High-precision real estate surveying and mapping method in complex terrain environment based on optimization algorithm

Abstract

The invention discloses a high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm, and the method comprises the following steps: collecting abundant geographic information data through integrating a laser radar and a global navigation satellite system sensor, carrying out the preprocessing of the data, generating basic geographic information data, and constructing a digital elevation model; the method is used for representing three-dimensional fluctuating changes of the terrain, performing path planning and dynamic adjustment of an equipment running track through an optimization algorithm to adapt to the complex terrain, integrating multi-source data into a unified geographic information model by using a fusion algorithm, and performing error analysis and correction to obtain an optimized geographic information model. A real estate surveying and mapping map is generated based on the model, and real estate boundaries and topographic features are accurately marked. Through a series of optimization technologies, the target of high-precision surveying and mapping in complex terrains is achieved.

Description

Technical Field

[0001] The present invention relates to the field of real estate surveying and mapping technology, and in particular to a high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm. Background Art

[0002] In modern real estate surveying and mapping, especially in complex terrain environments, traditional surveying and mapping methods often face many challenges. With the development of geographic information systems, remote sensing technology, drone technology, and data processing technology, the accuracy and efficiency of surveying and mapping have improved, but the existing technology still has significant deficiencies in meeting the needs of high-precision surveying and mapping in complex terrain environments.

[0003] Traditional surveying and mapping methods mainly rely on a single data source, such as the Global Navigation Satellite System (GNSS) or LiDAR, but the use of a single data source is easily limited by environmental conditions, equipment accuracy, and data processing algorithms. In complex terrains such as mountains and canyons, GNSS signals may be blocked or reflected, resulting in reduced positioning accuracy; and although LiDAR can provide high-precision three-dimensional point cloud data, the integrity and accuracy of the data are still difficult to guarantee in areas with dense vegetation or large terrain undulations. Due to the lack of effective fusion of multi-source data, traditional methods often cannot achieve sufficient coverage and accuracy in complex terrain.

[0004] Existing path planning algorithms, such as those based on the traditional Dijkstra algorithm or A* algorithm, can solve the optimization problem of surveying and mapping paths to a certain extent, but when faced with complex terrain, they are often unable to dynamically adapt to terrain changes, resulting in low surveying and mapping efficiency, and may even miss key terrain features. These algorithms usually only consider the length of the path and simple terrain features, lack a comprehensive understanding of the terrain and the ability to optimize multi-modal paths, and therefore often show inadequacies in practical applications.

[0005] Traditional multi-source data fusion technologies are mostly based on linear interpolation or simple weighted average methods. These methods often ignore the nonlinear characteristics and spatial relationships of data when processing data from different sensors, resulting in insufficient accuracy and consistency of fusion results. In addition, existing error correction methods, such as least squares and Kalman filtering, can effectively reduce errors in some simple scenarios, but in complex terrain environments, especially when multi-source data fusion and multi-modal path optimization are involved, they cannot fully eliminate system errors, affecting the accuracy of the final mapping results.

[0006] Therefore, how to provide a high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm is an urgent problem that technicians in this field need to solve. Summary of the invention

[0007] One purpose of the present invention is to propose a high-precision real estate mapping method under complex terrain environment based on optimization algorithm. The present invention describes in detail the whole process of realizing high-precision mapping under complex terrain conditions by integrating multi-source data acquisition technology, adopting adaptive multi-modal evolutionary optimization algorithm and high-dimensional nonlinear dynamic optimization fusion algorithm. It has the advantages of extensive data acquisition, accurate path planning, high consistency of data fusion and high error correction accuracy.

[0008] According to an embodiment of the present invention, a high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm includes the following steps:

[0009] S1. In complex terrain environments, multi-source data acquisition technology is used to integrate LiDAR and GNSS sensors to collect geographic information data to achieve data diversity and wide coverage;

[0010] S2. Preprocess the collected multi-source data, including data cleaning and coordinate conversion, to generate basic geographic information data;

[0011] S3, based on the processed geographic information data, construct a digital elevation model to represent the three-dimensional undulations of complex terrain;

[0012] S4. Based on the generated three-dimensional terrain model, an adaptive multi-modal evolutionary optimization algorithm is used for surveying and mapping path planning. By identifying and optimizing multiple path modes, the optimization strategy is dynamically adjusted to select the optimal path in complex terrain;

[0013] S5. During the surveying and mapping process, according to the path planning results, the running trajectory of the surveying and mapping equipment is dynamically adjusted to adapt to terrain changes and perform surveying and mapping;

[0014] S6, fusing the processed multi-source data with the generated three-dimensional terrain model, using a high-dimensional nonlinear dynamic optimization fusion algorithm to integrate the data of different sensors into a unified geographic information model, and performing data fusion and optimization processing;

[0015] S7, performing error analysis and correction on the generated geographic information model, applying a deep residual regression algorithm to correct the error, correcting the system error through a layer-by-layer progressive regression method, and finally obtaining an optimized geographic information model;

[0016] S8. Generate real estate surveying and mapping maps based on the optimized geographic information model, marking real estate boundaries and terrain features.

[0017] Optionally, the S1 specifically includes:

[0018] S11. In complex terrain environments, multi-source data acquisition technology is used to integrate lidar sensors and global navigation satellite system sensors to collect three-dimensional point cloud data and geographic location information respectively;

[0019] S12, emit a laser beam through the LiDAR sensor and receive the echo signal, and calculate the distance based on the echo time t and the speed of light c Generate 3D point cloud data of the target area, including the 3D coordinates (x i ,y i ,z i );

[0020] S13, use the GNSS sensor to receive satellite signals, calculate and record the geographic coordinates of the surveying points (x i ,y i ,z i ), providing geographic location information;

[0021] S14, using an adaptive data synchronization algorithm based on topology preservation, synchronizing the three-dimensional point cloud data collected by the LiDAR sensor with the geographic coordinate data acquired by the GNSS sensor, and automatically adjusting the time and space synchronization of the data by maintaining the topological structure of the terrain data;

[0022] S15. Integrate the synchronized data with topology preservation to form an initial geographic information dataset with wide coverage and diversity, which includes three-dimensional terrain relief, coordinates and topological structure.

