JP2018100945A - Method for route search and route search system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for route search and a route search system which have an unknown environment for search for an image and can make a search with limited resources for the search.SOLUTION: The present invention relates to a method for route search by generating a route for image information based on the self-similarity. The method includes the steps of: controlling generation of a route by resources to be used; and setting a search range. Desired searches are sequentially made to fit the environments on images. The step of controlling the route generation controls growth and reduction of a search node and changes the rate of the reduction according to the remaining amount of the resources to be used.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a route search method and a search system for image data.

  It is known that the self-similarity of biological structures can efficiently fill a space. The branching structure of blood vessels and the inner wall of the intestine that require a large surface area with a small volume have a self-similar structure, and a technique for applying this property to a search problem has been proposed.

  For example, there is a space filling curve method (see, for example, Non-Patent Document 1). The space filling curve method is used as an approximate solution to the traveling salesman problem, and is characterized by being able to find a solution at high speed.

  A typical method for solving the search problem is a tree search algorithm, and it is necessary to generate a tree dynamically when the data is not clearly shown in the tree structure from the beginning. Since the search space becomes very large, it is desirable to use an evaluation function. However, the search environment such as the search space or the distribution of the target state is unknown, or the search area is described by a mathematical function. In a complicated environment such as when it is difficult, it is difficult to set an appropriate evaluation function, and there are cases where the solution cannot be reached due to search costs and resource constraints.

  On the other hand, Lindenmayer System (hereinafter referred to as L-system) is used as a technique for generating images of buildings and roads that are artifacts in geographic information, trees and rivers that are natural objects (for example, Non-Patent Document 2). See). L-system is a kind of formal grammar that can describe the growth process of plants. Although it is used to generate images of plants and the like, L-system as an image generation technique is based on a generation grammar, so that it is difficult to search for existing image components.

“Space-filling curves and fractals”, Zergan, H.C. Written and translated by Seiichiro Kamada, Springer Japan, published December 1998 Gabriela Ochoa, “An Introduction to Lindenmayer Systems“ Internet (URL: https://s10.lite.msu.edu/res/msu/botonl/b_online/e28_3/lsys.html)

  The above-described conventional technique has a problem that the search cannot be performed sufficiently when the search environment is unknown and the search resources are limited (hereinafter, referred to as a complex environment).

  The problem to be solved by the present invention is made in order to overcome the above problem, and based on a grammar (hereinafter, simply referred to as a rule) that can describe a structure showing self-similarity. To provide a route search method and a route search system capable of obtaining a solution even in a complex environment.

  Specifically, the route search according to the present invention generates a search route by assigning rules for growth and contraction based on the L-system, and performs a desired search on the image while evaluating the resource amount. It is adapted to the environment and performed sequentially.

A route search method according to the present invention is a route search method for generating a route based on self-similarity to image information,
A step of controlling generation of a route according to a resource to be used, and a step of setting a search range,
A desired search is sequentially performed according to the environment on the image.

Further, the step of controlling the generation of the path is a step of controlling growth and contraction,
The contraction probability is increased or decreased according to the remaining amount of resources to be used.

A route search system according to the present invention is a route search system that generates a route based on self-similarity to image information,
Means for controlling generation of a route according to resources used, and means for setting a search range;
A desired search is sequentially performed according to the environment on the image.

Further, the means for controlling the generation of the path is a means for controlling growth and contraction,
The contraction probability is increased or decreased according to the remaining amount of resources to be used.

According to the route search method and route search system according to the present invention, a route can be obtained even if the search environment is unknown.

It is a flowchart of the route search method which concerns on this invention. It is an example of the description of the production | generation and shrinkage | contraction rule which used slime mold as a model in the route search method which concerns on this invention. It is a schematic diagram which shows the result of the route search which concerns on the Example of this invention. It is an example of a route search when the moving means is changed from an automobile to a walk at an intermediate point. It is a schematic diagram which shows the result of the route search which concerns on the Example of this invention. It is an example of a route search when the height difference is used as a change in the resource amount. It is a schematic diagram which shows the result of the route search which concerns on the Example of this invention. It is an example of the route search in the case of a detour.

  The form for implementing this invention is as follows. In the route search method according to the present invention, the algorithm is described using a programming language, and necessary hardware is controlled on a computer in which an arithmetic processing device, a memory, an external storage device, a display device, and the like are electrically connected. Thus, the route search system according to the present invention is realized by the computer system in which a program that implements the route search method operates.

  A route search method according to the present invention will be described below with reference to the drawings. In the route search method according to the present invention, a method simulating slime mold is implemented. Self-similarity is observed in the growth of the slime mold, and the growth of slime mold is described as a generation rule using an arbitrary programming language based on the L-system.

