Point feature label placement method based on maximum entropy principle

Abstract

The application discloses a point element annotation configuration method based on a maximum entropy principle. The essence of map annotation configuration is a combination optimization problem of competition on a limited map space. Maximum entropy is introduced as an optimization constraint. Each annotation configuration is expressed as a probability distribution. The maximum entropy distribution is optimal and unique. In order to construct a probability distribution mathematical model of the annotation configuration, a probability calculation method of geometric information entropy is introduced. The method is related to the annotation position, size, shape and spatial relationship with surrounding annotations. Finally, the geometric information entropy is used as a fitness function. A genetic algorithm is applied to solve the optimal annotation configuration. The number of annotation conflicts and coverings of the optimal individual of each generation of genetic evolution gradually decreases. When the number of conflicts and coverings is the least, the optimal annotation configuration can be continuously optimized to the optimal annotation configuration with the uniform annotation configuration distribution as the target. The multiple optimal solution condition is avoided. The uniqueness requirement of solving the optimal annotation configuration is met.

Description

Technical Field

[0001] This invention belongs to the field of map annotation configuration technology, specifically, it relates to a point feature annotation configuration method based on the principle of maximum entropy. Background Technology

[0002] Map annotations are an integral part of map language, playing a crucial supplementary role to map symbols. Through map annotations, readers can not only identify important symbols on the map but also obtain qualitative and quantitative information about each symbol. This makes map annotations an important tool for information transmission between cartographers and readers; it can be said that maps only become readable and translatable with annotations.

[0003] The rationality of map annotation configuration significantly impacts the efficiency of map information transmission, and the efficiency of configuration directly affects map production efficiency. Statistics show that configuring map annotations accounts for approximately 50% or more of the entire cartographic production time. Therefore, the efficiency of annotation configuration severely affects the speed of map production and publication. Map annotation configuration can be done manually or automatically by computer, and the more efficient and lower-cost automatic computer-configured map annotation is gradually replacing traditional manual annotation methods.

[0004] Depending on the type of map feature, automatic map annotation configuration can be divided into automatic annotation configuration for point, line, and polygon features, with automatic annotation configuration for point features being the most general. Automatic annotation configuration for point features first requires formalizing map annotation configuration rules for computer modeling processing.

[0005] There are four basic rules for the configuration of point feature annotations: (1) There should be no conflict between the annotations of each point feature (overlapping annotations is called conflict); (2) Annotations should not cover other point features (overlapping annotations with features is called overlay); (3) The visual relationship between the annotations of a point feature and the feature should be clear and unambiguous; (4) Different candidate positions for annotations of point features have different priorities. Under the above rules, the position with the highest priority should be selected. Controlling conflicts and overlays is the most commonly used rule. Existing research has shown that solving the optimal annotation configuration of point features is an NP (unsolved problem) problem, that is, a complex problem that cannot be solved accurately in polynomial time, but can only be solved approximately. Moreover, the complexity of solving the problem increases sharply when the problem size increases. To this end, many studies have adopted heuristic optimization algorithms, such as genetic algorithms, ant colony algorithms, simulated annealing, and gradient descent. These algorithms significantly improve the speed of problem solving while ensuring the quality of annotation configuration.

[0006] The optimal point feature annotation configuration problem is solved by using an optimization algorithm, and a reasonable and effective objective function needs to be established. The commonly used objective function in current research is to minimize the number of annotation conflicts and the number of caps. The specific method is to count the number of annotation conflicts and the number of caps for each point feature one by one, and then add the statistical results of each point feature. When the number of conflicts and the number of caps cannot be further reduced (ideally, the number of conflicts and the number of caps is zero), it is considered that the annotation configuration quality is the highest. It is easy to know that there are usually more than one annotation configuration without conflicts and caps, which leads to the situation that there are multiple "optimal solutions" for these commonly used objective functions, which is not consistent with the "uniqueness" of solving the optimal annotation configuration. Moreover, multiple "optimal solutions" have different distribution situations, such as the three optimal solutions shown in Figures 15 to 17 The annotation configuration situation corresponding to the three optimal solutions has local sparseness, local over-densification, and overall uneven distribution, which leads to the fact that the "optimal solution" is not the real optimal solution. SUMMARY

