CN116894400B - Spatial combination site selection method based on visual sense network - Google Patents
Spatial combination site selection method based on visual sense network Download PDFInfo
- Publication number
- CN116894400B CN116894400B CN202311159848.4A CN202311159848A CN116894400B CN 116894400 B CN116894400 B CN 116894400B CN 202311159848 A CN202311159848 A CN 202311159848A CN 116894400 B CN116894400 B CN 116894400B
- Authority
- CN
- China
- Prior art keywords
- visual
- sparse
- operator
- vsn
- space
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000000007 visual effect Effects 0.000 title claims abstract description 133
- 238000010187 selection method Methods 0.000 title claims abstract description 9
- 238000004458 analytical method Methods 0.000 claims abstract description 68
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 56
- 239000011159 matrix material Substances 0.000 claims abstract description 47
- 238000005457 optimization Methods 0.000 claims abstract description 38
- 238000010276 construction Methods 0.000 claims abstract description 34
- 238000004364 calculation method Methods 0.000 claims abstract description 24
- 238000000034 method Methods 0.000 claims description 39
- 230000035772 mutation Effects 0.000 claims description 31
- 238000012544 monitoring process Methods 0.000 claims description 25
- 108090000623 proteins and genes Proteins 0.000 claims description 19
- 210000000349 chromosome Anatomy 0.000 claims description 17
- 230000008439 repair process Effects 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 11
- 230000002068 genetic effect Effects 0.000 claims description 8
- 238000013461 design Methods 0.000 claims description 7
- 230000014759 maintenance of location Effects 0.000 claims description 6
- 230000006978 adaptation Effects 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 238000010353 genetic engineering Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000008263 repair mechanism Effects 0.000 claims description 2
- 238000012732 spatial analysis Methods 0.000 claims description 2
- 101150084044 P gene Proteins 0.000 claims 1
- 230000006870 function Effects 0.000 description 17
- 238000002474 experimental method Methods 0.000 description 10
- 238000011160 research Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000006467 substitution reaction Methods 0.000 description 3
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000004880 explosion Methods 0.000 description 2
- 238000012876 topography Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000001983 electron spin resonance imaging Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000008407 joint function Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000013215 result calculation Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Abstract
The invention discloses a space combination site selection method based on a visual sense network, which comprises the steps of firstly, providing a sparse multi-view analysis algorithm, calculating to obtain the space coverage range of each camera in each visual sense network, constructing a space visual field cost matrix, and reducing the multi-view space analysis in a three-dimensional geographic space to be the numerical calculation of the visual field cost matrix; secondly, taking a visual area cost matrix as an input condition, and designing a visual sensing network maximum coverage site selection model; and (3) providing an optimization algorithm for the MCLP-VSN model, solving a spatial combination addressing scheme of the visual sensing network, and finally forming a spatial combination addressing result of the visual sensing network. The invention realizes the quantitative calculation and analysis of the spatial combination site selection of the visual sensing network with high efficiency, can improve the visual coverage of VSN, can reduce the VSN engineering construction cost, and can be widely applied to VSN engineering construction of parks, universities, commercial complexes, traffic road networks, forest areas, natural protection areas and the like.
Description
Technical Field
The invention belongs to the field of spatial combination site selection and decision scheme making of a visual sense network, and particularly relates to a spatial combination site selection method based on a visual sense network.
Background
A vision sensing network (Visual Senser Networks, VSN) is a system that enables global video coverage and management of an observation region of interest (Region of Interesting, ROI) by connecting multiple cameras into the network. These cameras may be deployed in different locations, such as buildings, public places, businesses or homes. Through a network connection, the cameras may stream real-time video to one or more monitoring centers or monitoring devices for real-time viewing and recording of the video. The vision sensing network space combination site selection is a typical space combination optimization problem, can be divided into a high carrier and a low carrier due to the difference of laid carriers, and can be widely applied to smart cities, public security and the like. At present, the research of cameras is concentrated in the directions of monitoring system design, monitoring equipment control and management system design, space site selection optimization and the like, wherein the space layout optimization is mainly based on a mode of combining qualitative and quantitative GIS coverage to obtain the optimal site selection scheme. The spatial combination site selection scheme of the visual sensing network is established by combining GIS visual field analysis and manual qualitative decision, such as modeling by utilizing GIS visual field analysis in the urban fire control field, and establishing an urban fire control high-altitude observation system site selection optimization method.
As the situation of urban management becomes more complex, the management range becomes more diversified and the management objects become more refined, the design of the existing camera layout scheme presents the basic characteristics of large scale of the number of planned construction and focusing of planned observation objects, which also brings great challenges to the mode formulated by the space combination site selection scheme of the traditional vision sensing network, and a heuristic optimization algorithm capable of quantitatively searching needs to be introduced to realize combination optimization solution, so that a configuration scheme with higher quality is obtained. However, the spatial combination optimization calculation of the visual sensing network has bottleneck problems, namely: the traditional multi-view analysis is high in time complexity and time-consuming in single-generation calculation, iterative search of a heuristic optimization algorithm cannot be met, ROI cannot be integrated into the traditional multi-view analysis, and the application value of an analysis result is difficult to meet the requirement. Summarizing challenges faced by visual sensor network deployment: (1) Large-scale space calculation, wherein the larger the calculation scale is, the more complex the solution is; (2) traditional multi-view analysis is difficult to integrate into the ROI; (3) The method is suitable for the space optimization algorithm deletion of larger scale and larger area. Therefore, the method for analyzing and optimizing the VSN visual field is applicable to large-scale analysis and optimization of the VSN visual field.
Disclosure of Invention
The invention aims to: the invention aims to provide a space combination site selection method based on a visual sense network, which can achieve efficient and quantitative visual sense network combination space layout calculation, has higher configuration solution quality than manual qualitative decision, and can be widely used for urban fire control monitoring, forest area monitoring, high-speed monitoring and ecological protection.
