Landslide TLS site planning method coupling local and global optimization

Abstract

The invention discloses a landslide TLS site planning method coupled with local and global optimization, which comprises the following steps: firstly, constructing a three-dimensional model of an observation area, delimiting a candidate area for placing a TLS scanner in the three-dimensional model of the observation area, setting a screening grid in the candidate area to obtain candidate sites, and then carrying out observation simulation on each candidate site to obtain a landslide TLS site planning model; then selecting a local optimal TLS site set from all candidate sites by adopting a greedy algorithm, and finally selecting an optimal TLS site layout based on the local optimal TLS site set by adopting a genetic algorithm; according to the method, the influence of a real three-dimensional environment on the visibility between a candidate site and an observation target point is simulated, so that the reliability of the TLS observation network obtained by the method is ensured; according to the method, the global optimization process of the genetic algorithm is performed under the guidance of the local optimal solution obtained by the greedy algorithm, and the optimal TLS site layout is efficiently obtained, so that the quality and efficiency of the obtained TLS site layout of the large-range surface deformation area are ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of ground laser radar point cloud acquisition, and in particular relates to a landslide TLS site planning method coupling local and global optimization, which is suitable for obtaining an optimal TLS site layout. Background Art

[0002] Terrestrial three-dimensional laser scanning (TLS) technology is an important means to obtain high-density and high-precision three-dimensional coordinates of the surface. It can observe the surface deformation area in multiple phases and obtain the high-resolution deformation field of the entire area through subsequent processing. In this process, it is crucial to obtain scanning data with high observation accuracy at low cost and high efficiency. For this reason, many scholars have studied the impact of scanning geometry of different TLS site layouts on observation accuracy. However, in the current mainstream monitoring of surface deformation, TLS observation sites are still determined by users subjectively, which is easy to cause redundancy and voids in scanning data, and make the observation accuracy of collected data low, which seriously affects efficiency, cost and data quality. When the observation range becomes larger, such problems will become more serious. Therefore, it is of obvious scientific and economic value to scientifically plan the layout of TLS sites in large-scale potential areas of surface deformation that require long-term monitoring.

[0003] Existing technologies mainly use model-based methods to plan TLS site layouts. Based on the existing observation scenario model, TLS scans of candidate stations are simulated, constraint equations such as observation completeness and cost are established, and optimization algorithms such as greedy, genetic, and particle swarm are used to determine the TLS site layout. However, these studies are usually based on small-scale regional simulation planning based on two-dimensional topographic maps, ignoring the height difference characteristics of the surface deformation area, resulting in simulation distortion and affecting the reliability of TLS site planning. In addition, existing methods cannot efficiently handle the large amount of calculations in the three-dimensional global optimization site selection of large-scale TLS observation networks. It is necessary to consider how to guide the global optimization process based on local optimal solutions to improve the efficiency of planning methods and make them applicable to large-scale scenarios.

[0004] The existing technology has the problem of insufficient observation accuracy due to the lack of TLS site planning methods for surface deformation areas, and the problem of low efficiency of the global optimization algorithm for TLS site layout. Summary of the invention

[0005] The object of the present invention is to provide a landslide TLS site planning method coupling local and global optimization in view of the above problems existing in the prior art.

[0006] The above-mentioned purpose of the present invention is achieved by the following technical means:

[0007] A landslide TLS site planning method coupled with local and global optimization includes the following steps:

[0008] Step 1, obtain the three-dimensional information of the terrain and objects in the observation area, and reconstruct the three-dimensional information into three-dimensional point cloud information, set the size of the voxel, and convert the three-dimensional point cloud information into voxels: delete the voxels that do not include the point cloud, and the remaining voxels constitute the three-dimensional model of the observation area, mark all voxels in the landslide area and the surface deformation area as observation target points, and obtain the digital elevation model DEM and digital surface model DSM of the observation area based on the three-dimensional information and the three-dimensional model;

[0009] The three dimensions are x-direction, y-direction, and z-direction, where x-direction and y-direction are horizontal directions;

[0010] Step 2: Delineate candidate areas for placing TLS scanners in the three-dimensional model of the observation area;

[0011] Step 3: Set a screening grid with a set resolution in the candidate area, and use all grid points in the screening grid as candidate sites, obtain the sensor parameters of the TLS scanner, and perform observation simulation on each candidate site under the constraints of the sensor parameters and the three-dimensional model, and use the observation target points that meet all constraints as the measured observation target points;

[0012] Step 4: Use a greedy algorithm to select the local optimal TLS site set from all candidate sites;

[0013] Step 5: Use a genetic algorithm to select the optimal TLS site layout based on the local optimal TLS site set.

