest core point. The angle between the normal vector and the dominant plane of the cluster is denoted as .

[0085] Step 31, for each candidate building area Constructing a hierarchical seed point set : .

[0086] The process involves seed activation based on the high-quality seed point set, absorption decision-making for candidate points based on a structural convergence dynamic evaluation method, and directional growth based on the structural entropy field to obtain the building point cloud aggregation region, including: A weighted undirected graph is constructed based on the high-quality seed point set, and confidence propagation iteration is performed to obtain the activation sequence. An initial growth region is created starting from the seed point with the highest confidence. The baseline value of the region structure entropy and the dominant direction of the region normal vector are calculated. The active boundary set, candidate point buffer, growth history and dynamic convergence threshold are initialized. Starting from each point in the active boundary set, search for unlabeled candidate points. For each unlabeled candidate point, calculate the local structure compatibility, regional entropy change prediction, and topological continuity index. Construct an absorption decision state vector, combine it with a pre-trained policy network and rule thresholds to make absorption decisions, and perform batch absorption to update the growing region.

[0087] In one specific embodiment, starting from high-quality seed points in the high-quality seed point set, multi-level seed activation and growth initialization are performed based on confidence propagation, including the following steps: Step 32, for candidate building areas seed point set Construct a weighted undirected graph ,node Includes core seed point and expanding seed points ,side The connection space distance is less than the connection radius ( Seed point pairs (with average point spacing). Edge weights. Based on three-dimensional spatial distance and feature similarity:

[0088] ; in, For nodes The normalized eigenvectors.

[0089] Step 33: Initialize the confidence level of each seed point.

[0090] Perform confidence propagation iterations (usually 5 times).

[0091] ,in The damping coefficient is... For nodes The domain points. Based on the final confidence level. Sort all seed points in descending order to obtain the activation sequence. .

[0092] Step 34: Starting with the seed point with the highest confidence, create the initial growth region. ,calculate Regional structure entropy benchmark value : Simultaneously calculate the dominant direction of the region normal vector. , Obtained through the first principal component of the region point cloud using PCA. Initialize the region growth state parameters, including:

[0093] Active Boundary Set Initially, it is a seed point set; Candidate point buffer Initially empty; Growth history Record the change in region entropy during each iteration; Dynamic convergence threshold Initially set to .

[0094] Based on the dynamic evaluation method of structural convergence, an absorption decision is made for candidate points, including the following steps: Step 35, from the active boundary set Each point in Departure, within radius (First analytical scale) and Search for unlabeled candidate points within the (third analytical scale). For each candidate point... Calculate its absorption assessment characteristics:

[0095] Local structural compatibility : ,in, The local normal vector of the candidate point; Regional entropy change prediction : ; Topological continuity index Calculate the minimum connection distance between candidate points and active boundary points. Calculate the k-nearest neighbor overlap rate between candidate points and points within the region. , .

[0096] Step 36, construct the state vector for absorption decision: ; Calculate the absorption probability using a pre-trained Q-learning policy network: The network is trained offline using a large number of building point cloud samples, and the reward function is designed as follows: ; The final decision is made by combining rule thresholds, and its absorption condition is: ; Step 37: Perform batch optimization on the candidate point set that has passed the initial screening: Construct a candidate point conflict graph: If the absorption spheres of two candidate points (with radius ) If they intersect, then it is considered a conflict.

[0097] To solve the maximum weighted independent set problem, select the non-conflicting subset that maximizes the total absorbed revenue: ; Among them, weight For all conflict subsets, and The sum is not greater than 1. Perform batch absorption and update the growth region. And regional attributes.

[0098] The directional growth guided by entropy gradient field and adaptive parameter adjustment based on historical feedback includes the following steps: Step 38, calculate the current growth region The direction of the average entropy gradient at the boundary points: ; Establish a growth direction preference model, including: Main growth direction: along The direction of fastest entropy decrease; Secondary growth direction: extending along the plane of the current region; Prohibited direction: along , the direction of increasing entropy.

[0099] Assign growth direction weights to each boundary point: ; in, This is the preset preference direction.

[0100] Step 39: Monitor convergence metrics during the growth process, including: Entropy change trend: the average rate of change of entropy in the region over the most recent k iterations. ; Flatness preservation: the consistency of the normal vectors of the point cloud in the region. ; in, The eigenvalues ​​are the covariance matrix features of the regional point cloud.