The technical scheme is as follows: according to the spatial combination site selection method based on the visual sensing network, firstly, a sparse multi-view analysis algorithm is provided, the spatial coverage range of each camera in each visual sensing network is calculated and obtained, a spatial visual area cost matrix is constructed, and multi-view spatial analysis in a three-dimensional geographic space is reduced to be the numerical calculation of the visual area cost matrix; secondly, taking a visual area cost matrix as an input condition, and designing a visual sensing network maximum coverage site selection model; furthermore, an optimization algorithm for the MCLP-VSN model is provided, a vision sensing network space combination addressing scheme is solved, and finally, a vision sensing network space combination addressing result is formed, wherein the method comprises the following specific steps:
step 1, providing a sparse multi-view analysis algorithm, calculating to obtain the space coverage of each camera in each visual sensing network, constructing a space visual cost matrix, and reducing the multi-view space analysis dimension in a three-dimensional geographic space into a visual cost matrix numerical calculation; the core of the sparse multi-view analysis algorithm is to draw the idea of spatial division reduction of map making synthesis by reference, thin all candidate set point visual areas into limited and regular grids and grid centroids, spatially associate interested observation areas (Region of Interesting, ROI) with the grid centroids, screen out a sparse grid centroids set in the ROI range, and form a sparse visual matrix associated with the candidate set point-sparse grid centroids set, so that complex and dense three-dimensional geographic space multi-view analysis is converted into simple matrix operation;
step 2, designing a visual sense network maximum coverage site selection MCLP-VSN model: starting from the two aspects of Objective function and constraint condition, defining an Objective function MCLP-VSN Objective of a maximum coverage site selection model of the visual sensing network by taking a visual field cost matrix as an input condition, and establishing constraint conditions MCLP-VSNConstates of the maximum coverage site selection model of the visual sensing network;
step 3, designing an optimization algorithm of a visual sense network-oriented maximum coverage site selection MCLP-VSN model: an optimization algorithm for a visual sense network maximum coverage site selection MCLP-VSN model is constructed, and a subset of layout schemes which meet the constraint condition MCLP-VSN Constraints of the visual sense network maximum coverage site selection model and have high fitness of a visual sense network maximum coverage site selection model objective function MCLP-VSNOBjective is searched and evolved from a problem set;
step 4, evaluating a spatial combination addressing scheme of the visual sensing network and performing spatial visual analysis: firstly, evaluating a spatial combination site selection scheme of a visual sense network; secondly, carrying out space visual analysis on the space combination site selection result of the visual sensing network; and finally, selecting the optimal layout scheme from the scheme subset.
Further, the step 1 specifically includes the following steps: the sparse multi-view analysis algorithm design is that a three-dimensional geographic space or a digital elevation model is thinned into a limited and regular grid and a grid centroid, and the ROI is fused into the multi-view analysis process to form a sparse visibility matrix:
step 1.1, visual field analysis of all candidate observation points: traversing a set of candidate observation pointsEvery point, get->Candidate observation points->Performing a visibility analysis to form a visible region v；
Step 1.2, constructing the ROI visible region, which is marked as；
Step 1.3, meshing expression of a VSN construction area, namely thinning a three-dimensional geographic space or a digital elevation model of the VSN construction area into a limited and regular grid and a grid centroid;
step 1.4, the ROI visible region is associated with the geometric space of the centroid point of the regular grid to form a sparse grid centroid point set, the ROI visible region of each candidate observation point is traversed and is associated with the geometric space of the centroid point of the regular grid to complete the construction of the multi-view association relationship, and the sparse grid centroid point set is formed；
Step 1.5, constructing a visual area cost matrix: sparse grid centroid point set of complete record ROI visible area information obtained according to step 1.4Establishing a cost matrix between the camera and the VSN construction area>；
Step 1.6, designing a visual area cost matrix operator: in order to improve the evolutionary computing efficiency of the MCLP-VSN model and realize the rapid statistical computation of the objective function under different layout schemes, a single candidate observation point visual field statistical operator, a multi-candidate observation point visual field statistical operator and a single candidate observation point are required to be designedThe visual field statistics operator of (2) is shown in formula (1):
；
in the formula (1)Representing the visual field result of single candidate observation point, +.>Representing sparse centroid point index->Index representing candidate observation point,/->Represents the total number of candidate observation points, +.>Representing the number of sparse centroid points;
multiple candidate observation pointsThe visual field statistics operator of (2) is as shown in formula (2):
；
in the formula (2)Representing the visual field result of the multiple candidate observation points,for a plurality of candidate observation point sequences, the operator is mainly applied to evaluating and optimizing various body multi-view visual fields in the algorithm population.
Further, the step 2 specifically includes the following steps:
step 2.1, defining an objective function of an MCLP-VSN model, wherein the MCLP-VSN model is a maximum coverage model, namely, on the premise of giving the number P of newly built cameras, the ROI map spots in a VSN construction area are covered to the maximum extent, and the calculation level is equivalent to: at candidate observation point setSelecting the total P construction sequences so that the sparse multi-view matrix is givenThe number of middle coverage sparse centroid points is the largest, and the MCLP-VSN model objective function is described as follows:
；
in the method, in the process of the invention,representing sparse centroid points->Importance weight, ROI importance level is set to 1; />Representing sparse centroid pointsWhether or not it is to be covered or not,
；
step 2.2, defining constraints of the MCLP-VSN model, can be described as:
；
in the method, in the process of the invention,the representation may cover sparse centroid points +.>Wherein +.>Refers to the actual distance of the sparse centroid point from the camera, S refers to the maximum observed coverage radius of the camera, +.>Representing visibility analysis, +.>1 represents a visual; />Representation ofCamera layout decision variables, +.>Representing the number of newly built cameras +.>Representing the total number of candidate observation points;
；
constraint (4) and (5) are implemented: traversing each sparse centroid pointConstructing a coverage sparse centroid point +.>Candidate observation point set->Will->Adding all candidate observation points to obtain a sparse centroid point +.>Actual construction results of candidate observation point set of (2) if sparse centroid point + ->The actual construction result of the candidate observation point set is more than or equal to 1, at the moment, the sparse centroid point +.>Is covered, at this time will +.>Set to 1; if sparse centroid point->The actual construction result of the candidate observation point set is large0 or more, sparse centroid point->Is not covered by the service, at this point will +.>Set to 0; constraint (6) defines that the total amount of newly laid out VSNs is limited to p; constraints (7) and (8) limit whether the overlay variable, the decision variable, is binary.