[0014] As mentioned above, step 2 specifically includes the following steps:

[0015] Step 2.1, deleting the voxels included in the area of ​​all buildings from the three-dimensional model of the observation area;

[0016] Step 2.2, deleting the voxels included in the forest area from the three-dimensional model of the remaining area obtained in step 2.1;

[0017] Step 2.3, calculating the difference between the digital surface model DSM and the digital elevation model DEM corresponding to the remaining area obtained in step 2.2, and deleting the voxels included in the area where all the differences are not 0 from the three-dimensional model of the remaining area obtained in step 2.2;

[0018] Step 2.4, using the digital elevation model DEM to calculate the slope values ​​of the voxels in the remaining area obtained in step 2.3, and deleting the voxels included in the area with all slope values ​​greater than the set slope value from the three-dimensional model of the remaining area obtained in step 2.3, to obtain the candidate area for placing the TLS scanner.

[0019] The slope value of the voxel in step 2.4 as described above is calculated based on the following formula:

[0020]

[0021] Wherein, slop(i,j) is the slope value of the voxel with horizontal coordinates (i,j+1); dx and dy are the unit length of the voxel in the x direction and the unit length of the voxel in the 3D model, respectively; i is the coordinate in the x direction, and j is the coordinate in the y direction;

[0022] z i,j+1 is the elevation of the voxel with horizontal coordinate (i, j+1) in the digital elevation model DEM;

[0023] z i,j-1 is the elevation of the voxel with horizontal coordinate (i, j-1) in the digital elevation model DEM;

[0024] z i+1,j is the elevation of the voxel with horizontal coordinate (i+1, j) in the digital elevation model DEM;

[0025] z i-1,j is the elevation of the voxel with horizontal coordinate (i-1, j) in the digital elevation model DEM.

[0026] As mentioned above, step 3 specifically includes the following steps:

[0027] Step 3.1, set a screening grid of a set resolution in the candidate area, and regard all grid points in the screening grid as candidate sites, number all candidate sites, and then obtain the sensor parameters of the TLS scanner, determine the maximum scanning range and the minimum scanning range of the candidate site based on the sensor parameters, complete the simulation of the scanning range, and determine the scanning range of the candidate site;

[0028] Step 3.2, traverse all observation target points, calculate the angle between the direction vector from the candidate site to each observation target point and the normal vector of the surface where the observation target point is located, and select the observation target points that are less than or equal to the preset angle;

[0029] Step 3.3, draw a straight line with the candidate site as the starting point and the observation target point as the end point to obtain the observation sight line between the candidate site and the observation target point, insert an observation sight line point on the observation sight line every step length s, and then filter out the observation target points whose sight lines are not blocked;

[0030] Step 3.4, traverse all candidate sites in sequence according to the number of the candidate sites, and obtain the observation target points that meet steps 3.1-3.4 within the scanning range of each candidate site as the measured observation target points of each candidate site. The union of the measured observation target points of all candidate sites constitutes the simulated observation content.

[0031] As mentioned above, step 4 specifically includes the following steps:

[0032] Step 4.1, construct a local optimal TLS site set, and then use the number of measured observation target points of each candidate site as the weight of the corresponding candidate site;

[0033] Step 4.2: Select the candidate site with the largest current weight and add it to the local optimal TLS site set;

[0034] Step 4.3, deleting the measured observation target point of the candidate site with the largest current weight from the model observation content to obtain a new model observation content;

[0035] Step 4.4, recalculate the number of measured observation target points in all remaining candidate sites that are still in the new model observation content and use it as the weight of the corresponding candidate site;

[0036] Step 4.5: Set a weight threshold, and repeat steps 4.2-4.4 until the weight of the candidate site with the largest current weight is lower than the set weight threshold.