[0023] Optionally, the S3 specifically includes:

[0024] S31, based on the formed initial geographic information data set, extract the three-dimensional point cloud data (x i ,y i ,z i ) and corresponding geographic coordinate information;

[0025] S32, gridding the extracted three-dimensional point cloud data according to geographic coordinates to generate a preliminary elevation data model, and adjusting the size and distribution of the grid cells according to the complexity of the terrain;

[0026] S33, using the digital elevation model generation algorithm, based on the gridded three-dimensional point cloud data (x i ,y i ,z i ), calculate the elevation value Z in each grid cell DEM ;

[0027] S34, introduce the sparse matrix interpolation algorithm, use the calculated elevation value Z DEM Construct a sparse matrix M:

[0028]

[0029] In the interpolation process, the sparse matrix M is processed by an optimization algorithm:

[0030]

[0031] Where W is the interpolation weight matrix, λ is the regularization coefficient, and R(M) is the regularization term of the sparse matrix M;

[0032] S35, the elevation value Z′ processed by the sparse matrix interpolation algorithm DEM With geographic coordinate information (x i ,y i ) to build a complete three-dimensional digital elevation model.

[0033] Optionally, the S5 specifically includes:

[0034] S41. Extract terrain feature data based on the constructed 3D digital elevation model, including elevation value Z′ DEM , terrain slope θ and terrain curvature κ;

[0035] S42, input the extracted terrain feature data into the adaptive multi-modal evolutionary optimization algorithm to preliminarily generate multiple candidate paths in the complex terrain, each candidate path consists of a series of path points P i The data of each path point includes the elevation Z′ DEM (P i ), slope θ(P i ), curvature κ(P i ), and other possible topographic features;

[0036] S43. In the process of path optimization, a path evaluation function f(P) is introduced, which not only considers the traditional terrain characteristics, but also includes the evolutionary stability of the path and the adaptability under complex terrain:

[0037] f(P)=∫ P [α·Z' DEM (x,y)+β·θ(x,y)+γ·κ(x,y)+δ·S(x,y)+∈·E(P)]ds;

[0038] Among them, α, β, and γ are weight factors in path optimization, which control the influence of elevation, slope, and curvature on path selection respectively. S(x, y) represents the evolution stability at the path point (x, y). E(P) represents the global evolution complexity of the path. ds is the path element in the path integral.

[0039] S44. Based on the result of the path evaluation function f(P), identify and optimize the optimal path mode, and dynamically adjust the key parameters α, β, γ, δ, ∈ of the path to adapt to different terrain characteristics through an adaptive multi-modal evolution strategy:

[0040]

[0041] S45, the optimal path P obtained by optimization opt Projected into a 3D digital elevation model, 3D path data is generated for navigation of surveying and mapping equipment.

[0042] Optionally, the S6 specifically includes:

[0043] S61, based on the processed multi-source data and the generated three-dimensional path data, extracting geographic information data from different sensors, including three-dimensional point cloud data D LiDAR , geographic coordinate data D GNSS and elevation value Z′ DEM ;

[0044] S62, the extracted data and the three-dimensional terrain model M DEM To perform fusion, a high-dimensional nonlinear dynamic optimization fusion algorithm is applied to map the feature spaces of different data sources into a unified high-dimensional feature space by constructing a nonlinear mapping function F in the high-dimensional space:

[0045] M fused =F(D aligned ,M DEM );

[0046] Among them, F represents the nonlinear mapping function, M fused Represents the fused geographic information model;

[0047] S63. In the process of data fusion, a dynamic optimization mechanism is used to adjust the parameters of the nonlinear mapping function F in real time, and the fusion result maintains global consistency and adaptability to local terrain features:

[0048] min F {∑ i,j (Z' fused (x i ,y j )-Z target (x i ,y j )) 2 +λR(F)};

[0049] Among them, Z′ fused is the fused elevation value, Z target is the target elevation value, λ is the regularization coefficient, and R(F) is the regularization term of the mapping function F;

[0050] S64, the geographic information model M that has undergone high-dimensional nonlinear dynamic optimization fusion processing fused Combined with geographic coordinate information (x i ,yi ) and elevation value Z′ fused , generate a comprehensive geographic information model M composite .

[0051] Optionally, the S7 specifically includes:

[0052] S71. Generative integrated geographic information model M composite , extract the geographic coordinates (x i ,y i ), elevation value Z′ fused And the corresponding target elevation value Z target ;

[0053] S72, input the extracted data into the deep residual regression model, and establish an initial prediction model f res (x i ,y i )Calculate each coordinate point (x i ,y i )’s initial residual value r i :

[0054] r i =Z target (x i ,y i )-Z′ fused (x i ,y i );

[0055] S73. In the deep residual regression model, the system error is corrected by using a layer-by-layer progressive regression method, and the residual value r is fitted layer by layer through a multi-layer neural network. i , the goal of each layer of regression model is to minimize the residual value of the current layer Updated elevation value Z″ fused It is expressed as:

[0056]

[0057] in, represents the residual regression function of the lth layer. The layer-by-layer progressive regression algorithm gradually reduces the system error through multiple iterations. L is the number of regression layers.