Further, the step 3 specifically includes the following steps:
step 3.1, initializing a population: initializing a first generation based on an integer encoding operator and a capacity constraint patching operatorGenerating an initial population->；
Step 3.2, pair ISubstitution population->Until a stopping criterion is met;
and 3.3, stopping population evolution and outputting optimal individuals, and stopping population evolution when the optimal fitness reaches a preset termination threshold value or the iteration number reaches a preset termination iteration algebra through iterative evolution in the step 3.2.
Further, the step 3.2 specifically includes the following steps:
step 3.2.1, calculatingTarget function adaptation value of each individual: based on the maximum space coverage objective function of the visual area cost matrix, the computing operator invokes the visual area cost matrix operator in the step 1.6 of the sparse multi-visual area analysis algorithm, and performs individual sequencing according to the adaptive value;
step 3.2.2, elite retention strategy: will beThe individuals with the highest fitness value are added with +.>In the population, the retention of elite individuals is realized;
step 3.2.3, selecting a male parent individual based on a probability selection operator: according to the principle of random selection of probability of roulette wheelTwo father individuals are selected as crossed variation to form new offspring individuals;
step 3.2.4, genetic manipulation: genetic operation is carried out on two male parent individuals based on a multipoint crossover operator and a mutation operator, capacity adjustment operation is carried out by utilizing a capacity constraint repair operator, the total number of cameras represented by the individuals is equal to a capacity constraint condition, and the generated individuals are added；
Step 3.2.5, population scale discrimination: if it isThe number of individuals reaching the population size +.>And go to step 3.2.2, otherwise, loop step 3.2.3 until +.>The number of individuals reached the standard.
Further, in step 4, the spatial combination addressing scheme of the visual sensing network includes: integer coding operators, capacity constraint patching operators, selection operators, crossover operators and mutation operators;
integer encoding operator: first, theSubstitution population->By->Chromosome constitution, wherein the first generation population chromosomes are generated by adopting a method of 'first random and then adjustment', and then each generation of chromosomes are generated by adopting a genetic operator operation mode; generate->Substitute for the firstChromosome->The basic steps of (a) are as follows: in the case of the first population, a random selection method is used, i.e.)>Chromosome with length of candidate point number +.>There is->The value of each gene locus is 1,/o>The number of the individual gene positions is 0, and the capacity constraint condition is satisfied; otherwise, operating through a selection operator, a crossover operator, a mutation operator and a capacity constraint repair operator to generate a new chromosome;
capacity constraint repair operator: the capacity constraint repair operator can realize facility configuration capacity limitation, belongs to a repair mechanism of a chromosome, and is mainly used for solving the problem that after crossing or mutation, the number of the selected monitoring points is out of limit and the capacity constraint condition is not satisfied; comprises 2 steps of difference degree calculation and constraint repair, and the set capacity constraint quantityCalculating the number of gene locus value 1 in the current chromosome +.>Calculate->And->Taking the difference as a reference, and performing constraint repair;
selecting an operator: selecting a certain number of 'good' individuals and 'bad' individuals from the population, performing genetic operation, selecting a random traversal selection operator to screen the individuals, and placing the individuals on a wheel disc and rotating the individuals once by introducing a rotator with 4 pointers which are uniformly distributed;
crossover operator: the crossover operator uses multi-point crossover logic as follows: randomly selecting two chromosomes from a population by a roulette method, if the crossing probability condition is met, randomly generating a crossing position sequence by the selected two individuals according to the set crossing number, executing crossing, leveling the gene value at the crossing position by using a capacity constraint repair operator, realizing that the sum of the gene values at the crossing position is equal to the sum of genes at the corresponding position before crossing, and performing crossing operation, otherwise, not performing any operation by the two individuals;
mutation operator: the mutation operator uses polynomial mutation logic with the following rules: traversing the whole population, if the mutation probability is met, performing mutation operation by the individual according to the set mutation quantity, and calculating a new value for the mutation gene position through polynomial mutation; leveling the gene values at the crossing positions by using a capacity constraint repair operator, and realizing that the sum of the gene values at the mutation positions is equal to the sum of the genes at the corresponding positions before mutation.
The beneficial effects are that: compared with the prior art, the invention has the following remarkable advantages: (1) The method realizes quantitative calculation of the spatial coverage, and provides a sparse multi-view analysis algorithm and an optimization algorithm for combined spatial layout of a visual sensing network aiming at the challenge of difficult quantitative calculation of the spatial coverage of the camera. (2) The method realizes the calculation of the combined spatial layout result of the visual sensing network and the rapid spatial visual analysis under the real geographic scene, thereby greatly improving the limitation of the traditional ideal visual field model or the buffer zone model and enabling the model to accord with the real geographic environment. (3) The quality of the calculated solution scheme is higher than that of a manual qualitative decision, and the method can be widely used for urban fire control monitoring, forest area monitoring, high-speed monitoring and ecological protection so as to realize video monitoring of surrounding geographic objects, and has higher application value in video monitoring layout space analysis.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of sparse multi-view analysis;
FIG. 3 is a diagram of a capacity constraint patching algorithm;
FIG. 4 is a cross-over process diagram;
FIG. 5 is a graph of the result of the mutation operation;
FIG. 6 is a graph of study area versus experimental data;
FIG. 7 is a graph of monitoring coverage area variation;
FIG. 8 is a visual result diagram of site selection optimization.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings.
As shown in FIG. 1, the method and the device for spatial combination site selection of the high-efficiency visual sense network have the following experimental steps:
preparation before experiment: according to the invention, a city is selected as a VSN construction area, a planting area is taken as an ROI, a VSN space layout experiment is carried out, the maximum coverage of the ROI in the construction area is taken as an application target, and crop protection is served. The invention selects an open source data set of a national earth system science data center, which comprises the following steps: the data overview of VSN construction area (a), plowing map spot (a), DEM elevation data (b), ROI area (c) and candidate observation point data (d) is shown in figure 6. Experiment system hardware environment: CPU: intel (R) Core (TM) i7-7590; memory: 16GB; GPU: NVIDIA GeForce GT1080TI; system software environment: operating system: operating system: windows10 64 Bit; programming language Python, GDAL, GIS platform: ESRI ArcGIS 10.6.