[0101] but: like (Entropy increases), tightening the convergence threshold: ; like Increase the weight of the normal vector constraint; If the growth rate is too slow (more than 50 iterations), expand the search radius. .

[0102] A multi-condition joint convergence criterion is constructed to intelligently terminate growth, and the comprehensive quality index of the final growth region is calculated, constructing a topological relationship graph between regions. This includes the following steps:

[0103] Step 40: Establish a convergence criterion system. Growth is considered convergent when one of the following conditions is met: ①Convergence condition of structural entropy: The relative change over the past 10 iterations has been less than 1%. ② Boundary stability condition: Calculate the Hösdorf distance change of the boundary point set: ; Take twice the average point spacing.

[0104] ③ Candidate point depletion conditions: The ratio of candidate buffers to active boundary points is less than 0.1.

[0105] ④ Maximum Iteration Protection: In this embodiment .

[0106] Step 41, calculate the overall quality index of the final growth region: ① Structural consistency score : ,in The standard deviation of the point entropy values ​​within the region; ② Geometric integrity score : Projected area Area of ​​the minimum bounding rectangle The ratio measures the compactness of the area; ③Boundary sharpness score : .

[0107] The growth results are classified based on the scores above: High-quality area: All scores > 0.7, output directly; Area requiring optimization: A certain score is between 0.5 and 0.7, and will be optimized subsequently; Problem area: If a score is less than 0.5, it is marked as a failure.

[0108] Step 42, each successfully generated growth region is

[0109] The method based on minimum description length metric performs optimal fusion of the building point cloud aggregation region, converts the fused 3D point cloud building object into a 2D vector contour, and performs quality evaluation on the 2D vector contour to obtain a structured building representation, including: The optimal fusion of the building point cloud aggregation region is performed based on the minimum description length metric method between regions. The total description length is defined as the weighted sum of the geometric description length and the structural entropy description length. The fusion benefit and fusion confidence of candidate fusion pairs are calculated, and the fusion of regions that meet the conditions is performed to obtain the final set of building objects. Based on adaptive α-shape contour extraction, the final set of building objects is subjected to contour polygon extraction. The interior angle size is calculated by detecting the contour angle, and the interior angle that is close to 90 degrees is adjusted to 90 degrees to obtain the two-dimensional vector contour. The geometric fidelity and area consistency are calculated to perform the quality assessment and generate the structured building representation.

[0110] The method based on minimum description length between regions performs optimal fusion of the building point cloud aggregation regions. The total description length is defined as the weighted sum of the geometric description length and the structural entropy description length. The fusion benefit and fusion confidence of candidate fusion pairs are calculated, and region fusion that meets the conditions is performed to obtain the final set of building objects, including: The geometric description length includes boundary complexity and internal structural complexity. The structural entropy description length is calculated based on the covariance of the regional point entropy value distribution and the KL divergence between the regional distribution and the global building point distribution. Based on spatial adjacency, regions with effective adjacency distances less than a preset value are retained. Regions with merging benefits less than a preset value and merging confidence greater than a preset value are merged.

[0111] In one specific embodiment, optimal fusion of the same building is performed on the growth regions based on the minimum description length metric between regions. This includes the following steps:

[0112] Step 43, for a growth region Define its total description length The weighted sum of the two components: ; in, For adjustable weights, Geometric description length : , [ = [0.6, 0.4]. Boundary complexity ,in, Let the perimeter of the projected boundary be . The circumference of a circle with equal area This is the ratio of the convex hull area to the actual area, used to measure the concavity / convexity of the boundary. Internal structural complexity. ,in, Local normal entropy, used to measure changes in surface orientation. Used to measure elevation fluctuations.

[0113] Structural entropy describes length The first term is the covariance based on the distribution of regional point entropy values. The information entropy formula follows a Gaussian distribution; the second term represents the regional distribution. Global building point distribution The KL divergence penalizes areas that deviate from typical architectural features.

[0114] Step 44: Calculate all region pairs based on spatial adjacency relationships. Effective adjacency distance: ; in, This represents the average radius of the region's horizontal projection. (Retain) Region pairs. For each candidate merge pair Calculate the combined revenue:

[0115] ; Introducing a combined confidence assessment: ; The confidence assessment takes into account both the significance of the reduction in description length and the balance of regional size.