[0037] The above step 5 specifically includes the following steps:

[0038] Step 5.1, find and calculate the distance d between each candidate site and the nearest candidate site in the local optimal TLS site set through the KD tree;

[0039] Step 5.2: Construct a Gaussian kernel function to calculate the selection probability p. The Gaussian kernel function is based on the following formula:

[0040]

[0041] Where σ is the standard deviation of the Gaussian kernel function;

[0042] Step 5.3, calculate the selection probability p of each candidate site, and normalize the selection probability p of all candidate sites, then set the population size M, and perform M samplings according to the normalized selection probability p of each candidate site. The number of candidate sites sampled each time is equal to the number of candidate sites in the local optimal TLS site set obtained in step 4. The sampling result of each sampling is regarded as an individual, and all the individuals sampled constitute the initial population;

[0043] Step 5.4, calculate the fitness of each individual through the fitness function Fitness;

[0044] Step 5.5, a set of binary codes is set for each individual based on the numbers of all candidate sites, the length of the binary code is the number of all candidate sites, the encoding order of the binary code is the number order of all candidate sites, the binary code of the corresponding number of the candidate site included in the individual is 1, and the binary code of the corresponding number of the candidate site not included in the individual is 0;

[0045] Step 5.6, iterate the initial population n times, and put the candidate sites including the individuals with the largest fitness in the offspring population obtained in the nth iteration into the optimal TLS site layout;

[0046] Where n is the maximum number of iterations.

[0047] As mentioned above, the fitness function Fitness is calculated based on the following formula:

[0048]

[0049] In the formula, t is the individual; · is the multiplication sign; num vp is the number of candidate sites included in individual t; cov is the coverage rate of individual t to the observed target points; num lv is the number of measured observation target points of all candidate sites included in individual t; k is the index of the measured observation target point; A is the number set of candidate sites included in individual t, a is the number of candidate sites in individual t; θ a,k is the incident angle between the candidate site numbered a and the kth measured observation target point; r a,k is the distance between the candidate site numbered a and the kth measured observation target point; r max is the maximum range of the TLS scanner; r min The minimum range for TLS scanners.

[0050] As mentioned above, step 5.6 specifically includes the following steps:

[0051] Step 5.6.1, obtain the parent population, and generate a child population of the same size as the population size based on the parent population through the roulette algorithm;

[0052] In the first iteration, the parent population is the initial population; in the second to nth iterations, the parent population is the iterated child population obtained in the previous iteration;

[0053] Step 5.6.2, randomly pair the offspring population and exchange some genes between the paired chromosomes through crossover operation. The chromosomes are individuals, and the genes of the chromosomes are the binary codes of the individuals;

[0054] Step 5.6.3, set the mutation rate, then randomly select a candidate site number as the gene mutation position, and then modify the gene at the gene mutation position according to the mutation rate to obtain the iterated offspring population;

[0055] Step 5.6.4, repeat steps 5.4 to 5.9 n times, perform n iterations, convert the binary codes corresponding to each individual in the offspring population after the nth iteration into a set of candidate sites, and obtain the candidate sites included in each individual;

[0056] Step 5.6.5, calculate the fitness of each individual, then output the individual with the largest fitness, and put the candidate sites included in the individual with the largest fitness into the optimal TLS site layout.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] (1) Three-dimensionalization and reliability: The present invention simulates the impact of the real three-dimensional environment on the visibility between the candidate site and the observation target point, thereby ensuring the reliability of the TLS observation network obtained by the present invention;

[0059] (2) Global optimization and high efficiency: The present invention conducts a global optimization process of a genetic algorithm under the guidance of a local optimal solution obtained by a greedy algorithm, and efficiently obtains the optimal TLS site layout, thereby ensuring the quality and efficiency of the TLS site layout in a large-scale surface deformation area.