And step 1, designing a sparse multi-view analysis algorithm. The three-dimensional geographic space or the digital elevation model is thinned into a limited and regular grid and a grid centroid, and the ROI is fused into a multi-view analysis process to form a sparse visibility matrix, which comprises 6 substeps, and the steps are shown in the figure 2, and are as follows:
step 1.1, visual field analysis of all candidate observation points: traversing candidate observation point set) Every point, get->Candidate observation points->Performing a visibility analysis to form a visible region +.>. The method comprises the following specific steps:
step 1.1.1, selecting an observation pointFor observation point->Performing visibility analysis to obtain visibility grid area +.>Storing the file as a grid file;
step 1.1.2, calling GDAL grid-to-vector function GDAL FPolygonize (), converting the visibility grid region into a visibility vector graphic, denoted asStoring the file as a single vector shape file;
step 1.2, construction of the ROI visible region (denoted as). The method comprises the following specific steps:
step 1.2.1, vector graphics to visibilityWith region of ROI ()>) Performing spatial correlation analysis to obtain ∈>Visible ROI area->And index value of candidate observation point +.>Fill in to visibility vector graphics +.>In the field; if the target pattern spot is not considered->That is, the universe is the maximum coverage analysis application scene of the visibility of the target image spots or the topography multiple views, then；
Step 1.2.2, get the firstCandidate observation points->And repeatedly executing the steps 2.1 to 2.2 until the visibility analysis of all candidate observation points and the construction of the ROI visible area thereof are completed.
Step 1.3, meshing expression of VSN construction areas. The three-dimensional geographic space or the digital elevation model of the VSN construction area is thinned into a limited and regular grid and a grid centroid, and the specific steps are as follows:
step 1.3.1 EiAccording to the mesh sizeGenerating regular grid surface pattern spots covering an observation area(invoking the fishing net generating function ogr.geometry (ogr.wkbpolygon)) of the OGR package;
step 1.3.2, realizing sparse expression of a research area, and mapping spots on a regular grid surfaceExtracting centroid points to obtain a regular grid centroid point set +.>Storing the vector file as a vector file;
step 1.4, geometrically associating the ROI visible region with the centroid point of the regular grid to form a sparse grid centroid point set. Traversing the ROI visible region of each candidate observation point, and spatially associating the ROI visible region with the centroid point of the regular grid to complete multi-view association relation construction and form a sparse grid centroid point set; the method comprises the following specific steps:
step 1.4.1, reading candidate observation pointsIs of the region of interest (ROI)It is combined with administrative district rule grid centroid point +.>Spatial correlation analysis (calling GDAL vector image spot Spatial correlation Joint function), indexing candidate observation points +.>Information is stored to->In the associated candidate observation point field in the layer, completing spatial association;
step 1.4.2, traversing the visible areas of the target spots of the candidate observation points one by one, repeatedly executing the step 2.4.1 until all the candidate observation points finish the geometric space association of the ROI visible areas and the centroid points of the regular grids, and forming a sparse grid centroid point set for completely recording the information of the ROI visible areas；
Step 1.5, constructing a visual area cost matrix: sparse grid centroid point set of complete record ROI visible area information obtained according to step 2.4The cost matrix between the camera and the VSN construction area is established, and the specific steps are as follows:
step 1.5.1, setThere is +.>Strip record (s)/(s)>The total number of records in the layer is +.>Initializing a sparse multi-view matrix +.>Matrix dimension->Initialization values are all assigned 0;
step 1.5.2 traversalEach record acquires the attribute values of the ID field of the centroid point and the field of the associated candidate observation point, if the attribute value of the field of the associated candidate observation point is not null, the record indicates that the point can be covered by the candidate observation point, and the +.>The corresponding (centroid point ID, candidate observation point index) element value is 1; if the attribute value of the field of the associated candidate observation point is null, the sparse centroid point is not covered by the candidate observation point;
step 1.5.3, it is worth noting that ifThe corresponding (centroid point ID, candidate observation point index) element is set to 1 a plurality of times and still recorded as 1, and the physical meaning is that the same region is recorded as 1 only after being visible by a plurality of candidate points, so that the staggered condition of multi-view overlapping is simplified. Up to->Recording all traversals to complete the sparse multi-view matrix +.>And (5) assigning and updating.
And step 1.6, designing a visual area cost matrix operator. In order to improve the evolutionary computing efficiency of the MCLP-VSN model and realize the rapid statistical computation of the objective function under different layout schemes, a single candidate observation point visual field statistical operator and a multi-candidate observation point visual field statistical operator are required to be designed. Single candidate observation point
The visual field statistics operator of (2) is shown in formula (17):
in the formula (17)Representing the visual field result of single candidate observation point, +.>The sparse centroid point index is represented,。
multiple candidate observation pointsThe visual field statistics operator of (2) is shown in formula (18):
in the formula (18)Representing the visual field result of multiple candidate observation points, +.>Is a sequence of a plurality of candidate observation points. The operator is mainly applied to evaluating and optimizing the multi-view visual fields of various bodies in the algorithm population. The operator is mainly applied to evaluating and optimizing the multi-view visual fields of various bodies in the algorithm population.