[0007] To solve the problem of multiple "optimal solutions" of the existing objective function, a point feature annotation configuration method based on the maximum entropy principle is proposed, which takes the geometric information entropy as the objective function, introduces the maximum entropy principle as the optimization constraint, and applies the genetic algorithm to solve the probability distribution with the maximum geometric information entropy. The number of conflicts and the number of caps of the optimal individual of each generation gradually decrease, and the annotation configuration can be further optimized until the optimal solution is obtained when the number of conflicts and the number of caps are the smallest. The method can further obtain a higher quality configuration when the number of conflicts and the number of caps cannot be minimized, and avoid the situation of multiple "optimal solutions", which meets the "uniqueness" requirement of solving the optimal annotation configuration.

[0008] The application is implemented by using the following technical solutions:

[0009] A point feature annotation configuration method based on the maximum entropy principle is proposed, which includes:

[0010] A probability distribution model representing the proportion of annotation contribution to map information is constructed;

[0011] A candidate annotation set of point features is constructed, and a plurality of annotation configurations are obtained according to the candidate annotation set;

[0012] The constructed probability distribution model is taken as the objective function, the maximum entropy is taken as the optimization constraint of the annotation configuration, and the genetic algorithm is applied to solve the optimal annotation configuration.

[0013] In some embodiments of the application, the probability distribution model representing the proportion of annotation contribution to map information is constructed, specifically including: generating Voronoi regions of each annotation; taking the area of the Voronoi region of each annotation as the contribution to the map information; and taking the area proportion of all annotations as the probability distribution of the annotation configuration.

[0014] In some embodiments of the present application, the note Voronoi region is generated, specifically including: generating the minimum circumscribed rectangle of the note text, converting it into a planar feature; representing the minimum circumscribed rectangle of the note text with a point feature; generating the Voronoi region of all point features; and fusing the Voronoi regions of all point features belonging to the same note to obtain the note Voronoi region.

[0015] In some embodiments of the present application, the maximum entropy principle is introduced as a note configuration optimization constraint, specifically including: setting the contribution ratio of the i-th placed note to the map information as p(x i ); constructing a probability distribution P(X) with all note contribution ratios p(x1), p(x2), …, p(x n ); setting a probability distribution constraint condition; and solving the probability distribution with maximum entropy based on the constraint condition, which corresponds to the optimal note configuration quality.

[0016] In some embodiments of the present application, a candidate note set of point features is constructed, including: obtaining point feature data; setting the label text font, size, position, and candidate position according to the mapping rules; wherein the candidate position is set as eight labels generated in the right upper, upper, left upper, left, lower left, lower, right lower, and right directions of the point feature, each candidate position has different priority, and the priority is represented by an index, the smaller the index value, the higher the priority; selecting the fields to be expressed to generate the labels of the point features; and converting the labels of the point features to obtain the candidate note set.

[0017] In some embodiments of the present application, in the application of the genetic algorithm to solve the optimal note configuration, the steps of determining the genetic algorithm coding and decoding method are included, specifically including: setting each chromosome in the population to represent a note configuration, wherein the length of the chromosome is the same as the number of point features; determining the coding of each gene on the chromosome based on the index of the candidate note; and determining the candidate note of each point feature according to the coding of each gene of the chromosome.

[0018] In some embodiments of the present application, in the application of the genetic algorithm to solve the optimal note configuration, the step of generating an initial population is included, specifically including: setting the size of the initial population according to the number of point features; and generating the initial population by taking the index with the highest priority for the coding of the genes corresponding to all free note point features, and randomly selecting the candidate note index for the rest of the gene coding.

[0019] In some embodiments of the present application, in the application of the genetic algorithm to solve the optimal note configuration, the step of calculating the fitness is included, specifically including: setting the fitness function as the geometric information entropy; and calculating the area ratio of the note Voronoi region and applying the information entropy formula to calculate the geometric information entropy.