  In the conventional L-system, only irreversible generation was expressed, but in order to search in a complex environment, a death rule that incorporates the concept of slime mold contraction is described.

A flowchart of the route search according to the present invention is shown in FIG.
When a program in which search information according to the present invention is installed is started, a configuration file is first read.

  Next, the parameters described in the slime mold production rule shown in FIG. 2 are input, and the search area is displayed on the display device. When the search start point and the target point are specified and determined by the user, a search node (slime mold node in FIG. 2) is set at the start point. The search node executes a search according to the generation rule. A search node is displayed on the display device according to the execution of the search.

  After searching for an area delimited by a predetermined interval, the position of the search node is compared with the target point. If the coordinates are not the same, the search and the display of the search node are repeatedly executed.

  On the other hand, when the position of the search node and the target point are the same coordinates, the search node other than the search node that generates the route by connecting the start point and the target point is shifted to the end state, and the search node is killed.

  For confirmation, search nodes that are in the end state and die are displayed on the display device, and the transition to the end state, death, and display are repeated until there are no unnecessary search nodes.

  Finally, only the search node that has reached the target point remains, and the generated route information (a set of search nodes), particularly the length of the set and the number of search steps, is displayed on the display device, and the program ends.

  The generation rule (FIG. 2) according to the implementation of the present invention is expressed by a set of variables, a set of constants, an initial state, and a tuple of replacement rules that change the set of variables, and is capable of self-similarity or recursive growth. It is a grammar to represent.

  As shown in FIG. 2, a set of variables is represented as 0, 1, 2, 3, d, and D. Here, let the death state be d and the destination D. Also, as a set of constants, generate forward (ie forward) character ^, generate right 90 ° direction], and generate left 90 ° direction character [, end of string $, any single character Represented by. The route search according to the present invention is expressed as the set of characters.

As the search control parameters, the death probability p _d , the generation promotion probability p _h , the branch probabilities p _a , p _b , and p _c (where p _a + p _b + p _c = 1) are used. Furthermore, a resource amount R used for search is defined. The resource amount is equal to or greater than the sum of the resources used in all search nodes. If the set of search nodes is N, the amount of resources used in each search node is r _n , and the sum is Σ, then R ≧ ΣNr _n .

The execution of the search is as follows.
When there is an obstacle in the search range next to the search node, the search node does not generate a route in that direction. Further, when there is a height difference in the search range, a difference is provided in the amount of used resources according to the height difference. For example, if the resource usage is 1 when flat, 2 when generating to a high point.

Branch probability p _a, p _b, p _c is set as follows.
Forward or search forward direction is p _a search, the search in the lateral direction is p _b, forward and lateral direction search to (hereinafter sometimes referred to as omnidirectional.) By setting the value of p _c You can control the direction. For example, if (p _a , p _b , p _c ) = (0, 0, 1) is set, the search node searches in all directions and generates a route if possible.

  As described above, even in a complex environment, it is possible to adapt to the environment and generate a route to the target point.

The route search on the map image was carried out. FIG. 3 shows the result of the route search when assuming a walk from the car. Since car to road width that can travel is limited, branch probability _{(p a, p b, p} c) = (0.3, 0.6, 0.1) was. After that, at the midpoint, the car was alighted and turned into a walk, so the branching probability (p _a , p _b , p _c ) = (0, 0, 1).

Next, FIG. 4 shows the result of route search when resources used for search are controlled in consideration of the height difference. When a route search is performed beyond a point having a height difference, the amount of consumed resources is set as ΔR = M _x1y1 -N _x2y2 +1. Here, M _x1y1 is the height of the current location (x1, y1), and N _x2y2 is the height of the point (x2, y2) where the route is generated.

  Finally, FIG. 5 shows the result of route search in the case of a detour. Limit resources used and shrink to search nodes. The limit value of the set resource is 1350 that causes contraction.

Claims (4)

A route search method for generating a route based on self-similarity for image information,
A step of controlling generation of a route according to a resource to be used, and a step of setting a search range,
A route search method characterized by sequentially performing a desired search in conformity with an environment on an image. Controlling the generation of the pathway is controlling growth and contraction,
2. The route search method according to claim 1, wherein the contraction probability is increased or decreased according to a remaining amount of resources to be used. A route search system that generates a route for image information based on self-similarity,
Means for controlling generation of a route according to resources used, and means for setting a search range;
A route search method characterized by sequentially performing a desired search in conformity with an environment on an image. The means for controlling the generation of the pathway is a means for controlling growth and contraction,
4. The route search system according to claim 3, wherein the contraction probability is increased or decreased according to a remaining amount of resources to be used.

Images