[0060] (3) The present invention can serve for long-term monitoring of large-scale surface deformation areas, significantly improving the deformation monitoring capability of TLS and providing important technical support for the study of disaster mechanisms and disaster prevention and mitigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 is a flow chart of the method of the present invention;

[0062] Figure 2 This is a schematic diagram of the structure of making an observation sight line between a candidate site and an observation target point in step 3.3 of embodiment 1 of the present invention. DETAILED DESCRIPTION

[0063] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below in conjunction with embodiments. The embodiments described herein are only used to illustrate and explain the present invention and are not intended to limit the present invention.

[0064] Embodiment 1:

[0065] A landslide TLS site planning method coupled with local and global optimization includes the following steps:

[0066] Step 1: Use UAV aerial photogrammetry to quickly obtain the three-dimensional information of the terrain and objects in the observation area, and reconstruct the three-dimensional information into three-dimensional point cloud information. Set the size of the voxel (such as 1 meter) and convert the three-dimensional point cloud information into voxels: delete the voxels that do not include the point cloud, and the remaining voxels constitute the three-dimensional model of the observation area; mark all voxels in the landslide area and the surface deformation area as observation target points, and make a digital elevation model (DEM, Digital Elevation Model) and a digital surface model (DSM, Digital Surface Model) of the observation area based on the three-dimensional information and the three-dimensional model.

[0067] The three dimensions include an x-direction, a y-direction, and a z-direction, wherein the x-direction and the y-direction are horizontal directions.

[0068] Step 2: Using the digital elevation model DEM and digital surface model DSM obtained in step 1, based on the principle of station feasibility, the candidate area for placing the TLS scanner is delineated in the three-dimensional model of the observation area, which specifically includes the following steps:

[0069] Step 2.1, deleting the voxels included in the area of ​​all buildings from the three-dimensional model of the observation area;

[0070] Step 2.2, deleting all voxels included in the area (such as forest area) that has a serious impact on the observation quality from the three-dimensional model of the remaining area obtained in step 2.1;

[0071] Step 2.3, calculating the difference between the digital surface model DSM and the digital elevation model DEM corresponding to the remaining area obtained in step 2.2, and marking all areas where the difference is not equal to 0 as areas where a station cannot be set up, and deleting the voxels included in the area where a station cannot be set up from the three-dimensional model of the remaining area obtained in step 2.2;

[0072] Step 2.4, using the digital elevation model DEM to calculate the slope value slop of the voxels in the remaining area obtained in step 2.3, and marking all areas with slope values ​​greater than 30 as areas that affect operator safety (such as cliffs or steep slopes), and removing the voxels included in all marked areas in this step from the three-dimensional model of the remaining area obtained in step 2.3, thereby obtaining a candidate area for placing the TLS scanner;

[0073] The slope value of a voxel is calculated based on the following formula:

[0074]

[0075] Wherein, slop(i,j) is the slope value of the voxel with horizontal coordinates (i,j+1); dx and dy are the unit length of the voxel in the x direction and the unit length of the voxel in the 3D model, respectively; i is the coordinate in the x direction, and j is the coordinate in the y direction;

[0076] z i,j+1 is the elevation of the voxel with horizontal coordinate (i, j+1) in the digital elevation model DEM;

[0077] z i,j-1 is the elevation of the voxel with horizontal coordinate (i, j-1) in the digital elevation model DEM;

[0078] z i+1,j is the elevation of the voxel with horizontal coordinate (i+1, j) in the digital elevation model DEM;

[0079] z i-1,j is the elevation of the voxel with horizontal coordinate (i-1, j) in the digital elevation model DEM.

[0080] Step 3: Set a screening grid with a resolution of 15m×15m in the candidate area obtained in step 2, and use all grid points in the screening grid as candidate sites, obtain the sensor parameters of the TLS scanner, and perform observation simulation on each candidate site under the constraints of the sensor parameters and the three-dimensional model, use the observation target points that meet all constraints as measured observation target points, and record all observed target points in the simulated observation content of the simulated candidate site, which specifically includes the following steps:

[0081] Step 3.1, set a screening grid with a resolution of 15m×15m in the candidate area obtained in step 2.3, and regard all grid points in the screening grid as candidate sites, number all candidate sites, and then input sensor parameters, and determine the maximum scanning range and minimum scanning range of the candidate site based on the sensor parameters, thereby completing the simulation of the scanning range and determining the scanning range of the candidate site;