[0058] S74. In the layer-by-layer regression process, the learning rate and regression coefficient parameters of the model are dynamically adjusted to adapt to the complexity and error characteristics of different terrain areas, including adaptively adjusting the number of neural network layers and the number of neurons in each layer, so that the regression of each layer can effectively capture the error characteristics of different scales, and finally gradually reduce the overall system error:

[0059] M opt =M composite+ΔM;

[0060] Among them, M opt is the final optimized model, ΔM represents the error correction matrix obtained by the deep residual regression algorithm, which contains the error correction matrix for all coordinate points (x i ,y i )

[0061] S75. After obtaining the error correction matrix ΔM, apply it to the comprehensive geographic information model M composite In the process, the final optimized geographic information model M is generated. opt This optimization model not only corrects the initial system error, but also significantly improves the overall accuracy of the model through layer-by-layer regression. The final high-precision geographic information model M opt It will be used in subsequent surveying, analysis and other related application scenarios to provide an accurate geographic information foundation for practical operations.

[0062] Optionally, the S8 specifically includes:

[0063] S81, based on the generated final optimized geographic information model M opt , extract the geographic coordinates (x i ,y i ), the final optimized elevation value Z″ fused , slope θ(x i ,y i ) and the curvature κ(x i ,y i )’s terrain feature data;

[0064] S82. Based on the extracted geographic information data, use the real estate mapping map generation algorithm to preliminarily identify the real estate boundaries in the terrain model, and analyze the elevation value Z″ fused , slope θ(x i ,y i ) and the curvature κ(x i ,y i ) changes, determine the initial position of the real estate boundary B init :

[0065] B init ={(x i ,y i )|f(B(x i ,y i ))=0};

[0066] Among them, f(B(x i ,y i )) is the boundary judgment function;

[0067] S83, introduce the topology sensitive boundary optimization algorithm, and optimize the initial real estate boundary B init Further fine-tuning is performed to dynamically adjust the position of the boundary line based on the extracted terrain feature data:

[0068] B opt =argmin B {∑ i,j [f(B(x i ,y j ))·w(x i ,y j )+λ·T(x i ,y j )]};

[0069] Among them, B opt is the optimized real estate boundary line, w(x i ,y j ) is a weight factor used to adjust the local position of the boundary line, λ is a smoothing parameter, T(x i ,y j ) is a topological sensitivity function, which is used to reflect the gradient change of terrain features to optimize the smoothness and adaptability of the boundary line in the area of ​​terrain mutation;

[0070] S84, optimize the real estate boundary line B opt Projection to 3D digital elevation model M opt In combination with the optimized terrain feature data θ(x i ,y i ) and κ(x i ,y i ) to generate a real estate mapping map containing real estate boundaries and terrain features;

[0071] S85. In the generated surveying and mapping map, the real estate boundary B opt The main terrain features are annotated in detail, and the annotated locations and attribute values ​​are expressed in geographic coordinates (x i ,y i ) and the corresponding elevation value Z″ fused Determine and optimize the final geographic information model M opt Make associations to ensure that every data item in the surveying and mapping map is derived from an optimized geographic information model;

[0072] S86. Save the generated real estate surveying and mapping map as a digital file in a specified format, which contains a complete geographic information model and detailed terrain feature annotations.

[0073] The beneficial effects of the present invention are:

[0074] The present invention realizes efficient synchronization of LiDAR and GNSS sensor data during multi-source data collection. The algorithm ensures the consistency and accuracy of data from different sensors in complex terrain environments by maintaining the topological structure of terrain data and automatically adjusting the time and space synchronization of data, thus overcoming the difficulty of multi-source data integration in traditional methods. An adaptive multi-modal evolutionary optimization algorithm is used for path planning. The algorithm can identify and optimize multiple path modes in complex terrain, dynamically adjust the path planning strategy, and thus select the optimal path for surveying and mapping. Compared with traditional path planning methods, the adaptive multi-modal evolutionary optimization algorithm can better adapt to terrain changes, significantly improve the efficiency of surveying and mapping and the accuracy of path planning, especially in environments with complex terrain and drastic changes.

[0075] The present invention introduces a sparse matrix interpolation algorithm for constructing a digital elevation model. The algorithm performs sparse matrix interpolation processing on three-dimensional point cloud data, which not only reduces the amount of data, but also significantly improves the detail accuracy of the terrain model. Compared with the traditional linear interpolation method, the sparse matrix interpolation algorithm can more accurately capture subtle changes in terrain, making the performance of the digital elevation model in complex terrain more refined and accurate. A high-dimensional nonlinear dynamic optimization fusion algorithm is used. The algorithm fuses data features from different sensors by constructing a nonlinear mapping function in a high-dimensional space, and uses a dynamic optimization mechanism to adjust the parameters of the mapping function in real time to ensure that the fusion result can adapt to local terrain features while being globally consistent. This innovation effectively solves the problem of insufficient nonlinear data processing in traditional fusion algorithms and greatly improves the accuracy of the fusion model.