In step 2: the MCLP-VSN model design, the spatial optimization problem is typically composed of 3 parts: objective function, constraint condition and decision algorithm, wherein step 2 describes the spatial combination addressing objective function of the vision sensing network, establishes constraint conditions such as series capacity and the like, and step 3 constructs an optimization decision algorithm oriented to the MCLP-VSN model. The method comprises the following specific steps:
step 2.1, defining an objective function of an MCLP-VSN model, wherein the model used in the method is a maximum coverage model, namely the number of newly built cameras is given) On the premise of maximizing the coverage of the ROI patch within the VSN build area. Computational aspect, equivalent to: in candidate observation point set->Is selected from the total amount->The construction sequence of the individual is such that in a given sparse multi-view matrix +.>The number of middle coverage sparse centroid points is the largest, and the MCLP-VSN model objective function can be described as:
in the method, in the process of the invention,representing sparse centroid points->The importance weight, the ROI importance level in the patent of the invention is consistent, can be set as 1; />Representing sparse centroid points->Whether or not it is to be covered or not,
step 2.2, defining constraints of the MCLP-VSN model, can be described as:
in the method, in the process of the invention,the representation may cover sparse centroid points +.>Is to be noted, wherein ∈is a set of candidate observation points (simply referred to as available candidate observation points), wherein ∈>Refers to the actual distance of the sparse centroid point from the camera,/->Refers to the maximum viewing coverage radius of the camera,representing visibility analysis, +.>1 represents a visual; />Representing camera layout decision variables;
constraints (20) and (21) are implemented: traversing each sparse centroid pointConstructing a sparse centroid point>Candidate observation point set->Will->Adding all candidate observation points to obtain a sparse centroid point +.>The actual construction results of the candidate observation point set are discussed in terms of cases. (1) If sparse centroid point->The actual construction result of the candidate observation point set is more than or equal to 1, at the moment, the sparse centroid point +.>Is covered, at this time will +.>Set to 1; (2) If sparse centroid point->The actual construction result of the candidate observation point set is more than or equal to 0, which indicates sparse centroid point +.>Is not covered by the service, at this point will +.>Set to 0. Constraint (22) defines that the total amount of newly deployed VSNs is limited to p. Constraints (23) and (24) limit whether the overlay variable, the decision variable, is binary.
In step 3: MCLP-VSN decision algorithm. The method improves and designs integer coding operators, capacity constraint patching operators, multipoint crossover mutation operators and the like to form a vision sensing network space combination site selection genetic optimization algorithm. The method comprises the following specific steps:
and 3.1, initializing a population. Initializing a first generation based on an integer encoding operator and a capacity constraint patching operatorGenerating an initial population->；
Step 3.2, pair ISubstitution population->Iteratively executing the steps 3.2.1-3.2.5 until a stopping criterion is met;
step 3.2.1, calculatingTarget function adaptation value of each individual: based on the maximum space coverage objective function of the visual area cost matrix, the computing operator invokes the visual area cost matrix operator in the step 1.6 of the sparse multi-visual area analysis algorithm, and performs individual sequencing according to the adaptive value;
step 3.2.2, elite retention strategy: will beThe individuals with the highest fitness value are added with +.>In the population, the retention of elite individuals is realized;
step 3.2.3, selecting a male parent individual based on a probability selection operator: according to the principle of random selection of probability of roulette wheelTwo father individuals are selected as crossed variation to form new offspring individuals;
step 3.2.4, genetic manipulation: genetic operation is carried out on two male parent individuals based on a multipoint crossover operator and a mutation operator, capacity adjustment operation is carried out by utilizing a capacity constraint repair operator, the total number of cameras represented by the individuals is equal to a capacity constraint condition, and the generated individuals are added；
Step 3.2.5, population scale discrimination: if it isThe number of individuals reaching the population size +.>And go to step 3.2.2, otherwise, loop step 3.2.3 until +.>The number of individuals reached the standard.
And 3.3, stopping population evolution and outputting the optimal individuals. And (3) through iterative evolution in the step 3.2, when the optimal fitness reaches a preset termination threshold value or the iteration number reaches a preset termination iteration algebra, the population evolution is terminated. The parameter setting of the optimization algorithm of the specific experiment is shown in table 1, and the optimization algorithm is written based on the python language.
Table 1 algorithm general parameter settings
And 4, forming a spatial combination site selection result of the visual sensing network, as shown in fig. 7 and 8.
First, a high-efficiency verification experiment of a sparse multi-view analysis method is carried out. The experiment forms four groups of comparison experiments (a 1-a4, b1-b4, c1-c4 and d1-d4 in table 2) by setting different candidate site numbers, capacity scales, visible radius of a camera and iteration frequency parameters of an optimization algorithm, and aims to illustrate the time efficiency superiority of a VSN space layout optimization algorithm based on sparse multi-view analysis, and the time efficiency analysis comparison of a classical visibility analysis algorithm and a sparse multi-view analysis algorithm is shown in table 2. As shown in table 2, in the four experiments a, b, c, d, as the visible radius of the camera increases, the single-generation adaptability time of the classical visual analysis algorithm also increases, and for this reason, because the VSN space layout optimization algorithm based on the classical visual analysis needs to repeatedly perform the steps of selecting points in candidate points, extracting elevation values to points, performing visual analysis, performing grid turn-plane, performing intersection analysis, performing calculation area and other space queries and performing space analysis during each generation of calculation, and these repeated space calculation operations greatly increase the calculation time of the VSN space layout optimization algorithm. In the data preprocessing link, the VSN space layout optimization algorithm based on sparse multi-view analysis only needs to complete one-time sparse multi-view analysis to form a space visual cost matrix, and the later iterative computation is completed based on the matrix, so that the single-generation fitness computation time is very low, and the time-consuming process of repeated space computation operation in the evolution search process is effectively avoided. In four groups of control experiments, the comprehensive time efficiency improvement rate of the VSN space layout optimization algorithm based on sparse multi-view analysis is 99.98%. In summary, the sparse multi-view analysis algorithm provided by the invention takes the space visual area cost matrix as a medium, and can be effectively applied to a VSN space layout optimization algorithm so as to improve the time efficiency of the algorithm and relieve the calculation explosion.
Table 2 comparison table of time efficiency of VSN space layout optimization algorithm based on different multi-view analysis algorithms. In four groups of control experiments, the same group controls three variable conditions, and one variable condition is regulated and changed to realize the observation of the control variable.