[0020] In some embodiments of the present application, in the application of genetic algorithm to solve the optimal mark configuration, the step of executing genetic operators is included, specifically comprising: the selection operator adopts the tournament selection strategy, and selects the individual with the highest geometric information entropy in each competition; the crossover operator adopts the single-point crossover method, randomly selects a gene position on the chromosome, and exchanges all genes after the gene position in the parent string to generate a new individual; the mutation operator adopts the conflict and grommet gene position mutation, and generates a candidate mark index for the conflict and grommet mark randomly to replace the original code.

[0021] In some embodiments of the present application, in the application of genetic algorithm to solve the optimal mark configuration, the step of selecting the optimal individual and decoding is included, specifically comprising: selecting the individual with the highest fitness in the last generation evolution result after the termination of the genetic algorithm as the optimal individual; decoding the chromosome gene of the optimal individual; and drawing the optimal mark configuration result graph based on the decoding result.

[0022] Compared with the prior art, the advantages and positive effects of the present application are: in the point feature mark configuration method based on the maximum entropy principle proposed by the present application, it is considered that the essence of map mark configuration is a combination optimization problem of competition in a limited map space, the maximum entropy is introduced as an optimization constraint, each mark configuration is expressed as a probability distribution, and the distribution with the maximum entropy is optimal and unique; in order to construct the probability distribution mathematical model of the mark configuration, the probability calculation method of the geometric information entropy is introduced, which is related to the mark position, size, shape, and spatial relationship with surrounding marks, finally, the genetic algorithm is applied to solve the optimal mark configuration by taking the geometric information entropy as the fitness function, not only the number of mark conflicts and grommets of the optimal individual in each generation of genetic evolution gradually decreases, but also when the number of conflicts and grommets is the least, the mark configuration can be optimized to the optimal with the goal of more uniform mark configuration distribution, the multiple "optimal solutions" are avoided, and the "uniqueness" requirement of solving the optimal mark configuration is met.

[0023] Other features and advantages of the present application will become more apparent after reading the detailed description of the embodiments of the present application in combination with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0024] The accompanying drawings, which are part of the present application, serve to provide a further understanding of the present application, and the schematic embodiments and descriptions of the present application are used to explain the present application, but do not constitute an improper limitation on the present application. Obviously, the accompanying drawings in the following description are only some embodiments, and other drawings can be obtained from these drawings without creative labor for those skilled in the art.

[0025] Figure 1 The flow chart of the point feature mark configuration method based on the maximum entropy principle proposed by the present application;

[0026] Figure 2 Point-like element annotation candidate position diagram for an embodiment of the present application;

[0027] Figure 3 Point-like element annotation candidate set diagram for an embodiment of the present application;

[0028] Figure 4 Genetic algorithm flowchart for an embodiment of the present application;

[0029] Figure 5 Genetic algorithm encoding method diagram for an embodiment of the present application;

[0030] Figure 6 Annotation planar element Voronoi region generation diagram for an embodiment of the present application;

[0031] Figure 7 Genetic algorithm crossover operator diagram for an embodiment of the present application;

[0032] Figure 8 Genetic algorithm mutation operator diagram for an embodiment of the present application;

[0033] Figure 9 Point-like element annotation configuration diagram based on the maximum entropy principle for an embodiment of the present application;

[0034] Figure 10 Annotation conversion from planar elements to point-like elements diagram for an embodiment of the present application;

[0035] Figure 11 Voronoi region division of each point-like element of an annotation diagram for an embodiment of the present application;

[0036] Figure 12 Annotation Voronoi region division diagram for an embodiment of the present application;

[0037] Figure 13 Annotation configuration Voronoi region division diagram for an embodiment of the present application;

[0038] Figure 14 Voronoi region area size of each annotation, and annotation position, size, shape, and spatial relationship with surrounding annotations diagram for an embodiment of the present application;

[0039] Figure 15 One of the existing commonly used objective function exists multiple "optimal solution" distribution diagram;

[0040] Figure 16 Two of the existing commonly used objective function exists multiple "optimal solution" distribution diagram;

[0041] Figure 17Fig. 3 is a schematic diagram of a plurality of "optimal solution" distributions of an existing commonly used objective function.

[0042] It should be noted that the drawings and the written description are not intended to limit the scope of the inventive concept in any way, but are merely to illustrate the inventive concept to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0043] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments will be described clearly and completely below with reference to the drawings of the embodiments of the present application. The following embodiments are used to illustrate the present application, but are not used to limit the scope of the present application.

[0044] The design idea of the present application is to represent each note configuration as a probability distribution, introduce the maximum entropy principle as an optimization constraint, find the probability distribution with the maximum entropy, and find the optimal distribution, i.e., the optimal note configuration.

[0045] Based on the above idea, as shown in Fig. 1, the point feature note configuration method based on the maximum entropy principle is described by taking 229 point feature sets as an example, including the following steps. Figure 1

[0046] Step S11: Introduce the maximum entropy principle as an optimization constraint for note configuration.

[0047] The maximum entropy principle indicates that when predicting the probability distribution of a random event, all known conditions should be met, and no subjective assumption should be made for unknown conditions. In this case, the probability distribution is the most uniform, and the risk of prediction is the smallest. Among all probability distributions that meet the specific conditions, the probability distribution with the maximum entropy is unique. Among all possible probability distributions, the distribution with the maximum entropy is the best.

[0048] In the embodiments of the present application, the contribution ratio of the i-th placed note to the map information is p(x i ), and the contribution ratios of all notes to the map information p(x n ) constitute a probability distribution P(X). The probability distribution has two constraint conditions, which are ∑p(x i ) = 1 and p(x i )≠0. According to the maximum entropy principle, the note configuration optimization problem is to solve the probability distribution with the maximum entropy based on the constraint conditions, and the corresponding note configuration quality is optimal.

[0049] Step S12: Construct a probability distribution model representing the contribution ratio of the note to the map information.

[0050] ​The premise of solving the probability distribution of maximum entropy is to build a mathematical model of the probability distribution. In order to build a probability distribution representing the proportion of map information contributed by the annotation, the present application introduces geometric information entropy to measure map information, regards the annotation as a planar feature, and converts the planar feature into a point feature. The Voronoi area of each annotation is built by representing the point feature of each annotation, the contribution of each annotation to the map information is represented by the area of the Voronoi area of each annotation, and the probability distribution of the annotation allocation is composed of the proportion of the area of the Voronoi area of each annotation. According to the maximum entropy principle, the probability distribution with the maximum geometric information entropy is solved. When the annotation distribution is most uniform, the geometric information entropy is maximum, and the function of avoiding annotation conflict and overlay is implicitly realized.

[0051] Specifically includes:

[0052] (1) Building a Voronoi area to link annotations with surrounding annotations.

[0053] As shown in Figure 10 , for each annotation, a minimum bounding rectangle of the annotation text is generated (shown on the left of the figure), and the annotation text is converted into a planar feature. Because the Voronoi area cannot be generated based on the planar feature, the present application uses a point feature to represent the minimum bounding rectangle of the annotation text (shown on the right of the figure); then, for all point features, the Voronoi area of each point feature is generated, as shown in Figure 11 , 1 is a point feature of an example, and 2 is the Voronoi area of the point feature.

[0054] After generating the Voronoi area of each point feature, the Voronoi areas of all point features belonging to the same annotation need to be fused to obtain the Voronoi area of the annotation, so as to realize the construction of the Voronoi area of the planar feature, as shown in Figure 12 , the Voronoi areas of all annotations finally divided are shown in Figure 13 .

[0055] As can be seen, the area of the Voronoi area of each annotation is related to the position, size, shape and spatial relationship with surrounding annotations. If a certain annotation is close to surrounding annotations, the spatial distribution of the area of the annotation is compact, and their corresponding Voronoi area is small, as shown in Figure 14 , the left and right boxes respectively represent the cases of compact and loose spatial distribution of point features.