[0076] The present invention applies a deep residual regression algorithm for error correction. Through a layer-by-layer progressive regression method, the present invention can gradually reduce the systematic error and eventually generate an optimized high-precision geographic information model. Compared with traditional error correction methods, the deep residual regression algorithm can more effectively capture the systematic errors in complex terrain, and accurately correct them through a deep learning model, greatly improving the accuracy of the final surveying and mapping results. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0078] Figure 1 A flow chart of the high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm proposed by the present invention;

[0079] Figure 2 A schematic diagram of the adaptive multi-modal evolutionary optimization algorithm used for path planning in the present invention;

[0080] Figure 3 It is a joint schematic diagram of the high-dimensional nonlinear dynamic optimization fusion algorithm and the deep residual regression algorithm in the present invention. DETAILED DESCRIPTION

[0081] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0082] refer to Figure 1-3 , a high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm, comprising the following steps:

[0083] S1. In complex terrain environments, multi-source data acquisition technology is used to integrate LiDAR and GNSS sensors to collect geographic information data to achieve data diversity and wide coverage;

[0084] S2. Preprocess the collected multi-source data, including data cleaning and coordinate conversion, to generate basic geographic information data;

[0085] S3, based on the processed geographic information data, construct a digital elevation model to represent the three-dimensional undulations of complex terrain;

[0086] S4. Based on the generated three-dimensional terrain model, an adaptive multi-modal evolutionary optimization algorithm is used for surveying and mapping path planning. By identifying and optimizing multiple path modes, the optimization strategy is dynamically adjusted to select the optimal path in complex terrain;

[0087] S5. During the surveying and mapping process, according to the path planning results, the running trajectory of the surveying and mapping equipment is dynamically adjusted to adapt to terrain changes and perform surveying and mapping;

[0088] S6, fusing the processed multi-source data with the generated three-dimensional terrain model, using a high-dimensional nonlinear dynamic optimization fusion algorithm to integrate the data of different sensors into a unified geographic information model, and performing data fusion and optimization processing;

[0089] S7, performing error analysis and correction on the generated geographic information model, applying a deep residual regression algorithm to correct the error, correcting the system error through a layer-by-layer progressive regression method, and finally obtaining an optimized geographic information model;

[0090] S8. Generate real estate surveying and mapping maps based on the optimized geographic information model, marking real estate boundaries and terrain features.

[0091] In this implementation, S1 specifically includes:

[0092] S11. In complex terrain environments, multi-source data acquisition technology is used to integrate lidar sensors and global navigation satellite system sensors to collect three-dimensional point cloud data and geographic location information respectively;

[0093] S12, emit a laser beam through the LiDAR sensor and receive the echo signal, and calculate the distance based on the echo time t and the speed of light c Generate 3D point cloud data of the target area, including the 3D coordinates (x i ,y i ,z i );

[0094] S13, use the GNSS sensor to receive satellite signals, calculate and record the geographic coordinates of the surveying points (x i ,y i ,z i ), providing geographic location information;

[0095] S14, using an adaptive data synchronization algorithm based on topology preservation, synchronizing the three-dimensional point cloud data collected by the LiDAR sensor with the geographic coordinate data acquired by the GNSS sensor, and automatically adjusting the time and space synchronization of the data by maintaining the topological structure of the terrain data;

[0096] S15. Integrate the synchronized data with topology preservation to form an initial geographic information dataset with wide coverage and diversity, which includes three-dimensional terrain relief, coordinates and topological structure.

[0097] In this implementation, S3 specifically includes:

[0098] S31, based on the formed initial geographic information data set, extract the three-dimensional point cloud data (x i ,y i ,z i ) and corresponding geographic coordinate information;

[0099] S32, gridding the extracted three-dimensional point cloud data according to geographic coordinates to generate a preliminary elevation data model, and adjusting the size and distribution of the grid cells according to the complexity of the terrain;

[0100] S33, using the digital elevation model generation algorithm, based on the gridded three-dimensional point cloud data (x i ,y i ,z i ), calculate the elevation value Z in each grid cell DEM ;

[0101] S34, introduce the sparse matrix interpolation algorithm, use the calculated elevation value Z DEM Construct a sparse matrix M:

[0102]

[0103] In the interpolation process, the sparse matrix M is processed by an optimization algorithm:

[0104]

[0105] Where W is the interpolation weight matrix, λ is the regularization coefficient, and R(M) is the regularization term of the sparse matrix M;

[0106] S35, the elevation value Z′ processed by the sparse matrix interpolation algorithm DEM With geographic coordinate information (x i ,y i ) to build a complete three-dimensional digital elevation model.

[0107] In this implementation manner, S5 specifically includes:

[0108] S41. Extract terrain feature data based on the constructed 3D digital elevation model, including elevation value Z′ DEM , terrain slope θ and terrain curvature κ;

[0109] S42, input the extracted terrain feature data into the adaptive multi-modal evolutionary optimization algorithm to preliminarily generate multiple candidate paths in the complex terrain, each candidate path consists of a series of path points P i The data of each path point includes the elevation Z′ DEM (P i ), slope θ(P i ), curvature κ(P i ), and other possible topographic features;

[0110] S43. In the process of path optimization, a path evaluation function f(P) is introduced, which not only considers the traditional terrain characteristics, but also includes the evolutionary stability of the path and the adaptability under complex terrain:

[0111] f(P)=∫ P [α·Z' DEM (x,y)+β·θ(x,y)+γ·κ(x,y)+δ·S(x,y)·+∈·E(P)]ds;

[0112] Among them, α, β, and γ are weight factors in path optimization, which control the influence of elevation, slope, and curvature on path selection respectively. S(x, y) represents the evolution stability at the path point (x, y). E(P) represents the global evolution complexity of the path. ds is the path element in the path integral.

[0113] S44. Based on the result of the path evaluation function f(P), identify and optimize the optimal path mode, and dynamically adjust the key parameters α, β, γ, δ, ∈ of the path to adapt to different terrain characteristics through an adaptive multi-modal evolution strategy:

[0114]

[0115] S45, the optimal path P obtained by optimization opt Projected into a 3D digital elevation model, 3D path data is generated for navigation of surveying and mapping equipment.