[0082] Step 3.2, traverse all observation target points, calculate the angle between the direction vector from the candidate site to each observation target point and the normal vector of the surface where the observation target point is located, and select the observation target points that are less than or equal to the preset angle condition;

[0083] Step 3.3, draw a straight line with the candidate site as the starting point and the observation target point as the end point to obtain the observation line of sight between the candidate site and the observation target point, insert an observation line of sight point on the observation line of sight every step length s, compare the elevations of all observation line of sight points with the elevations corresponding to the horizontal coordinates of the observation line of sight points in the digital surface model DSM, if the elevation of any observation line of sight point is less than or equal to the elevation corresponding to the horizontal coordinates of the observation line of sight point in the digital surface model DSM, then mark the observation target point as an observation target point whose line of sight is blocked; if the elevations of all observation line of sight points are greater than the elevation corresponding to the horizontal coordinates of the observation line of sight points in the digital surface model DSM, then mark the observation target point as an observation target point whose line of sight is not blocked; traverse all observation target points within the scanning range of the candidate site, and filter out the observation target points whose line of sight is not blocked. In this embodiment, the fixed step length s is equal to the voxel size.

[0084] The computational complexity of this method does not increase exponentially with the increase of scanning range.

[0085] Step 3.4, traverse all candidate sites in turn according to the number of the candidate sites, and take the observation target points that meet steps 3.1-3.4 within the scanning range of all candidate sites as measured observation target points, and all observed target points constitute the model observation content.

[0086] The sensor parameters of the TLS scanner that are mainly considered in the present invention are: scanning range, that is, the maximum measurement distance at which the target reflectivity is above 90%; incident angle (the preset angle in step 3.2 indirectly reflects the incident angle), that is, the angle between the laser beam of the scanner and the surface normal of the observed object. The scanning range and elevation angle are determined when the instrument leaves the factory, and can be directly determined by querying the parameters. There are many factors that affect the incident angle. From the perspective of echo, the present invention sets the maximum incident angle to 75°.

[0087] Step 4: Use a greedy algorithm to select the local optimal TLS site set, which specifically includes the following steps:

[0088] Step 4.1, construct the local optimal TLS site set, take the union of the measured observation target points of all candidate sites in step 3.4, obtain the union of the measured observation target points, and then use the number of measured observation target points of each candidate site as the weight of the corresponding candidate site;

[0089] Step 4.2: Select the candidate site with the largest current weight and add it to the local optimal TLS site set;

[0090] Step 4.3, deleting the measured observation target point of the candidate site with the largest current weight from the measured observation target point union, to obtain a new measured observation target point union;

[0091] Step 4.4, recalculate the number of measured observation target points in the new measured observation target point set among the measured observation target points of all remaining candidate sites and use it as the weight of the corresponding candidate site;

[0092] Step 4.5: Repeat steps 4.2-4.4 until the weight of the candidate site with the largest current weight is lower than the set weight threshold.

[0093] Ideally, the measurement should stop when the union of the measured observation target points is an empty set. However, due to the extremely complex real environment, it is difficult to achieve full coverage of the measurement area in most cases, so consider setting a weight threshold.

[0094] Step 5: Use a genetic algorithm to select the optimal TLS site layout under the guidance of the local optimal TLS site set. The specific steps are as follows:

[0095] Step 5.1: Use the KD tree to quickly find the distance d between all candidate sites and the nearest candidate site in the local optimal TLS site set.

[0096] Step 5.2: Construct a Gaussian kernel function to calculate the selection probability p. The Gaussian kernel function is based on the following formula:

[0097]

[0098] Wherein, d is the distance from the candidate site to the nearest candidate site in the local optimal TLS site set, and σ is the standard deviation of the Gaussian kernel function, which is used to control the width of the probability distribution. In this embodiment, it is set to 1.5. Adjusting σ can affect the shape of the probability distribution. A smaller σ will make the probability distribution more concentrated near the zero point, and a larger σ will make the probability distribution more dispersed.

[0099] In order to make the probability distribution smoother and similar to the normal distribution, we use the Gaussian kernel function to adjust the probability distribution. The Gaussian kernel function can generate an exponentially decaying probability distribution based on the distance from the discrete point to the target point, which ensures that the probability of selecting points farther away is lower, rather than a simple linear decay.

[0100] Step 5.3, calculate the selection probability p of each candidate site, and normalize the selection probability p of all candidate sites, then set the population size M, perform M samplings according to the normalized selection probability p of each candidate site, each sampling result is regarded as an individual, and all the individuals drawn constitute the initial population, specifically including the following steps:

[0101] Step 5.3.1, calculate the selection probability p of each candidate site, and normalize the selection probabilities p of all candidate sites so that the sum of the selection probabilities of all candidate sites is equal to 1;

[0102] Step 5.3.2, set the population size M, and perform M samplings according to the normalized selection probability p. The number of candidate sites sampled each time is equal to the number of candidate sites in the local optimal TLS site set obtained in step 4. The sampling result of each sampling is regarded as an individual;

[0103] Step 5.3.3: All individuals constitute the initial population.

[0104] Step 5.4: Calculate the fitness of each individual through the fitness function Fitness, so as to evaluate the quality of each individual. The individual with greater fitness is better.

[0105] The fitness function Fitness is based on the following formula:

[0106]

[0107] In the formula, t is the individual; · is the multiplication sign; num vp is the number of candidate sites included by individual t; cov is the coverage rate of individual t for the observed target points, which is the number of target observation points within the scanning range of all candidate sites included by individual t divided by the number of all observed target points in step 1; num lv is the number of measured observation target points of all candidate sites included in individual t; k is the index of the measured observation target point; A is the number set of candidate sites included in individual t, a is the number of candidate sites in individual t; θ a,k is the incident angle between the candidate site numbered a and the kth measured observation target point; r a,k is the distance between the candidate site numbered a and the kth measured observation target point; r max is the maximum range of the TLS scanner used; r min The minimum range of the TLS scanner used.

[0108] Step 5.5, set a set of binary codes for each individual based on the numbers of all candidate sites, the length of the binary code is the total number of all candidate sites, and the encoding order of the binary code is the numbering order of all candidate sites, wherein the binary code of the corresponding number of the candidate site included in the individual is 1, and the binary code of the corresponding number of the candidate site not included in the individual is 0, wherein 1 represents setting a TLS scanner at the corresponding candidate site, and 0 represents not setting a TLS scanner at the corresponding candidate site.

[0109] Step 5.6, iterate the initial population n times, and put the candidate sites including the individuals with the largest fitness in the offspring population obtained in the nth iteration into the optimal TLS site layout, which specifically includes the following steps:

[0110] Where n is the maximum number of iterations.

[0111] Step 5.6.1, obtain the parent population, and generate a child population of the same size as the population size based on the parent population through the "roulette wheel" algorithm.

[0112] Among them, in the first iteration, the parent population is the initial population; in the second to nth iterations, the parent population is the iterated child population obtained in the previous iteration.

[0113] Step 5.6.2: Randomly pair the offspring population and exchange some genes between the paired chromosomes through crossover operation. The chromosome is the individual, and the gene of the chromosome is the binary code of the individual.

[0114] Step 5.6.3, set the mutation rate (such as 0.2), randomly select the number of a candidate site as the gene mutation position, and then modify the gene at the gene mutation position according to the mutation rate to obtain the iterated offspring population.

[0115] Step 5.6.4, repeat steps 5.4 to 5.9 n times, thus performing n iterations, converting the binary codes corresponding to each individual in the offspring population after the nth iteration into a set of candidate sites, and obtaining the candidate sites included in each individual.

[0116] Step 5.6.5, calculate the fitness of each individual, then output the individual with the largest fitness, and put the candidate sites included in the individual with the largest fitness into the optimal TLS site layout.

[0117] It should be noted that the embodiments described in the present invention are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the described embodiments or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the attached claims.

Claims

1. A landslide TLS site planning method coupled with local and global optimization, characterized in that: The following steps are involved: Step 1, obtain the three-dimensional information of the terrain and objects in the observation area, and reconstruct the three-dimensional information into three-dimensional point cloud information, set the size of the voxel, and convert the three-dimensional point cloud information into voxels: delete the voxels that do not include the point cloud, and the remaining voxels constitute the three-dimensional model of the observation area, mark all voxels in the landslide area and the surface deformation area as observation target points, and obtain the digital elevation model DEM and digital surface model DSM of the observation area based on the three-dimensional information and the three-dimensional model; The three dimensions are x-direction, y-direction, and z-direction, where x-direction and y-direction are horizontal directions; Step 2: Delineate candidate areas for placing TLS scanners in the three-dimensional model of the observation area; Step 3: Set a screening grid with a set resolution in the candidate area, and use all grid points in the screening grid as candidate sites, obtain the sensor parameters of the TLS scanner, and perform observation simulation on each candidate site under the constraints of the sensor parameters and the three-dimensional model, and use the observation target points that meet all constraints as the measured observation target points; Step 4: Use a greedy algorithm to select the local optimal TLS site set from all candidate sites; Step 5: Use a genetic algorithm to select the optimal TLS site layout based on the local optimal TLS site set.

2. A landslide TLS site planning method coupled with local and global optimization according to claim 1, characterized in that: The step 2 specifically includes the following steps: Step 2.1, deleting the voxels included in the area of ​​all buildings from the three-dimensional model of the observation area; Step 2.2, deleting the voxels included in the forest area from the three-dimensional model of the remaining area obtained in step 2.1; Step 2.3, calculating the difference between the digital surface model DSM and the digital elevation model DEM corresponding to the remaining area obtained in step 2.2, and deleting the voxels included in the area where all the differences are not 0 from the three-dimensional model of the remaining area obtained in step 2.2; Step 2.4, using the digital elevation model DEM to calculate the slope values ​​of the voxels in the remaining area obtained in step 2.3, and deleting the voxels included in the area with all slope values ​​greater than the set slope value from the three-dimensional model of the remaining area obtained in step 2.3, to obtain the candidate area for placing the TLS scanner.

3. A landslide TLS site planning method coupled with local and global optimization according to claim 2, characterized in that: The slope value of the voxel in step 2.4 is calculated based on the following formula: Wherein, slop(i,j) is the slope value of the voxel with horizontal coordinates (i,j+1); dx and dy are the unit length of the voxel in the x direction and the unit length of the voxel in the 3D model, respectively; i is the coordinate in the x direction, and j is the coordinate in the y direction; z i,j+1 is the elevation of the voxel with horizontal coordinate (i, j+1) in the digital elevation model DEM; z i,j-1 is the elevation of the voxel with horizontal coordinate (i, j-1) in the digital elevation model DEM; z i+1,j is the elevation of the voxel with horizontal coordinate (i+1, j) in the digital elevation model DEM; z i-1,j is the elevation of the voxel with horizontal coordinate (i-1, j) in the digital elevation model DEM.

4. A landslide TLS site planning method coupled with local and global optimization according to claim 3, characterized in that: The step 3 specifically includes the following steps: Step 3.1, set a screening grid of a set resolution in the candidate area, and regard all grid points in the screening grid as candidate sites, number all candidate sites, and then obtain the sensor parameters of the TLS scanner, determine the maximum scanning range and the minimum scanning range of the candidate site based on the sensor parameters, complete the simulation of the scanning range, and determine the scanning range of the candidate site; Step 3.2, traverse all observation target points, calculate the angle between the direction vector from the candidate site to each observation target point and the normal vector of the surface where the observation target point is located, and select the observation target points that are less than or equal to the preset angle; Step 3.3, draw a straight line with the candidate site as the starting point and the observation target point as the end point to obtain the observation sight line between the candidate site and the observation target point, insert an observation sight line point on the observation sight line every step length s, and then filter out the observation target points whose sight lines are not blocked; Step 3.4, traverse all candidate sites in sequence according to the number of the candidate sites, and obtain the observation target points that meet steps 3.1-3.4 within the scanning range of each candidate site as the measured observation target points of each candidate site. The union of the measured observation target points of all candidate sites constitutes the simulated observation content.

5. A landslide TLS site planning method coupled with local and global optimization according to claim 4, characterized in that: The step 4 specifically comprises the following steps: Step 4.1, construct a local optimal TLS site set, and then use the number of measured observation target points of each candidate site as the weight of the corresponding candidate site; Step 4.2: Select the candidate site with the largest current weight and add it to the local optimal TLS site set; Step 4.3, deleting the measured observation target point of the candidate site with the largest current weight from the model observation content to obtain a new model observation content; Step 4.4, recalculate the number of measured observation target points in all remaining candidate sites that are still in the new model observation content and use it as the weight of the corresponding candidate site; Step 4.5: Set a weight threshold, and repeat steps 4.2-4.4 until the weight of the candidate site with the largest current weight is lower than the set weight threshold.

6. A landslide TLS site planning method coupled with local and global optimization according to claim 5, characterized in that: The step 5 specifically comprises the following steps: Step 5.1, find and calculate the distance d between each candidate site and the nearest candidate site in the local optimal TLS site set through the KD tree; Step 5.2: Construct a Gaussian kernel function to calculate the selection probability p. The Gaussian kernel function is based on the following formula: Where σ is the standard deviation of the Gaussian kernel function; Step 5.3, calculate the selection probability p of each candidate site, and normalize the selection probability p of all candidate sites, then set the population size M, and perform M samplings according to the normalized selection probability p of each candidate site. The number of candidate sites sampled each time is equal to the number of candidate sites in the local optimal TLS site set obtained in step 4. The sampling result of each sampling is regarded as an individual, and all the individuals sampled constitute the initial population; Step 5.4, calculate the fitness of each individual through the fitness function Fitness; Step 5.5, a set of binary codes is set for each individual based on the numbers of all candidate sites, the length of the binary code is the number of all candidate sites, the encoding order of the binary code is the number order of all candidate sites, the binary code of the corresponding number of the candidate site included in the individual is 1, and the binary code of the corresponding number of the candidate site not included in the individual is 0; Step 5.6, iterate the initial population n times, and put the candidate sites including the individuals with the largest fitness in the offspring population obtained in the nth iteration into the optimal TLS site layout; Among them, n is the maximum number of iterations.

7. A landslide TLS site planning method coupled with local and global optimization according to claim 6, characterized in that: The fitness function Fitness is calculated based on the following formula: In the formula, t is the individual; · is the multiplication sign; num vp is the number of candidate sites included in individual t; cov is the coverage rate of individual t to the observed target points; num lv is the number of measured observation target points of all candidate sites included in individual t; k is the index of the measured observation target point; A is the number set of candidate sites included in individual t, a is the number of candidate sites in individual t; θ a,k is the incident angle between the candidate site numbered a and the kth measured observation target point; r a,k is the distance between the candidate site numbered a and the kth measured observation target point; r max is the maximum range of the TLS scanner; r min The minimum range for TLS scanners.

8. A landslide TLS site planning method coupled with local and global optimization according to claim 7, characterized in that: The step 5.6 specifically includes the following steps: Step 5.6.1, obtain the parent population, and generate a child population of the same size as the population size based on the parent population through the roulette algorithm; In the first iteration, the parent population is the initial population; in the second to nth iterations, the parent population is the iterated child population obtained in the previous iteration; Step 5.6.2, randomly pair the offspring population and exchange some genes between the paired chromosomes through crossover operation. The chromosomes are individuals, and the genes of the chromosomes are the binary codes of the individuals; Step 5.6.3, set the mutation rate, then randomly select a candidate site number as the gene mutation position, and then modify the gene at the gene mutation position according to the mutation rate to obtain the iterated offspring population; Step 5.6.4, repeat steps 5.4 to 5.9 n times, perform n iterations, convert the binary codes corresponding to each individual in the offspring population after the nth iteration into a set of candidate sites, and obtain the candidate sites included in each individual; Step 5.6.5, calculate the fitness of each individual, then output the individual with the largest fitness, and put the candidate sites included in the individual with the largest fitness into the optimal TLS site layout.

Images