[0056] The originally dispersed points are irrelevant to each other, but after the Voronoi region is built for them, the points are connected with the surrounding points through the common edges of the Voronoi region, and based on this, the connection between each annotation and the surrounding annotations is indirectly connected through the common edges of the Voronoi region by converting the planar elements into point elements and fusing the Voronoi regions belonging to the same annotation according to the above method of the present application.

[0057] (2) The contribution of each annotation to the map information is represented by the area of the Voronoi region of the annotation.

[0058] (3) The probability distribution of the annotation configuration is constituted by the area proportion of the Voronoi region of all annotations.

[0059] Step S13: Building a candidate annotation set of point elements, and obtaining a plurality of annotation configurations according to the candidate annotation set.

[0060] The point element annotation automatic configuration method based on the maximum entropy principle is to select the one with the maximum entropy from various annotation configurations. Each annotation configuration is composed of a combination of annotations in the candidate annotation set of point elements. Therefore, building the candidate annotation set of point elements is the basis.

[0061] The parameters that need to be considered for generating the label include but are not limited to font, font size, font position (distance between the label and the point element), and candidate position scheme. The embodiment of the present application assumes that the annotation size and text content of all point elements are the same, the label font is set to Songti, the font size is 8 pounds, and the font position is 1 pound.

[0062] The candidate position scheme adopts an "8-annotation candidate position scheme", as shown in Figure 2 , that is, a label is generated in the right upper, upper, left upper, left, lower left, lower, lower right, and right direction of the point element, respectively, and corresponds to an index. The priority of labels in different directions is different, and the smaller the index value is, the higher the priority is. It should be noted that the generated label cannot exceed the drawing range. Finally, the candidate annotation set of the point element can be converted by the "label to annotation" function of ArcMap, as shown in Figure 3 .

[0063] As described above, each point element has 8 annotation candidate positions, and when combining annotations, a random annotation candidate position is selected for each point element, and these all randomly selected candidate positions constitute an annotation configuration. When the number of point elements is N, there are 8 N kinds of annotation configurations.

[0064] Step S14: Applying a genetic algorithm to solve the optimal annotation configuration.

[0065] To improve the efficiency of finding the optimal annotation configuration, this embodiment of the invention uses a genetic algorithm to quickly obtain an approximate optimal solution.

[0066] Genetic algorithms originated from computer simulations of biological systems and are a stochastic global search optimization method. The flowchart of the genetic algorithm in this embodiment is as follows: Figure 4 As shown, it specifically includes:

[0067] Step S141: Determine the encoding and decoding methods for the genetic algorithm. Encoding includes: setting the chromosome of an individual in the population to represent an annotation configuration; the length of the chromosome is the same as the number of point features, which is 229. The gene encoding on the chromosome is determined by the index of the candidate annotations, with values ​​ranging from 1 to 8. For example... Figure 5 As shown, gene 1 on the chromosome corresponds to dot element 1, and its candidate annotation direction is due left, with an index of 7. Therefore, the encoding of this gene is 7. The encoding method for the remaining genes is the same.

[0068] Decoding, the opposite of encoding, involves determining candidate annotations for each point element based on the encoding of each gene on the chromosome.

[0069] Step S142: Generate the initial population.

[0070] In this embodiment of the invention, the initial population size is set to 30. Based on the characteristics of the annotation configuration problem, an initial population is generated using a combination of random generation and elite retention strategies, as follows:

[0071] (1) On the chromosomes of each individual, the gene code corresponding to all free annotations (where no conflict or overlay occurs at any annotation position) is directly taken from the highest priority index, which is 1;

[0072] (2) Randomly select a candidate annotation index as the gene code corresponding to the non-free annotation point element on their chromosome.

[0073] Step S143: Calculate fitness.

[0074] After the initial population is generated, it evolves generation by generation according to the principles of survival of the fittest and natural selection, producing increasingly better approximate solutions. In each generation, individuals are selected based on their fitness in the problem domain, and genetic operators from natural genetics are used for combination, crossover, and mutation to generate a population representing a new set of solutions. The fitness function needs to be determined first.

[0075] This invention uses the geometric information entropy of annotation configuration as the fitness function to evaluate individuals. The calculation process is as described in step S12 above: First, the annotations are treated as planar features. Points on the contours of the planar features are extracted to approximate the boundaries of the features. Based on the point features, Voronoi regions of the annotation configuration are generated, such as... Figure 6as shown.

[0076] The following formula is then applied to calculate the geometric information entropy:

[0077]

[0078] where P i represents the proportion of the Voronoi region area of the i-th label, and N represents the number of labels.

[0079] To generate more accurate Voronoi regions, the contour of the areal feature can be densified to obtain more points (in the embodiment of the present application, the contour is densified at intervals of 10 meters), and then the Voronoi region of the densified points is further generated. Finally, the Voronoi regions belonging to the same areal feature are fused.

[0080] Sometimes, to compare the geometric information entropy of different feature numbers, the following formula is used for standardization:

[0081] H max = log N, (2)

[0082]

[0083] Step S144: Perform genetic operators.

[0084] The genetic operators include selection operators, crossover operators and mutation operators. The selection operators are used to select superior individuals from the initial population. According to the characteristics of the label configuration problem and the fitness function, the embodiment of the present application adopts a tournament selection strategy to select the individual with the highest geometric information entropy in each competition. The number of competitions is set to 30, and the number of participating individuals is set to 30% of the population size. The crossover operators combine the excellent features of the parent strings by exchanging two chromosomes, and pass the excellent features to the child strings, thereby generating new excellent individuals. The embodiment of the present application adopts a single-point crossover method. Specifically, the crossover probability is set to 0.8, and then each individual is assigned a value in the range of 0-1. Individuals with an assigned value not greater than 0.8 are selected to participate in the crossover and form pairs in order. If the number of individuals participating in the crossover is equal to 0, the first two individuals are directly used as parent strings. If the number of individuals participating in the crossover is odd, the last individual participating in the crossover is removed, and a certain gene position on the chromosome is further randomly selected. The parent strings are exchanged with each other at the gene position and all genes after the gene position, thereby generating new individuals, as shown in Figure 7 The mutation operators prevent the genetic algorithm from falling into a local optimal solution in the optimization process, and thus the individuals need to be mutated, i.e., part of the gene coding of the individual is randomly changed. For map labels, considering that label conflicts and overlaps are the main reasons for reducing the quality of map label configuration, the mutation operator adopts conflict and overlap gene position mutation, randomly generates a candidate label index for the conflict and overlap label genes, and replaces the original coding, as shown inFigure 8 are shown.

[0085] Step S145: select the optimal individual and decode.

[0086] After the genetic operators are executed, one generation of genetic evolution is completed. The result of each generation of evolution is used as the population of the next generation of evolution, and the iteration is repeated until the termination condition is reached. The termination condition can be that the highest geometric information entropy of each generation of individuals no longer changes or reaches a set number of evolution generations. The embodiment of the present application sets the termination condition to be 50 generations of evolution. After termination, the individual with the highest geometric information entropy in the last generation of evolution result is taken as the optimal result, and the corresponding annotation configuration is obtained by decoding the chromosome gene of the optimal individual and is plotted into a graph. Figure 9 The initial population and the optimal annotation configuration graph of part of the evolution generations are shown.

[0087] It should be pointed out that the above description is not a limitation of the present application, and the present application is also not limited to the above examples. Changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present application should also be within the scope of protection of the present application.