Secondly, under the conditions of the same monitoring radius and different capacities, the space layout optimization algorithm of the visual sensing network based on sparse multi-view analysis increases along with the increase of the capacity value, and the area occupation ratio of the target pattern spots covered by the calculated optimal layout scheme also increases. As the monitoring radius increases, so does the monitoring coverage area, as shown in fig. 7. However, the increase of the specific gravity tends to be limited, and is mainly limited by the limited mounting height of the camera or the existence of a monitoring shielding blind area in a part of unmonitored areas without candidate points.
Finally, calculating and spatially visualizing the site selection result for the further spatial combination of the visual sensing network: firstly, setting the density of a sparse grid as 100m, setting the capacity as 115, carrying out space combination site selection result calculation of a visual sensing network within the range of 1000m of monitoring radius, and under the current condition, monitoring the area of a cultivated land in a research area as 39.74 percent, wherein the coverage rate is higher; and secondly, carrying out space visual analysis on the space combination addressing result of the visual sensing network, wherein the addressing result is shown in figure 8. It can be seen that the point space distribution of the visual sensing network combined space distribution result distributed in the region is reasonable, and the coverage of the ROI monitoring area is improved.
In summary, the sparse multi-view analysis algorithm is designed, continuous topography is thinned into discrete points, and a specific observation range is fused into the multi-view analysis process to form a sparse visibility matrix, so that two problems that an optimization algorithm is difficult to combine with multi-view analysis in the space combination site selection problem of a visual sensing network and a traditional multi-view analysis nested optimization algorithm calculates scale explosion are solved, and the calculation population adaptation time of the optimization algorithm is greatly improved; and secondly, the optimization algorithm of the spatial combination site selection of the visual sensing network constructed by the invention estimates the spatial characteristic information of the point positions, improves the acquisition quality, finally obtains an ideal result by applying the spatial characteristic information to the actual layout research, can be widely used for the camera layout scheme formulation in the fields of urban fire control monitoring, forest area monitoring, high-speed monitoring and ecological protection, improves the monitoring coverage rate of target pattern spots, and has higher application value.
Claims (4)
1. The spatial combination site selection method based on the vision sensing network is characterized in that firstly, a sparse multi-view analysis algorithm is provided, the spatial coverage range of each camera in each vision sensing network is calculated and obtained, a spatial visual area cost matrix is constructed, and multi-view spatial analysis in a three-dimensional geographic space is reduced to be the numerical calculation of the visual area cost matrix; secondly, taking a visual area cost matrix as an input condition, and designing a visual sensing network maximum coverage site selection model; furthermore, an optimization algorithm for the MCLP-VSN model is provided, a vision sensing network space combination addressing scheme is solved, and finally, a vision sensing network space combination addressing result is formed, wherein the method comprises the following specific steps:
step 1, providing a sparse multi-view analysis algorithm, calculating to obtain the space coverage of each camera in each visual sensing network, constructing a space visual cost matrix, and reducing the multi-view space analysis dimension in a three-dimensional geographic space into a visual cost matrix numerical calculation;
step 2, designing a visual sense network maximum coverage site selection MCLP-VSN model: starting from the two aspects of Objective function and constraint condition, defining an Objective function MCLP-VSN Objective of a maximum coverage site selection model of the visual sensing network by taking a visual field cost matrix as an input condition, and establishing constraint conditions MCLP-VSN Constraints of the maximum coverage site selection model of the visual sensing network;
step 3, designing an optimization algorithm of a visual sense network-oriented maximum coverage site selection MCLP-VSN model: an optimization algorithm for a visual sense network maximum coverage site selection MCLP-VSN model is constructed, and a subset of layout schemes which meet the constraint condition MCLP-VSN Constraints of the visual sense network maximum coverage site selection model and have high fitness of a visual sense network maximum coverage site selection model Objective function MCLP-VSN Objective is searched and evolved from a problem set;
step 4, evaluating a spatial combination addressing scheme of the visual sensing network and performing spatial visual analysis: firstly, evaluating a spatial combination site selection scheme of a visual sense network; secondly, carrying out space visual analysis on the space combination site selection result of the visual sensing network; finally selecting an optimal layout scheme from the scheme subset;
the step 1 specifically comprises the following steps: the sparse multi-view analysis algorithm design is that a three-dimensional geographic space or a digital elevation model is thinned into a limited and regular grid and a grid centroid, and the ROI is fused into the multi-view analysis process to form a sparse visibility matrix:
step 1.1, visual field analysis of all candidate observation points: traversing candidate observation point set view points Every point takes the view of the j-th candidate observation point points (j) Performing visibility analysis to form a visible region view (view points (j))；
Step 1.2, constructing the ROI visible region, which is marked as target ROI ∩viewshed.view points (j))；
Step 1.3, meshing expression of a VSN construction area, namely thinning a three-dimensional geographic space or a digital elevation model of the VSN construction area into a limited and regular grid and a grid centroid;
step 1.4, the ROI visible region is associated with the geometric space of the centroid point of the regular grid to form a sparse grid centroid point set, the ROI visible region of each candidate observation point is traversed and is associated with the geometric space of the centroid point of the regular grid to complete the construction of multi-view association relationship, and the sparse grid centroid point set points are formed (sparse_fishnet) ；
Step 1.5, constructing a visual area cost matrix: sparse grid centroid point set points of complete record of ROI visible area information according to step 1.4 (sparse_fishnet) Establishing a cost matrix between the camera and the VSN construction area sparseVisible ；
Step 1.6, designing a visual area cost matrix operator: in order to improve the evolutionary computing efficiency of the MCLP-VSN model and realize the rapid statistical computation of the objective function under different layout schemes, a single candidate observation point visual field statistical operator and a multi-candidate observation point visual field statistical operator are required to be designed, and the visual field statistical operator of a single candidate observation point j is shown in a formula (1):