[0116] Step 45: Iteratively execute candidate merge pairs and select those that meet the requirements. and Find the best merge pair and perform the merge. When there are no more merge pairs that meet the criteria, or when the maximum number of merge rounds is reached (e.g., 10 rounds), output the final set of building objects.

[0117] In this embodiment, adaptive α-shape contour initial extraction is selected. The contour polygons of the 3D point cloud buildings obtained in steps 43-45 are extracted. The contour angles are detected, and the interior angles are calculated. Interior angles close to 90 degrees are directly adjusted to 90 degrees to ensure the accuracy of the building angles and obtain the final contour. .

[0118] For the final outline Perform quality assessment and topology consistency verification, and generate structured building representations, including the following steps: Step 46: Final outline of each building object Calculate their geometric fidelity respectively. and area consistency .

[0119] ,in, For point To the final outline The shortest distance, The average spacing of the point cloud. A value close to 1 indicates that the outline can cover the point cloud well.

[0120] ,in, Let be the area of ​​the convex hull of the point cloud.

[0121] Step 47: Output the final structured building representation for easy use later.

[0122] This invention also provides a building extraction system based on multidimensional point cloud structural entropy phase transition detection, the system comprising: The data preprocessing and structural entropy field construction module is used to denoise the original point cloud data and obtain non-ground points using a progressive triangulation encryption filtering algorithm. The structural entropy field is constructed by weighted fusion of geometric structure entropy, echo distribution entropy and elevation consistency entropy. The gradient field calculation and boundary identification module is used to calculate the gradient field of the non-ground points based on the structural entropy field, identify the boundary structure based on gradient flow tracking and persistent cohomology, and construct candidate building regions based on closed-loop boundary rings. The multimodal feature clustering module is used to construct a multimodal feature field for the candidate building region, and perform feature clustering based on Gaussian mixture model and density peak to obtain a high-quality seed point set; The seed activation and region growth module is used to activate seeds based on the high-quality seed point set, make absorption decisions on candidate points based on the structural convergence dynamic evaluation method, and perform directional growth based on the structural entropy field to obtain the building point cloud aggregation region. The region fusion and contour extraction module is used to perform optimal fusion of the building point cloud aggregation region based on the minimum description length metric method, convert the fused 3D point cloud building objects into 2D vector contours, evaluate the quality of the 2D vector contours, and obtain a structured building representation.

[0123] Specifically, this embodiment of a building extraction system based on multidimensional point cloud structural entropy phase transition detection achieves a fully automated processing flow from raw point cloud data to structured building representation through the collaborative work of five functional modules. Combining structural entropy field construction, gradient field boundary recognition, multimodal feature clustering, seed activation region growth, and region fusion contour extraction, it realizes efficient and intelligent processing of large-scale point cloud data and high-precision automatic building extraction.

[0124] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A building extraction method based on multi-dimensional point cloud structure entropy phase transition detection, characterized by, The method comprises the following steps: The original point cloud data is denoised and a progressive triangulation encryption filtering algorithm is used to obtain non-ground points, a structure entropy field is constructed by weighted fusion of geometric structure entropy, echo distribution entropy and height consistency entropy; The structure entropy field quantifies the structure order of the local neighborhood of the point cloud by fusing the three dimensions of geometric anisotropy, echo distribution complexity and height fluctuation consistency; Gradient field calculation is performed on the non-ground points based on the structure entropy field, boundary structures are identified based on gradient flow tracking and persistent homology, and a candidate building area is constructed based on a closed-loop boundary ring; A multi-modal feature field is constructed for the candidate building area, feature clustering is performed based on a Gaussian mixture model and a density peak value, and a high-quality seed point set is obtained; Seed activation is performed based on the high-quality seed point set, absorption decision is made on candidate points based on a structure convergence dynamic evaluation method, directional growth is performed based on the structure entropy field, and a building point cloud aggregation area is obtained. The optimal fusion of the building point cloud aggregation area is performed based on a minimum description length metric method, the fused three-dimensional point cloud building object is converted into a two-dimensional vector contour, the quality of the two-dimensional vector contour is evaluated, and a structured building expression is obtained.

2. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 1, characterized in that, The gradient field calculation on the non-ground points based on the structure entropy field, the identification of boundary structures based on gradient flow tracking and persistent homology, and the construction of a candidate building area based on a closed-loop boundary ring comprise: Gradient field calculation and significance analysis of the three-dimensional structure entropy field of the non-ground points are performed, a scalar entropy value field of the point neighborhood is fitted by using a moving least squares method, local linear fitting coefficients are obtained by solving a weighted least squares problem, an entropy gradient vector and a gradient amplitude at each point are calculated, and a significant gradient point set that meets the gradient amplitude condition is screened out; The significant gradient point set is taken as a seed point, bidirectional streamline tracking is performed along the gradient direction, a boundary segment is obtained by using a fourth-order Runge-Kutta method to numerically integrate the streamline equation, significant boundary structures in the boundary segment are identified based on a topological feature importance evaluation scheme of persistent homology, and the candidate building area is constructed based on a closed-loop boundary ring.

3. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 2, characterized in that, The gradient field calculation and significance analysis of the three-dimensional structure entropy field of the non-ground points, the fitting of the scalar entropy value field of the point neighborhood by using the moving least squares method, the obtaining of the local linear fitting coefficients by solving the weighted least squares problem, the calculation of the entropy gradient vector and the gradient amplitude at each point, and the screening out of the significant gradient point set that meets the gradient amplitude condition comprise: The dot product of the gradient direction and the local normal vector is calculated to determine the positive and negative gradients, the positive gradient of the positive and negative gradients indicates that the building points point to the environment corresponding to the outer boundary of the building, the negative gradient of the positive and negative gradients indicates that the environment points to the building interior corresponding to the hole or courtyard boundary, the statistics of the global gradient amplitude distribution is calculated, and the significant gradient point set that meets the condition of the gradient amplitude being greater than the mean value plus the standard deviation multiple is screened out.

4. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 2, characterized in that, The significant gradient point set is taken as a seed point, bidirectional streamline tracking is performed along a gradient direction, a fourth-order Runge-Kutta method numerical integral streamline equation is used to obtain a boundary segment, a topological feature importance evaluation scheme based on persistent homology is used to identify a significant boundary structure in the boundary segment, and a candidate building region is constructed based on a closed-loop boundary ring, including: The positive gradient points are integrated along the positive gradient direction to simulate the entropy increasing path from the building interior to the exterior, the negative gradient points are integrated along the negative gradient direction to simulate the entropy decreasing path from the exterior to the building interior, an ordered point set is generated on each streamline, and spline fitting is performed to obtain a smooth parameterized curve representation, and a α The complex is calculated, the birth time and death time of the generators of the homology groups of each dimension are calculated, the topological features satisfying the persistence threshold are screened, and the corresponding boundary segments are extracted according to the streamline point set corresponding to the topological features.

5. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 1, characterized in that, The building point cloud aggregation region is optimally fused based on a minimum description length metric method, a three-dimensional point cloud building object after fusion is converted into a two-dimensional vector contour, and quality evaluation is performed on the two-dimensional vector contour to obtain a structured building expression, including: The building point cloud aggregation region is optimally fused based on an inter-region minimum description length metric method, a total description length is defined as a weighted sum of a geometric description length and a structure entropy description length, a merging benefit and a merging confidence of a candidate merging pair are calculated, a region merging that meets a condition is performed, and a final building object set is obtained; A contour polygon is extracted from the final building object set based on an adaptive alpha-shape contour extraction, an inner angle size is calculated by detecting a contour angle, an inner angle close to 90 degrees is adjusted to 90 degrees, the two-dimensional vector contour is obtained, and the quality evaluation is performed by calculating a geometric fidelity and an area consistency to generate the structured building expression.

6. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 5, characterized in that, The building point cloud aggregation region is optimally fused based on an inter-region minimum description length metric method, a total description length is defined as a weighted sum of a geometric description length and a structure entropy description length, a merging benefit and a merging confidence of a candidate merging pair are calculated, a region merging that meets a condition is performed, and a final building object set is obtained, including: The geometric description length includes a boundary complexity and an internal structure complexity, the structure entropy description length is calculated based on a region point entropy value distribution covariance and a KL divergence of a region distribution and a global building point distribution, a region pair with an effective adjacent distance less than a preset value is retained based on a spatial adjacency relationship, and a region merging is performed on a region with a merging benefit less than a preset value and a merging confidence greater than a preset value.

7. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 1, characterized in that, The original point cloud data is denoised, and a progressive triangular net encryption filtering algorithm is used to obtain non-ground points, a structure entropy field is constructed by weighted fusion of a geometric structure entropy, a echo distribution entropy, and a height consistency entropy, including: A statistical outlier removal method is used on the original point cloud data to eliminate sensor noise and abnormal points, a voxel grid method is used to regularize and resample the point cloud to obtain resampled point cloud; A progressive triangular net encryption filtering algorithm is used to separate the resampled point cloud to obtain the non-ground points; A multi-scale adaptive neighborhood is constructed according to the characteristic scale of the building component, a final representative entropy value of each point is generated by weighted geometric mean fusion of the geometric structure entropy, the echo distribution entropy, and the height consistency entropy, and the structure entropy field is constructed based on a local terrain complexity to dynamically determine the credibility weight of the entropy value of each scale.

8. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 1, characterized in that, The multi-modal feature field is constructed for the candidate building region, feature clustering is performed based on a Gaussian mixture model and a density peak value, and a high-quality seed point set is obtained, including: The local point density, normalized relative height and final structure entropy are calculated for each point in the candidate building area, and the robust statistical quantity calculation and winsorization processing are performed on each feature dimension to establish the mapping from the three-dimensional coordinate space to the feature space, and the feature point set distribution is obtained; The optimal component number is determined by the Bayesian information criterion to fit the Gaussian mixture model, the local density and minimum distance of each point in the feature space are calculated, the posterior probability of the Gaussian mixture model is fused with the density peak value information, and the similar clustering clusters are merged based on the Mahalanobis distance, and the high-quality seed point set is obtained.

9. The building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to claim 1, characterized in that, The seed activation is performed based on the high-quality seed point set, the candidate points are absorbed based on the dynamic evaluation method of structure convergence, and the directional growth is performed based on the structure entropy field, and the building point cloud aggregation area is obtained, including: A weighted undirected graph is constructed based on the high-quality seed point set, and the confidence propagation iteration is performed to obtain an activation sequence, an initial growth area is created from the seed point with the highest confidence, the regional structure entropy reference value and the dominant direction of the regional normal vector are calculated, and the active boundary set, the candidate point buffer area, the growth history record and the dynamic convergence threshold are initialized; Each unmarked candidate point is searched from each point in the active boundary set, the local structure compatibility, the regional entropy change prediction and the topological continuity index are calculated for each unmarked candidate point, the absorption decision state vector is constructed, the absorption decision is made in combination with the pre-trained strategy network and the rule threshold, and the batch absorption updates the growth area.

10. A building extraction system based on multi-dimensional point cloud structure entropy phase transition detection, configured to perform a building extraction method based on multi-dimensional point cloud structure entropy phase transition detection according to any one of claims 1-9, characterized in that, The system comprises: A data preprocessing and structure entropy field construction module is configured to denoise the original point cloud data and obtain non-ground points by using a progressive triangular mesh encryption filtering algorithm, and to construct a structure entropy field by weighted fusion of geometric structure entropy, echo distribution entropy and elevation consistency entropy; A gradient field calculation and boundary identification module is configured to calculate a gradient field for the non-ground points based on the structure entropy field, identify boundary structures based on gradient flow tracking and persistent homology, and construct a candidate building area based on a closed-loop boundary ring; A multi-modal feature clustering module is configured to construct a multi-modal feature field for the candidate building area, perform feature clustering based on a Gaussian mixture model and a density peak value, and obtain a high-quality seed point set; A seed activation and region growing module is configured to perform seed activation based on the high-quality seed point set, make absorption decisions for candidate points based on a dynamic evaluation method of structure convergence, and perform directional growth based on the structure entropy field, and obtain a building point cloud aggregation area. A region fusion and contour extraction module is configured to perform optimal fusion on the building point cloud aggregation area based on a minimum description length metric method, convert the fused three-dimensional point cloud building object into a two-dimensional vector contour, perform quality evaluation on the two-dimensional vector contour, and obtain a structured building representation.

Images