Claims

1. A point feature label placement method based on the maximum entropy principle, characterized in that, The method comprises the following steps: constructing a probability distribution model representing the proportion of map information contributed by each annotation; constructing a candidate annotation set for point-like elements, and obtaining a plurality of annotation configurations according to the candidate annotation set; taking the constructed probability distribution model as a target function, taking the maximum entropy as an optimization constraint of the annotation configuration, and applying a genetic algorithm to solve the optimal annotation configuration; constructing a probability distribution model representing the proportion of map information contributed by each annotation, specifically comprising: generating a Voronoi region for each annotation; representing the contribution of each annotation to map information by the area of the Voronoi region of the annotation; representing the probability distribution of the annotation configuration by the area proportion of the Voronoi region of all annotations; taking the maximum entropy principle as an optimization constraint of the annotation configuration, specifically comprising: The contribution of the first placed marker to the map information is ; and the contribution of the second placed marker to the map information is Contribution of all annotations to map information Constitute a probability distribution ; setting a probability distribution constraint condition; solving the probability distribution with the maximum entropy based on the constraint condition; in the step of applying the genetic algorithm to solve the optimal annotation configuration, comprising a step of calculating the fitness, specifically comprising: setting the fitness function as the geometric information entropy; calculating the area proportion of the Voronoi region of the annotation and applying the information entropy formula to calculate the geometric information entropy; in the step of applying the genetic algorithm to solve the optimal annotation configuration, comprising a step of executing a genetic operator, specifically comprising: adopting a tournament selection strategy for the selection operator, and selecting the individual with the highest geometric information entropy in each competition; adopting a single-point crossover method for the crossover operator, and randomly selecting a gene position on the chromosome to exchange all genes after the gene position in the parent string, thereby generating a new individual; adopting a conflict and pressing gene position mutation for the mutation operator, and generating a candidate annotation index for the conflict and pressing annotation gene to replace the original code.

2. The point feature label placement method based on the maximum entropy principle according to claim 1, wherein, generating a Voronoi region for the annotation, specifically comprising: generating a minimum circumscribed rectangle for the annotation text and converting it into a surface-like element; representing the minimum circumscribed rectangle of the annotation text by a point-like element; generating a Voronoi region for all point-like elements; fusing the Voronoi regions of all point-like elements belonging to the same annotation to obtain the Voronoi region of the annotation.

3. The point feature label placement method based on the maximum entropy principle according to claim 1, wherein, constructing a candidate annotation set for point-like elements, comprising: obtaining point-like element data; setting the annotation text font, font size, character position, and candidate position according to the drawing rules; wherein the candidate position is set as eight annotations generated in the right upper, upper, left upper, left, lower left, lower, right lower, and right directions of the point-like element, each candidate position has a different priority, and the priority is represented by an index, and the smaller the index value, the higher the priority; selecting a field to be expressed to generate the annotation of the point-like element; converting the annotation of the point-like element to obtain the candidate annotation set.

4. The point feature label placement method based on the maximum entropy principle according to claim 3, wherein, in the step of applying the genetic algorithm to solve the optimal annotation configuration, comprising a step of determining the genetic algorithm coding and decoding method, specifically comprising: setting each chromosome in the population to represent an annotation configuration, wherein the length of the chromosome is the same as the number of point-like elements; determining the coding of each gene on the chromosome based on the index of the candidate annotation; and determining the candidate annotation of each point-like element according to the coding of each gene of the chromosome.

5. The point feature label placement method based on the maximum entropy principle according to claim 3, wherein, in the step of applying the genetic algorithm to solve the optimal annotation configuration, comprising a step of generating an initial population, specifically comprising: setting the size of the initial population according to the number of point-like elements; The initial population is generated by taking the index of the gene encoding corresponding to the point element of all free notes as the highest priority index, and randomly selecting the candidate note index for the rest of the gene encoding.

6. The point feature label placement method based on the maximum entropy principle according to claim 3, wherein, In the application of the genetic algorithm to solve the optimal note configuration, the steps of selecting the optimal individual and decoding are included, specifically comprising: The individual with the highest fitness in the last generation evolution result after the termination of the genetic algorithm is taken as the optimal individual; The chromosome gene of the optimal individual is decoded; Based on the decoding result, the optimal note configuration result graph is drawn.

Images