in the formula (1), V (J) represents a single candidate observation point visual field result, I represents a sparse centroid point index, J represents a candidate observation point index, J represents the total number of candidate observation points, and I represents the sparse centroid point number;
multiple candidate observation points (j) 1 ,j 2 ,j 3 ,…,j k ) The visual field statistics operator of (2) is as shown in formula (2):
v (j) in formula (2) 1 ,j 2 ,j 3 ,…,j k ) Representing the visual field result of multiple candidate observation points, (j) 1 ,j 2 ,j 3 ,…,j k ) For a plurality of candidate observation point sequences, the operator is applied to evaluating the multi-view visual fields of each body in the optimization algorithm population;
the step 2 specifically comprises the following steps:
step 2.1, defining an objective function of an MCLP-VSN model, wherein the MCLP-VSN model is a maximum coverage model, namely, on the premise of giving the number p of newly built cameras, the ROI map spots in a VSN construction area are covered to the maximum extent, and the calculation level is equivalent to: view is assembled at candidate observation points points The total p construction sequences are selected, so that matrix x of a given sparse multi-view matrix sparseVisible The number of middle coverage sparse centroid points is the largest,the MCLP-VSN model objective function is described as:
wherein a is i Representing important weight of sparse centroid point i, and setting the ROI important level as 1; y is i Indicating whether the sparse centroid point i is covered or not,
step 2.2, defining constraints of the MCLP-VSN model, can be described as:
N i ＝{j|d ij ≤S ∩ LOS(i，j)＝1} (5)
wherein N is i Representing a series of candidate observation point sets that may cover a sparse centroid point i, where d ij Referring to the actual distance of the sparse centroid point from the camera, S refers to the maximum observed coverage radius of the camera, LOS (i, j) represents the visibility analysis, LOS (i, j) =matrix sparseVisible [i，j]1 represents a visual; x is x j Representing camera layout decision variables, p-tableThe number of newly built cameras is shown, and J represents the total number of candidate observation points;
constraint (4) and (5) are implemented: traversing each sparse centroid point i to construct a candidate observation point set N capable of covering the sparse centroid point i i Will N i Adding all candidate observation points in the set to obtain an actual construction result of a candidate observation point set of a sparse centroid point i, and if the actual construction result of the candidate observation point set of the sparse centroid point i is more than or equal to 1, covering the sparse centroid point i, and covering y at the moment i Set to 1; if the actual construction result of the candidate observation point set of the sparse centroid point i is more than or equal to 0, the sparse centroid point i is not covered by the service, and y is determined at the moment i Set to 0; constraint (6) defines that the total amount of newly laid out VSNs is limited to p; constraints (7) and (8) limit whether the overlay variable, the decision variable, is binary.
2. The spatial combination addressing method based on the visual sense network as set forth in claim 1, wherein the step 3 specifically includes the steps of:
step 3.1, initializing a population: initializing a first generation t=0 based on an integer coding operator and a capacity constraint patching operator to generate an initial population P 0 ；
Step 3.2, for the t generation population P t Until a stopping criterion is met;
and 3.3, stopping population evolution and outputting optimal individuals, and stopping population evolution when the optimal fitness reaches a preset termination threshold value or the iteration number reaches a preset termination iteration algebra through iterative evolution in the step 3.2.
3. The spatial combination addressing method based on the visual sense network according to claim 2, wherein the step 3.2 specifically comprises the following steps:
step 3.2.1, calculatingP t Target function adaptation value of each individual: based on the maximum space coverage objective function of the visual area cost matrix, the computing operator invokes the visual area cost matrix operator in the step 1.6 of the sparse multi-visual area analysis algorithm, and performs individual sequencing according to the adaptive value;
step 3.2.2, elite retention strategy: will P t Adding P to the individual with highest fitness value t+1 In the population, the retention of elite individuals is realized;
step 3.2.3, selecting a male parent individual based on a probability selection operator: according to the principle of random selection of roulette wheel probability, at P t Two father individuals are selected as crossed variation to form new offspring individuals;
step 3.2.4, genetic manipulation: genetic operation is carried out on two male parent individuals based on a multipoint crossover operator and a mutation operator, capacity adjustment operation is carried out by utilizing a capacity constraint repair operator, the total number of cameras represented by the individuals is equal to a capacity constraint condition, and the generated individuals are added into P t+1 ；
Step 3.2.5, population scale discrimination: if P t+1 If the number of individuals reaches the population scale, t is greater than t+1 and the step is transferred to step 3.2.2, otherwise, the step 3.2.3 is circularly executed until P is reached t+1 The number of individuals reached the standard.
4. The spatial combination addressing method based on a visual sense network according to claim 1, wherein in step 4, the spatial combination addressing scheme of the visual sense network comprises: integer coding operators, capacity constraint patching operators, selection operators, crossover operators and mutation operators;
integer encoding operator: the t generation population P (t) is composed of pop size Chromosome constitution, wherein the first generation population chromosomes are generated by adopting a method of 'first random and then adjustment', and then each generation of chromosomes are generated by adopting a genetic operator operation mode; generation of the nth chromosome in the t th generationThe basic steps of (a) are as follows: in the case of the first generation population, a random selection method is adopted,i.e. < ->Chromosome, length is candidate point number J, p gene bit number is 1, J-p gene bit number is 0, meeting capacity constraint condition; otherwise, operating through a selection operator, a crossover operator, a mutation operator and a capacity constraint repair operator to generate a new chromosome;
capacity constraint repair operator: the capacity constraint repair operator can realize facility configuration capacity limitation, belongs to a repair mechanism of a chromosome, and is used for solving the problem that after crossing or mutation, the number of the selected monitoring points is out of limit and the capacity constraint condition is not satisfied; the method comprises 2 steps of difference degree calculation and constraint repair, namely calculating the number Num of gene bit values of 1 in the current chromosome by using the set capacity constraint quantity p, calculating the difference value between p and Num, and performing constraint repair by taking the difference value as a reference;