[0116] In this implementation manner, S6 specifically includes:

[0117] S61, based on the processed multi-source data and the generated three-dimensional path data, extracting geographic information data from different sensors, including three-dimensional point cloud data D LiDAR , geographic coordinate data D GNSS and elevation value Z′ DEM ;

[0118] S62, the extracted data and the three-dimensional terrain model M DEM To perform fusion, a high-dimensional nonlinear dynamic optimization fusion algorithm is applied to map the feature spaces of different data sources into a unified high-dimensional feature space by constructing a nonlinear mapping function F in the high-dimensional space:

[0119] M fused =F(D aligned ,M DEM );

[0120] Among them, F represents the nonlinear mapping function, M fused Represents the fused geographic information model;

[0121] S63. In the process of data fusion, a dynamic optimization mechanism is used to adjust the parameters of the nonlinear mapping function F in real time, and the fusion result maintains global consistency and adaptability to local terrain features:

[0122] min F {∑ i,j (Z' fused (x i ,y j )-Z target (x i ,y j )) 2 +λR(F)};

[0123] Among them, Z′ fused is the fused elevation value, Z target is the target elevation value, λ is the regularization coefficient, and R(F) is the regularization term of the mapping function F;

[0124] S64, the geographic information model M that has undergone high-dimensional nonlinear dynamic optimization fusion processing fused Combined with geographic coordinate information (x i ,y i ) and elevation value Z′ fused , generate a comprehensive geographic information model M composite .

[0125] In this implementation manner, the S7 specifically includes:

[0126] S71. Generative integrated geographic information model M composite , extract the geographic coordinates (x i ,y i ), elevation value Z′ fused And the corresponding target elevation value Z target ;

[0127] S72, input the extracted data into the deep residual regression model, and establish an initial prediction model f res (x i ,y i )Calculate each coordinate point (x i ,y i )’s initial residual value r i :

[0128] r i =Z target (x i ,y i )-Z′ fused (x i ,y i );

[0129] S73. In the deep residual regression model, the system error is corrected by using a layer-by-layer progressive regression method, and the residual value r is fitted layer by layer through a multi-layer neural network. i , the goal of each layer of regression model is to minimize the residual value of the current layer , updated elevation value Z″ fused It is expressed as:

[0130]

[0131] in, represents the residual regression function of the lth layer. The layer-by-layer progressive regression algorithm gradually reduces the system error through multiple iterations. L is the number of regression layers.

[0132] S74. In the layer-by-layer regression process, the learning rate and regression coefficient parameters of the model are dynamically adjusted to adapt to the complexity and error characteristics of different terrain areas, including adaptively adjusting the number of neural network layers and the number of neurons in each layer, so that the regression of each layer can effectively capture the error characteristics of different scales, and finally gradually reduce the overall system error:

[0133] M opt =M composite +ΔM;

[0134] Among them, M opt is the final optimized model, ΔM represents the error correction matrix obtained by the deep residual regression algorithm, which contains the error correction matrix for all coordinate points (x i ,y i )

[0135] S75. After obtaining the error correction matrix ΔM, apply it to the comprehensive geographic information model M composite In the process, the final optimized geographic information model M is generated. opt This optimization model not only corrects the initial system error, but also significantly improves the overall accuracy of the model through layer-by-layer regression. The final high-precision geographic information model M opt It will be used in subsequent surveying, analysis and other related application scenarios to provide an accurate geographic information foundation for practical operations.

[0136] In this implementation manner, S8 specifically includes:

[0137] S81, based on the generated final optimized geographic information model M opt , extract the geographic coordinates (x i ,y i ), the final optimized elevation value Z″ fused , slope θ(x i ,y i ) and the curvature κ(x i ,y i )’s terrain feature data;

[0138] S82. Based on the extracted geographic information data, use the real estate mapping map generation algorithm to preliminarily identify the real estate boundaries in the terrain model, and analyze the elevation value Z″ fused , slope θ(x i ,y i ) and the curvature κ(x i ,y i ) changes, determine the initial position of the real estate boundary B init :

[0139] B init ={(x i ,yi )|f(B(x i ,y i ))=0};

[0140] Among them, f(B(x i ,y i )) is the boundary judgment function;

[0141] S83, introduce the topology sensitive boundary optimization algorithm, and optimize the initial real estate boundary B init Further fine-tuning is performed to dynamically adjust the position of the boundary line based on the extracted terrain feature data:

[0142] B opt =argmin B {∑ i,j [f(B(x i ,y j ))·w(x i ,y j )+λ·T(x i ,y j )]};

[0143] Among them, B opt is the optimized real estate boundary line, w(x i ,y j ) is a weight factor used to adjust the local position of the boundary line, λ is a smoothing parameter, T(x i ,y j ) is a topological sensitivity function, which is used to reflect the gradient change of terrain features to optimize the smoothness and adaptability of the boundary line in the area of ​​terrain mutation;

[0144] S84, optimize the real estate boundary line B opt Projection to 3D digital elevation model M opt In combination with the optimized terrain feature data θ(x i ,y i ) and κ(x i ,y i ) to generate a real estate mapping map containing real estate boundaries and terrain features;

[0145] S85. In the generated surveying and mapping map, the real estate boundary B opt The main terrain features are annotated in detail, and the annotated locations and attribute values ​​are expressed in geographic coordinates (x i ,y i ) and the corresponding elevation value Z″ fused Determine and optimize the final geographic information model M opt Make associations to ensure that every data item in the surveying and mapping map is derived from an optimized geographic information model;

[0146] S86. Save the generated real estate surveying and mapping map as a digital file in a specified format, which contains a complete geographic information model and detailed terrain feature annotations.

[0147] Embodiment 1:

[0148] In order to verify the feasibility of the present invention in implementation, the present invention is applied to a mountainous area in the south, where the terrain is complex, the mountains are undulating, the vegetation is dense, and there are a large number of gullies and canyons. This area requires high-precision real estate mapping to determine the boundaries of real estate and provide data support for subsequent land management and resource planning. However, due to the complexity of the terrain, traditional mapping methods are difficult to achieve high-precision and high-efficiency mapping work in this area. In particular, GNSS signals are easily blocked in mountainous areas, LiDAR data may be incomplete in vegetation-covered areas, and conventional path planning algorithms are also difficult to find the optimal path in such complex terrain.

[0149] During the data collection phase, the present invention uses multi-source data collection technology, integrating LiDAR and global navigation satellite system sensors. In early July 2024, the surveying and mapping team selected a typical area in the southern mountainous area, with an area of ​​about 100 square kilometers, and used the LiDAR sensor carried by the drone and the ground GNSS receiver to synchronously collect data. In this process, the LiDAR collects three-dimensional point cloud data in the area by emitting laser beams and receiving echo signals, while the GNSS receiver records the geographic coordinates of each surveying and mapping point by receiving satellite signals. Through an adaptive data synchronization algorithm based on topology preservation, the surveying and mapping team accurately synchronized the data collected by the LiDAR and GNSS in time and space, generating a complete and accurate initial geographic information data set.

[0150] In the data processing stage, the surveying and mapping team first cleaned and converted the collected data and constructed a digital elevation model of the area. Due to the complex terrain and large undulations in the mountainous area, the surveying and mapping team used a sparse matrix interpolation algorithm to interpolate the gridded three-dimensional point cloud data and constructed a refined elevation model. Compared with traditional methods, the model shows higher accuracy in areas with drastic changes in altitude. According to the actual test data of the surveying and mapping team, using the method of the present invention, the error between the interpolated elevation model and the measured data is reduced by an average of 15%, especially in areas with high slopes and uneven terrain, the error is reduced by nearly 20%.

[0151] The surveying and mapping team planned the surveying and mapping path. Since the paths in mountainous areas are complex and changeable, conventional path planning algorithms are difficult to adapt. Through the adaptive multimodal evolutionary optimization algorithm, the surveying and mapping team selected the path with the best performance in terrain adaptability and surveying and mapping efficiency from multiple candidate paths. During the path planning process, the algorithm takes into account multiple features of the terrain, such as elevation, slope, and curvature, and adjusts the optimization strategy in real time to adapt to complex terrain changes. In actual surveying and mapping, the optimization algorithm enables the surveying and mapping equipment to avoid multiple signal interference areas and obstacle areas while still covering the key terrain features of the surveying and mapping area. Compared with traditional path planning algorithms, the path planning time of the present invention is reduced by about 30%, and the path coverage rate is increased by about 25%.

[0152] During the data fusion and error correction stage, the surveying and mapping team fused the data collected by LiDAR and GNSS with the three-dimensional terrain model, and used a high-dimensional nonlinear dynamic optimization fusion algorithm to integrate the data of different sensors into a unified geographic information model. The algorithm ensures the balance between global consistency and local feature adaptability of the fusion results by dynamically adjusting the parameters of the mapping function. After data fusion, the surveying and mapping team used a deep residual regression algorithm to correct the errors of the fused geographic information model. The test results show that after multi-layer regression processing, the overall error of the model was reduced by more than 40%, and the final generated geographic information model has significantly improved the mapping accuracy in multiple key terrain areas.

[0153] Based on the optimized geographic information model, the surveying and mapping team generated a real estate surveying and mapping map of the area. Through the topology-sensitive boundary optimization algorithm, the real estate boundary was further accurately delineated to ensure the rationality and accuracy of the boundary line in complex terrain. Actual data shows that using the method of the present invention, the deviation between the generated boundary line and the actual ground measurement result is less than 10 cm, which is about 50% higher than the traditional method.

[0154] Throughout the testing process, the present invention demonstrated great superiority. Compared with traditional surveying and mapping methods, the present invention not only greatly improves the accuracy and efficiency of surveying and mapping, but also effectively solves common problems in complex terrain, such as signal interference, incomplete data, and difficult path planning. When the area of ​​the surveying and mapping area is the same, the surveying and mapping time is shortened by about 40%, and the overall data processing and image generation time is reduced by nearly half. These achievements fully demonstrate the feasibility and advantages of the present invention in complex terrain environments, and provide a new high-precision solution for the field of real estate surveying and mapping.

[0155] Table 1 Comparison of surveying accuracy and path planning effect

[0156]

[0157] Table 2 Comparison of data fusion, error correction and map generation effects

[0158]

[0159] By comparing the test results of the traditional method and the method of the present invention on multiple key performance indicators, the present invention demonstrates significant technical advantages, especially in the application performance in complex terrain environments. In terms of data acquisition, the multi-source data acquisition and synchronous processing technology of the present invention increases the coverage of data acquisition by 11.8%. This means that under complex terrain conditions, the present invention can collect geographic information data more comprehensively to ensure the integrity of the surveying and mapping area. At the same time, the data synchronization error is reduced by 40%, reflecting that the present invention effectively improves the consistency and synchronization accuracy of the data in the process of multi-source data fusion. In terms of digital elevation model construction, the present invention reduces the average error between the elevation model and the measured data by 16.7% by introducing a sparse matrix interpolation algorithm, especially in areas with large terrain undulations, the accuracy of the model has been significantly improved. This lays a more reliable foundation for subsequent surveying and mapping work.

[0160] In the process of path planning, the adaptive multi-modal evolutionary optimization algorithm adopted by the present invention greatly improves the efficiency and accuracy of path planning. Compared with the traditional method, the path planning time is shortened by 30%, and the path coverage rate is increased by 25.7%. This shows that the present invention can not only formulate the optimal mapping path faster, but also cover important areas more comprehensively in complex terrain. At the same time, the obstacle avoidance success rate of the mapping equipment has also increased by 41.7%, further enhancing the adaptability and safety of the equipment in complex terrain.

[0161] In terms of data fusion and error correction, the present invention reduces the error of the fused geographic information model by 40% and improves the model accuracy by 15% after system error correction through the application of high-dimensional nonlinear dynamic optimization fusion algorithm and deep residual regression algorithm. This ensures that the final generated real estate surveying and mapping map has higher accuracy. The boundary line deviation is reduced by 50%, and the accuracy of terrain feature annotation is improved by 11.8%, which further demonstrates the excellent performance of the present invention in surveying and mapping accuracy.

[0162] In summary, the present invention significantly improves the comprehensiveness of data collection, the accuracy of path planning, the consistency of data fusion, and the accuracy and efficiency of the final mapping results in real estate surveying and mapping in complex terrain environments. Compared with traditional methods, the present invention not only overcomes many limitations in the prior art, but also provides a more reliable and efficient solution for practical applications, with broad application prospects and significant technical advantages.

[0163] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A high-precision real estate surveying and mapping method in a complex terrain environment based on an optimization algorithm, characterized in that: The steps include: S1. In complex terrain environments, multi-source data acquisition technology is used to integrate LiDAR and GNSS sensors to collect geographic information data to achieve data diversity and wide coverage; S2. Preprocess the collected multi-source data, including data cleaning and coordinate conversion, to generate basic geographic information data; S3, based on the processed geographic information data, construct a digital elevation model to represent the three-dimensional undulations of complex terrain; S4. Based on the generated three-dimensional terrain model, an adaptive multi-modal evolutionary optimization algorithm is used for surveying and mapping path planning. By identifying and optimizing multiple path modes, the optimization strategy is dynamically adjusted to select the optimal path in complex terrain; S5. During the surveying and mapping process, according to the path planning results, the running trajectory of the surveying and mapping equipment is dynamically adjusted to adapt to terrain changes and perform surveying and mapping; S6, fusing the processed multi-source data with the generated three-dimensional terrain model, using a high-dimensional nonlinear dynamic optimization fusion algorithm to integrate the data of different sensors into a unified geographic information model, and performing data fusion and optimization processing; S7, performing error analysis and correction on the generated geographic information model, applying a deep residual regression algorithm to correct the error, correcting the system error through a layer-by-layer progressive regression method, and finally obtaining an optimized geographic information model; S8. Generate real estate surveying and mapping maps based on the optimized geographic information model, marking real estate boundaries and terrain features.

2. The high-precision real estate surveying and mapping method based on an optimization algorithm in a complex terrain environment according to claim 1 is characterized in that: The S1 specifically includes: S11. In complex terrain environments, multi-source data acquisition technology is used to integrate lidar sensors and global navigation satellite system sensors to collect three-dimensional point cloud data and geographic location information respectively; S12, emit a laser beam through the LiDAR sensor and receive the echo signal, and calculate the distance based on the echo time t and the speed of light c Generate 3D point cloud data of the target area, including the 3D coordinates (x i ,y i ,z i ); S13, use the GNSS sensor to receive satellite signals, calculate and record the geographic coordinates of the surveying points (x i ,y i ,z i ), providing geographic location information; S14, using an adaptive data synchronization algorithm based on topology preservation, synchronizing the three-dimensional point cloud data collected by the LiDAR sensor with the geographic coordinate data acquired by the GNSS sensor, and automatically adjusting the time and space synchronization of the data by maintaining the topological structure of the terrain data; S15. Integrate the synchronized data with topology preservation to form an initial geographic information dataset with wide coverage and diversity, which includes three-dimensional terrain relief, coordinates and topological structure.

3. The high-precision real estate surveying and mapping method based on an optimization algorithm in a complex terrain environment according to claim 1 is characterized in that: The S3 specifically includes: S31, based on the formed initial geographic information data set, extract the three-dimensional point cloud data (x i ,y i ,z i ) and corresponding geographic coordinate information; S32, gridding the extracted three-dimensional point cloud data according to geographic coordinates to generate a preliminary elevation data model, and adjusting the size and distribution of the grid cells according to the complexity of the terrain; S33, using the digital elevation model generation algorithm, based on the gridded three-dimensional point cloud data (x i ,y i ,z i ), calculate the elevation value Z in each grid cell DEM ; S34, introduce the sparse matrix interpolation algorithm, use the calculated elevation value Z DEM Construct a sparse matrix M: In the interpolation process, the sparse matrix M is processed by an optimization algorithm: Where W is the interpolation weight matrix, λ is the regularization coefficient, and R(M) is the regularization term of the sparse matrix M; S35, the elevation value Z processed by the sparse matrix interpolation algorithm ′ DEM With geographic coordinate information (x i ,y i ) to build a complete three-dimensional digital elevation model.

4. The high-precision real estate surveying and mapping method based on an optimization algorithm in a complex terrain environment according to claim 1 is characterized in that: The S5 specifically includes: S41, based on the constructed 3D digital elevation model, extract terrain feature data, including elevation value Z ′ DEM , terrain slope θ and terrain curvature κ; S42, input the extracted terrain feature data into the adaptive multi-modal evolutionary optimization algorithm to preliminarily generate multiple candidate paths in the complex terrain, each candidate path consists of a series of path points P i The data of each path point includes the elevation Z′ DEM (P i ), slope θ(P i ), curvature κ(P i ), and other topographic features; S43. In the process of path optimization, a path evaluation function f(P) is introduced: f(P)=∫ P [α·Z' DEM (x,y)+β·θ(x,y)+γ·κ(x,y)+δ·S(x,y)+∈·E(P)]ds; Among them, α, β, and γ are weight factors in path optimization, which control the influence of elevation, slope, and curvature on path selection respectively. S(x, y) represents the evolution stability at the path point (x, y). E(P) represents the global evolution complexity of the path. ds is the path element in the path integral. S44. Based on the result of the path evaluation function f(P), identify and optimize the optimal path mode, and dynamically adjust the key parameters α, β, γ, δ, ∈ of the path to adapt to different terrain characteristics through an adaptive multi-modal evolution strategy: S45, the optimal path P obtained by optimization opt Projected into a 3D digital elevation model, 3D path data is generated for navigation of surveying and mapping equipment.

5. The high-precision real estate surveying and mapping method based on an optimization algorithm in a complex terrain environment according to claim 1 is characterized in that: The S6 specifically includes: S61, based on the processed multi-source data and the generated three-dimensional path data, extracting geographic information data from different sensors, including three-dimensional point cloud data D LiDAR , geographic coordinate data D GNSS and elevation value Z ′ DEM ; S62, the extracted data and the three-dimensional terrain model M DEM To perform fusion, a high-dimensional nonlinear dynamic optimization fusion algorithm is applied to map the feature spaces of different data sources into a unified high-dimensional feature space by constructing a nonlinear mapping function F in the high-dimensional space: M fused =F(D aligned ,M DEM ); Among them, F represents the nonlinear mapping function, M fused Represents the fused geographic information model; S63. In the process of data fusion, a dynamic optimization mechanism is used to adjust the parameters of the nonlinear mapping function F in real time, and the fusion result maintains global consistency and adaptability to local terrain features: min F {∑ i,j (Z' fused (x i ,y j )-Z target (x i ,y j )) 2 +λR(F)}; Among them, Z′ fused is the fused elevation value, Z target is the target elevation value, λ is the regularization coefficient, and R(F) is the regularization term of the mapping function F; S64, the geographic information model M that has been fused by high-dimensional nonlinear dynamic optimization fused Combined with geographic coordinate information (x i ,y i ) and elevation value Z ′ fused , generate a comprehensive geographic information model M composite .

6. The high-precision real estate surveying and mapping method based on an optimization algorithm in a complex terrain environment according to claim 1 is characterized in that: The S7 specifically includes: S71. Generative integrated geographic information model M composite , extract the geographic coordinates (x i ,y i ), elevation value Z ′ fused And the corresponding target elevation value Z target ; S72, input the extracted data into the deep residual regression model, and establish an initial prediction model f res (x i ,y i )Calculate each coordinate point (x i ,y i )’s initial residual value r i : r i =Z target (x i ,y i )-Z′ fused (x i ,y i ); S73. In the deep residual regression model, the system error is corrected by using a layer-by-layer progressive regression method, and the residual value r is fitted layer by layer through a multi-layer neural network. i , the goal of each layer of regression model is to minimize the residual value of the current layer Updated elevation value Z ″ fused It is expressed as: in, represents the residual regression function of the lth layer, L is the number of regression layers; S74. In the layer-by-layer regression process, the learning rate and regression coefficient parameters of the model are dynamically adjusted to adapt to the complexity and error characteristics of different terrain areas, including adaptively adjusting the number of neural network layers and the number of neurons in each layer, so that the regression of each layer can effectively capture the error characteristics of different scales: Mopt=Mcomposite+ΔM; Among them, M opt is the final optimized model, ΔM represents the error correction matrix obtained by the deep residual regression algorithm; S75. After obtaining the error correction matrix ΔM, apply it to the comprehensive geographic information model M composite In the process, the final optimized geographic information model M is generated. opt .

7. The high-precision real estate surveying and mapping method based on an optimization algorithm in a complex terrain environment according to claim 1 is characterized in that: The S8 specifically includes: S81, based on the generated final optimized geographic information model M opt , extract the geographic coordinates (x i ,y i ), the final optimized elevation value Z ″ fused , slope θ(x i ,y i ) and the curvature κ(x i ,y i )’s terrain feature data; S82, based on the extracted geographic information data, use the real estate mapping map generation algorithm to preliminarily identify the real estate boundaries in the terrain model, and analyze the elevation value Z ″ fused , slope θ(x i ,y i ) and the curvature κ(x i ,y i ) changes, determine the initial position of the real estate boundary B init : B init ={(x i ,y i )|f(B(x i ,y i ))=0}; Among them, f(B(x i ,y i )) is the boundary judgment function; S83, introduce the topology sensitive boundary optimization algorithm, and optimize the initial real estate boundary B init Further adjustments are made to dynamically adjust the position of the boundary line based on the extracted terrain feature data: B opt =argmin B {∑ i,j [f(B(x i ,y j ))·w(x i ,y j )+λ·T(x i ,y j )]}; Among them, B opt is the optimized real estate boundary line, w(x i ,y j ) is the weight factor, λ is the smoothing parameter, T(x i ,y j ) is a topology sensitive function; S84, optimize the real estate boundary line B opt Projection to 3D digital elevation model M opt In combination with the optimized terrain feature data θ(x i ,y i ) and κ(x i ,y i ) to generate a real estate mapping map containing real estate boundaries and terrain features; S85. In the generated surveying and mapping map, the real estate boundary B opt The main terrain features are annotated in detail, and the annotated locations and attribute values ​​are expressed in geographic coordinates (x i ,y i ) and the corresponding elevation value Z ″ fused Determine and optimize the final geographic information model M opt Make associations; S86. Save the generated real estate surveying and mapping map as a digital file in a specified format, wherein the digital file includes a geographic information model and terrain feature annotations.

Images