selecting an operator: selecting a certain number of 'good' individuals and 'bad' individuals from the population, performing genetic operation, selecting a random traversal selection operator to screen the individuals, and placing the individuals on a wheel disc and rotating the individuals once by introducing a rotator with 4 pointers which are uniformly distributed;
crossover operator: the crossover operator uses multi-point crossover logic as follows: randomly selecting two chromosomes from a population by a roulette method, if the crossing probability condition is met, randomly generating a crossing position sequence by the selected two individuals according to the set crossing number, executing crossing, leveling the gene value at the crossing position by using a capacity constraint repair operator, realizing that the sum of the gene values at the crossing position is equal to the sum of genes at the corresponding position before crossing, and performing crossing operation, otherwise, not performing any operation by the two individuals;
mutation operator: the mutation operator uses polynomial mutation logic with the following rules: traversing the whole population, if the mutation probability is met, performing mutation operation by the individual according to the set mutation quantity, and calculating a new value for the mutation gene position through polynomial mutation; leveling the gene values at the crossing positions by using a capacity constraint repair operator, and realizing that the sum of the gene values at the mutation positions is equal to the sum of the genes at the corresponding positions before mutation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311159848.4A CN116894400B (en) | 2023-09-11 | 2023-09-11 | Spatial combination site selection method based on visual sense network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311159848.4A CN116894400B (en) | 2023-09-11 | 2023-09-11 | Spatial combination site selection method based on visual sense network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116894400A CN116894400A (en) | 2023-10-17 |
| CN116894400B true CN116894400B (en) | 2023-12-12 |
Family
ID=88315246
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311159848.4A Active CN116894400B (en) | 2023-09-11 | 2023-09-11 | Spatial combination site selection method based on visual sense network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116894400B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268240A (en) * | 2014-09-29 | 2015-01-07 | 南京国图信息产业股份有限公司 | Implementation method for point feature cartographic label placement based on cartographic related group ant colony algorithm |
| CN110457997A (en) * | 2019-06-27 | 2019-11-15 | 中国地质大学（武汉） | A kind of star based on improved adaptive GA-IAGA ground space-time maximal cover problem solving method |
| AU2020102550A4 (en) * | 2020-10-14 | 2020-12-24 | Wuhan University | The location optimization of 5G base stations (BSs) in urban outdoor area that considers signal blocking effect |
| CN114329854A (en) * | 2020-11-06 | 2022-04-12 | 北京航空航天大学 | Two-dimensional space vision sensor layout optimization method based on multi-objective constraint |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016133924A1 (en) * | 2015-02-18 | 2016-08-25 | Siemens Healthcare Diagnostics Inc. | Image-based tube slot circle detection for a vision system |
-
2023
- 2023-09-11 CN CN202311159848.4A patent/CN116894400B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268240A (en) * | 2014-09-29 | 2015-01-07 | 南京国图信息产业股份有限公司 | Implementation method for point feature cartographic label placement based on cartographic related group ant colony algorithm |
| CN110457997A (en) * | 2019-06-27 | 2019-11-15 | 中国地质大学（武汉） | A kind of star based on improved adaptive GA-IAGA ground space-time maximal cover problem solving method |
| AU2020102550A4 (en) * | 2020-10-14 | 2020-12-24 | Wuhan University | The location optimization of 5G base stations (BSs) in urban outdoor area that considers signal blocking effect |
| CN114329854A (en) * | 2020-11-06 | 2022-04-12 | 北京航空航天大学 | Two-dimensional space vision sensor layout optimization method based on multi-objective constraint |
Non-Patent Citations (3)
| Title |
|---|
| An Integrated AHP-GIS-MCLP Method to Locate Bank Branches;Sajjad Allahi 等;《Proceedings of the 2015 Industrial and Systems Engineering Research Conference》;1-11 * |
| 卫星城镇避难场所层级选址与布局优化——以河北燕郊为例;李海君 等;《科学技术与工程》;第17卷(第18期);96-104 * |
| 战区装备保障点动态选址仿真问题研究;董鹏 等;《计算机仿真》;第29卷(第9期);32-36 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116894400A (en) | 2023-10-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Saeidian et al. | Evaluation and comparison of Genetic Algorithm and Bees Algorithm for location–allocation of earthquake relief centers | |
| Huang et al. | A land-use spatial optimum allocation model coupling a multi-agent system with the shuffled frog leaping algorithm | |
| Shaygan et al. | Spatial multi-objective optimization approach for land use allocation using NSGA-II | |
| CN108520557B (en) | Massive building drawing method with graphic and image fusion | |
| CN107895031B (en) | Self-adaptive dynamic scheduling method of building information model in three-dimensional digital city scene | |
| CN111210517B (en) | Multi-grid terrain generation method based on neural network | |
| CN106780262A (en) | A kind of same bit pattern for considering urban road network constraint finds method and device | |
| CN110415346A (en) | The method for carrying out soil erosion simulation using the three-dimensional cellular automaton of object-oriented | |
| CN112528508A (en) | Electromagnetic visualization method and device | |
| CN115272637B (en) | Large-area-oriented three-dimensional virtual ecological environment visual integration and optimization system | |
| Zhang et al. | Basic farmland zoning and protection under spatial constraints with a particle swarm optimisation multiobjective decision model: a case study of Yicheng, China | |
| CN110569616B (en) | SOM-based building multi-objective optimization design decision support method | |
| CN110084865A (en) | A kind of method of discrete point classification weighted fitting regular grid | |
| CN109872390A (en) | A kind of method for organizing of magnanimity threedimensional model | |
| CN108495252A (en) | Indoor positioning network element Optimal Deployment Method based on genetic algorithm and simulated annealing | |
| Chandramouli et al. | Spatial change optimization | |
| CN116894400B (en) | Spatial combination site selection method based on visual sense network | |
| CN112488413A (en) | AWA-DRCN-based population spatialization method | |
| Demetriou et al. | LandParcelS: A module for automated land partitioning | |
| Heyns | A multi-objective approach towards geospatial facility location | |
| CN108304576A (en) | A kind of the Ecological Control line demarcation method and device of intelligent interactive | |
| Yue et al. | A Genetic Optimization Method for Spatial Layout of Cameras in Video Sensor Networks | |
| Veselý | Building massing generation using GAN trained on Dutch 3D city models | |
| CN117523063B (en) | Rendering method of high-performance ray tracing simulation based on three-dimensional space | |
| Lazoglou et al. | Investigating the use of a modified NSGA-II solution for land-use planning in mediterranean islands |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |