JP4782521B2 - Feature shape simplification device, simplification method and program thereof - Google Patents

Description

  In the present invention, the shape of a feature such as a building or land expressed as a polygon with the number of vertices N (N is a natural number of 5 or more) is converted into a figure with the number of vertices M (M is a natural number of 3 or more and less than N). It relates to the technology to convert.

  In recent years, map information has been digitized and used in automobile navigation systems and the like. One such map information is the shape of a building. A so-called residential map or the like reproduces the planar shape of a building existing at the place, thereby enhancing the convenience of the map. It is possible to input the planar shape of a building based on architectural drawings, but to reduce the labor of inputting the shape of a large number of buildings, for example, aerial photography is used and a method such as figure recognition is used. It is used to generate a polygon (polygonal shape) representing a planar shape of a building.

  The polygon generated in this way is based on the planar shape of the actual building (projected shape in the projected view). It is not uncommon for the number to be 10 or more. When handling such feature shapes as map information, there is a limit on the size of data allowed as map information, and overly complex shapes may not make sense for use as maps. Appropriate complexity as a shape (in other words, appropriate simplicity) is required. Therefore, conventionally, techniques for simplifying polygons having complicated shapes have been proposed (see Patent Documents 1 and 2 below).

JP 2003-141554 A JP 2003-178299 A

  However, in the conventional simplification technique, since it is repeatedly determined whether or not each side meets a predetermined condition, the processing takes time, and the number of vertices can be greatly reduced. could not.

  The present invention solves the problem that it is difficult to greatly simplify a complicated feature shape, and an object thereof is to realize a simplified shape of a building in a short time.

The feature shape simplification device of the present invention that realizes such an object is:
This is a feature shape simplification device for converting a feature shape expressed as a polygon having a vertex number N (N is a natural number of 5 or more) into a figure having a vertex number M (M is a natural number of 3 or more and less than N). And
Edge information input means for inputting information on consecutive N edges constituting the feature shape;
Edge classification means for classifying the N edges to which the information is input into M types based on a direction from the start point to the end point of the edge;
Edge extraction means for extracting one edge for each classification from the classified edges;
Candidate shape forming means for configuring, as a candidate shape, a polygon having at least three or more vertices by the extracted side or a side obtained by extending the side;
An evaluation value calculation means for comparing the formed candidate shape with the feature shape to obtain an evaluation value;
The gist of the invention is that it includes simplified shape determining means for obtaining a simplified shape obtained by simplifying the feature shape based on the evaluation value.

Moreover, the simplification method corresponding to this feature shape simplification device is:
A method of converting a feature shape expressed as a polygon having a vertex number N (N is a natural number of 5 or more) into a figure having a vertex number M (M is a natural number of 3 or more and less than N),
Input information on consecutive N sides constituting the feature shape,
The N sides to which the information is input are classified into M types based on the direction from the start point to the end point of the side,
From the classified edges, one edge is extracted for each classification,
A polygon having at least three or more vertices is formed as a candidate shape by the extracted side or a side obtained by extending the side,
An evaluation value is obtained by comparing the formed candidate shape with the feature shape,
The gist is to obtain a simplified shape obtained by simplifying the feature shape based on the evaluation value.

  According to the feature shape simplification apparatus or the method thereof, a feature shape expressed as a polygon having N vertices can be converted into a figure having M vertices (3 ≦ M <N). According to this apparatus or method, the N sides for which information is input are classified into M types based on the direction from the start point to the end point of the side, and one side is extracted for each classification from the classified sides. Then, a polygon having at least three vertices is formed as a candidate shape by the extracted side or a side obtained by extending the side. Then, an evaluation value is obtained by comparing the formed candidate shape with the feature shape, and a simplified shape obtained by simplifying the feature shape is obtained based on the evaluation value. Therefore, simplification of the feature shape can be realized with a simple configuration. The direction from the start point of the side to the end point can be used as it is when the side is defined as a vector, and either one of them can be used when the side is simply defined by the coordinates of both ends. One may be defined as the starting point and the other as the ending point. At this time, if one of the sides is the end point, the connected end point is the start point and the other is the end point in the other side connected thereto. The multiple sides that make up the feature shape are connected in sequence to form a closed space, so if the start and end points are determined for one side, the start and end points of the other side can be uniquely determined. it can.

  The classification of the sides can be performed depending on which range the direction from the starting point to the ending point of the side belongs to M divided in all directions. As a shape having an area, the triangle has the fewest number of sides, so it belongs to all directions, that is, in the range of 120 degrees by dividing 360 degrees (2π) into 3 parts, so that all sides are classified into three types This is the simplest configuration. In view of the fact that the feature shape is often based on a quadrangle, the shape may be determined depending on which range is divided into four in all directions. Of course, it is possible to increase all the cases from M = 3 to M-1 in order. In this case, all cases may be exhausted, and the highest evaluation may be obtained as a simplified shape of the feature shape, or a simplified shape that is evaluated more than a predetermined value can be obtained sequentially. Then, the processing may be stopped there, and the processing below the edge classification may be terminated.

  When dividing all directions into several parts, it may be divided into M pieces based on a predetermined specific direction (for example, an angle corresponding to the east direction). The longest side may be found and divided into M pieces based on the direction of the side. In general, since a feature is often formed in consideration of sunlight, it is meaningful to classify a side by using a specific direction such as the east as a reference. In addition, since the longest side among the sides constituting the feature often indicates the directionality of the basic shape of the feature, all directions are divided into M pieces based on the longest side. This is effective as a side classification for simplifying the feature shape. The direction of the longest side may be set to 0 degrees, and the first range may be 0 to 360 / M degrees. However, in consideration of the fact that there is an error in the direction information of the side, the direction of the longest side is set to 180 / It is also practical to set M degrees and set the initial range from 0 to 360 / M degrees. Of course, considering that the actual division number M is about 10 or less, the longest side may be regarded as a predetermined angle of about 10 to 15 degrees, and all directions may be divided into M pieces.

  Various methods are assumed to extract the sides for forming the candidate shape from the M types of sides. For example, the extraction can be performed by paying attention to the length of the sides. In this case, if the extraction is performed in order from the long side, it is possible to increase the possibility that the shape close to the original feature shape with a small number of sides is set as the candidate shape. Alternatively, when one side is extracted from one class, it is possible to select a side having a specific relationship with the previously extracted side, for example, a side that is as orthogonal as possible, from the other class.

  In addition, for example, out of edges belonging to each of the M categories, an edge having an absolute value larger than a predetermined value and a ratio larger than a predetermined ratio with respect to the perimeter of the feature shape is extracted. The target is also effective in making a shape approximate to the original feature shape as a candidate shape.

  When forming a candidate shape using the extracted sides, it is also practical to configure only the shapes that do not protrude from the original feature shape as candidate shapes when using the extended sides of the extracted sides. It is. This is because the feature shape originally does not overlap with other adjacent feature shapes, and if only the shapes that do not protrude are used as candidate shapes, the characteristics of such feature shapes can be maintained. Of course, up to a certain ratio, it is possible to allow the protrusion or allow the protrusion depending on the simplified shape of the adjacent building.

  When composing a candidate shape with the extracted sides, only those shapes in which all the two adjacent sides form inside the candidate shape are smaller than a predetermined angle of at least 270 degrees (1.5π) or more are selected. It can also be configured as a shape. In the case of a quadrangular shape or more, it is generally possible to form a convex corner toward the inside of the shape. However, as a normal feature shape, a shape having an inner angle of 270 degrees or more that is an inwardly convex corner is a special shape, and a shape having such a corner is not considered as a candidate shape. There are cases where it is useful. This angle can be set as a predetermined angle of 270 degrees or more, and may be set in advance according to actual processing such as 280 degrees and 290 degrees.

  Evaluation of the candidate shape thus obtained may be performed by calculating an evaluation value using each area of the feature shape and the candidate shape and an area based on the overlap of both shapes. Various methods can be considered for evaluation using such evaluation values. For example, as an area based on the overlap of both shapes, at least one area is obtained from the area of the overlapping portion of both shapes and the area of the portion where the candidate shape does not overlap the feature shape, and the feature of this area A configuration in which the evaluation value is obtained using the ratio of the shape to the area may be employed. For the calculation of the evaluation value, various methods can be determined in advance from the viewpoint of what candidate shape is desirable as a simplified shape of the original feature shape. In the above example, obtaining the ratio of the feature shape to the area corresponds to a so-called normalization process. In this way, large and small shapes can be evaluated using the same method.

  In such simplification of the feature shape, in the process of obtaining the simplified shape based on the input edge information, it is confirmed that a predetermined condition for stopping simplification of the feature shape of the number of vertices N is satisfied. It may be possible to stop the simplification process by making a judgment. The simplification of actual feature shapes is not possible in all cases, and there are cases where it is desirable in total to quit based on predetermined conditions without performing unnecessary search processing. It is. When it is determined that the stop condition is satisfied, the information on the side of the feature shape described in the format prepared for describing the feature shape may be output as it is. On the other hand, when the simplified information is determined, the simplified information may be output in the same format as the format in which the feature shape is described. Of course, a flag indicating whether or not the data to be output has been canceled may be added.

  The predetermined conditions for stopping the simplification process include a case where the length of all the input sides is equal to or smaller than a predetermined determination value, or a straight line whose outer shape is a curved feature shape based on the input side information. It is possible to consider a case where there is an edge approximated by the above, or a case where no candidate shape satisfying a preset condition is obtained in the process of obtaining a simplified shape based on input edge information.

  In the present invention, when a plurality of shapes are obtained as candidate shapes, and any of the candidate shapes has an evaluation value that does not exceed a predetermined threshold value, A shape obtained by combining two or more shapes that satisfy the condition is obtained as a new candidate shape, and if the evaluation value satisfies a predetermined composition condition, the combined candidate shape may be obtained as a simplified shape. . This is because a building having a complicated shape may not be simplified with a single candidate shape.

  Naturally, there can be any number of candidate shapes to be combined, and as a predetermined condition to be satisfied by two or more shapes to be combined, the area of each shape is a predetermined ratio with respect to the area of the feature shape. The condition that it is the above can be provided. Alternatively, the two shapes overlap each other, and all the angles formed by two adjacent sides inside the candidate shape after the composition are smaller than a predetermined angle of at least 270 degrees or more. It is safe to set such conditions. If composition of a plurality of shapes is allowed, simplification of a feature shape having a large number of sides may be extremely complicated. Therefore, it may be limited to the condition of combining two shapes based on the shape of a normal feature. This is because, for an “L” -shaped feature or the like, a simplified shape can be found by synthesizing almost two shapes. Of course, since there are “U” shaped features, it is possible to combine up to three shapes or more.

  Note that the present invention is not only grasped as such a feature shape simplification device and method thereof, but also as an invention of a program for realizing the simplification method by a computer, an invention as a rabbit medium recording the program, etc. It is also possible to grasp.

    The feature shape simplification apparatus and method as one embodiment of the present invention will be described below with reference to embodiments. FIG. 1 is a schematic configuration diagram of a feature shape simplifying device 10 as an embodiment. The apparatus 10 includes a CPU 21 that performs arithmetic processing, a RAM 22 that is used for a work area for processing, a ROM 23 that stores a monitor program, a keyboard 25 with a mouse for inputting user instructions, a feature shape being processed, and the like. And a hard disk 30 for storing a feature shape database BDB and the like. In this embodiment, the feature shape that is the object of the simplification process is defined in advance as a set of a plurality of line segments constituting the building, and constitutes a database.

  FIG. 2 shows an example of the data structure of the feature shape database BDB. As shown in the figure, in this example, the feature shape database BDB mainly stores “element type”, “number of hollow surfaces”, and “outer frame surface information”. The element type is information indicating which element corresponds to the map database GDB since the database BDB constitutes a part of a higher-level map database. In the present embodiment, the element type has a value indicating that it is surface or outline information in the map database GDB. The number of hollow surfaces is information indicating the number of hollow portions when there is a hollow portion in one feature shape. In this database BDB, since there is no hollow, the number of hollow surfaces is 0 in all feature shapes.

  The outer frame surface information is composed of coordinate data of n points constituting the outer frame indicating the feature shape. The outer frame surface information is composed of information on the number of coordinate points on the outer frame surface and data of coordinates corresponding to the number n defined here. FIG. 3 is an explanatory diagram showing a correspondence relationship with outer frame surface information about one building registered in the feature shape database BDB. As shown in the figure, this feature shape is composed of 17 sides, and each side is connected by a connection point. This connection point CP, which is represented by a set of x and y coordinates (x1, y1) (xn, yn) from the connection point CP1 to the connection point CPn, is “outside of the feature shape database BDB. Frame surface information ”. In this example, n = 17, and coordinate data of a total of 17 connection points CP1 to CP17 are recorded. In this example, the coordinates of the connection point CP are described. However, the connection point CP1 and the connection point CPn are different points, and it is assumed that the next of the connection point CPn is treated as the connection point CP1. There will be edges between successive connecting points. Therefore, this database BDB stores the coordinates of the connection point CP, but essentially defines the sides constituting the feature shape. Since the shape is different for each building and the number of connection points is also different, this database BDB is a so-called variable length database.

  The CPU 21 simplifies the feature shape by temporarily loading a simplified program, which will be described later, from the hard disk 30 to the RAM 22 constituting the main memory and executing the program on the main memory. In the example shown in FIG. 3, the simplified shape is a shape with six connecting points as shown in FIG. 4, for example. In addition, since the connection point after simplification does not necessarily correspond with the connection point with which the original feature shape was provided, in FIG. 4, the connection point after simplification is represented by VP. The simplified shape in this case is also defined in the same format as the database shown in FIG. The simplified feature shape database is referred to as a simplified feature shape database SDB. This simplified feature shape database SDB is also stored in the hard disk 30 in the same manner as the feature shape database BDB. Although details will be described later, there may naturally be cases where simplification is not possible. In this case, the original data (see FIG. 2) that has not been simplified is stored in the simplified feature shape database SDB as it is.

  Next, processing executed by the feature shape simplification apparatus 10 will be described. FIG. 5 is a flowchart showing processing executed by the apparatus 10. When the feature shape database BDB is prepared and the simplification apparatus 10 is instructed to start the simplification process via the keyboard 25, the simplification device 10 first accesses the feature shape database BDB, and from the database BDB, the first feature A process of reading shape data is performed (step S100). In short, data corresponding to one building shown in FIG. 2 is read from the hard disk 30. Subsequently, a process of reading the simplification threshold value TS is performed (step S110). This threshold value TS is a threshold value for performing an evaluation when the original feature shape is simplified. As will be described in detail later, a criterion for determining whether or not the candidate shape obtained under a predetermined condition can be regarded as simplified is given.

  Such a threshold value TS may be uniformly set to the same value for all feature shapes, but may be changed depending on characteristics of the feature shapes. For example, it may be changed in accordance with the complexity of the feature shape, that is, the number of connection points CP. This is because if the shape becomes complicated, it may be difficult to determine that simplification has been achieved unless the threshold value is lowered to some extent. Of course, the threshold value TS depends not on the complexity of the feature shape, but on the size of the feature shape (occupied area), the distance to the adjacent building, and the type of building (type of target or individual house). It is good also as what changes. In the embodiment, as shown in FIG. 6, a value between the lower limit value TSl and the upper limit value TSu is taken according to the area SS of the feature shape of interest. That is, in the embodiment, the threshold value TS is equal to the value TS1 if the area SS of the feature shape is less than or equal to the lower limit SL, and equal to the value TSu if the area SS of the feature shape is greater than or equal to the upper limit SU. Is defined as a value obtained by linear interpolation between the lower limit value TSl and the upper limit value TSu. Of course, instead of processing such as linear interpolation, the threshold value TS is held in the form of a table, and when data of one feature shape is read from the feature shape database BDB (step S100), the area is calculated, The threshold value TS may be read with reference to the table.

  Thus, data of one building is acquired from the feature shape database BDB (step S100), a threshold value TS for simplification of the feature shape is read (step S110), and whether simplification is possible next. (Step S120). The determination as to whether or not simplification is possible will be described later in detail with reference to FIG. 7, but not all feature shapes can be simplified. If it is determined in step S120 that simplification is not possible, the process proceeds to step S125, and the data obtained from the feature shape database BDB (see FIG. 2) is used as it is after the simplified feature shape database SDB after simplification. To output as. Even when the simplification process cannot be performed in this way, the data is written in the database SDB in the present embodiment in order to draw a map. This is to keep the data aligned. If the object is simply to simplify the feature shape, the feature shape that has not been simplified may be configured not to output data.

The contents of determination as to whether simplification is possible will be described with reference to FIG. FIG. 7 is a flowchart showing details of the simplification determination processing routine (step S120). As shown in the figure, when this process is started, first, based on the data of one building acquired from the feature shape database BDB, the length in the X direction and the Y direction of each side constituting the feature shape It is determined whether all the lengths are smaller than the predetermined value L1 (step S121). The lengths Lxi and Lyi in the X direction and Y direction of the side existing between the i-th connection point CP _i and the next connection point CP _{i + 1} are:
Lxi = | x _{i + 1} −x _i |
Lyi = | y _{i + 1} −y _i |
Can be easily obtained. Therefore, if the lengths Lxi and Lyi of these sides are all equal to or smaller than the predetermined value L1, it is determined that the simplification process is not performed because the feature shape is a small building.

  If it is determined that each side representing the feature shape has a length L1 or more (step S121), then whether or not a curved figure is included in a part of the feature shape of interest is determined. A determination is made (step S122). In this embodiment, when a curved shape is included in at least a part of the outer shape such as a circular hole, this is described as a series of short straight lines. Therefore, when the short vectors of the predetermined value L2 or less continue for the predetermined value N1 or more, it is determined that the short vector approximates the curve shape and is excluded from the target of simplification processing. . In the present embodiment, since simplification is based on straight line approximation, it is originally assumed that a curved figure is not simplified. In the example, the actual length of 25 cm was set as one unit, and L1 was set as 40 units. Similarly, in the example, L2 was 10 units (corresponding to 2.5 meters), and N1 = 10, and the curve figure was judged. Of course, the curved figure is not excluded from the simplification process, and may be the object of the simplification process in the same manner as other figures.

  If it is determined in step S121 or S122 that the candidate shape cannot be generated, the process returns to “1” and this routine is temporarily terminated. “1” is a connection point of the process shown in FIG. 5. In this case, the process proceeds to the above-described step S125, and a candidate shape cannot be generated, and data of the original feature shape is copied. It is. On the other hand, if it is determined in steps S121 and S122 that the simplification process is possible, the process goes to “NEXT”, the process routine is terminated, and the process returns to the process of FIG.

  If it is determined that the simplification process is possible, next, a process of classifying all sides included in the data acquired from the feature shape database BDB into four (step S130). In the present embodiment, the classification into four is processing in which each side is regarded as a vector having a start point and an end point, and the angle θ of the vector belongs to which of the four quadrants. In the present embodiment, the four quadrants are determined based on the longest side. That is, first, all sides are searched to find the longest side, the direction of this side is set to an angle of 0, −0.25π ≦ θ <0.25π is the first quadrant, and 0.25π ≦ θ <0.75π is the first. Two quadrants, 0.75π ≦ θ <1.25π are defined as the third quadrant, and 1.25π ≦ θ <1.75π are defined as the fourth quadrant. Then, classification is made according to which of these four quadrants each side belongs to. FIG. 8 shows an example of a feature shape including the sides S1 to S8. In this example, the side S5 is the longest. Therefore, when classification is performed based on this side, the result shown in FIG. 9 is obtained.

  The classification into four in this embodiment is for convenience, and there is no problem even if it is classified into five. Further, it is sufficient that each classification is performed based on the direction of the side, and even if the four classifications are performed, each angle range does not necessarily need to be 90 degrees. The angle range of each classification may be determined in advance one by one. Further, in this embodiment, the range of each quadrant is determined based on the direction of the longest side, but each quadrant is determined by setting a predetermined direction, for example, a direction corresponding to the east on the actual map as an angle 0. Also good.

  After the sides are classified into four in this way, it is determined whether or not a candidate shape can be generated (step S135). Candidate shapes are generated by taking one side from each of the four categories and combining the four sides for convenience. Of course, if the original feature shape has five or more connecting points CP, each side is not necessarily connected even if the four classified sides are taken out one by one. Therefore, when generating candidate shapes, the candidate shapes are generated by appropriately extending the four sides extracted from each classification.

  Generation of candidate shapes will be described along the feature shapes illustrated in FIG. Since the feature shape shown in FIG. 8 has eight connecting points CP1 to CP8, eight sides S1 connecting the adjacent connecting points CP clockwise from the smaller number to the larger one. Or S8. Each side S1 to S8 is regarded as a vector, and its direction is classified into the above four angle ranges. For easy understanding of the direction, each side is shown as a vector having an arrow on the end point side in FIG. The classified results are shown in FIG. In FIG. 9, the four sides classified into the first quadrant to the fourth quadrant are classified with the longer side as the top. That is, S5 which is the longest side among all sides belongs to the first quadrant, sides S4 and S2 belong to the second quadrant, and the upper side S4 is long in this order, and the third quadrant has sides. S1, S3, S7 belong, the upper side S1 is the longest in this order, sides S8, S6 belong to the fourth quadrant, and the upper side S8 is longer in this order. Therefore, out of the sides classified into each quadrant, the longest one is sequentially extracted, and a candidate shape is generated by combining them. Examples of generated candidate shapes are shown in FIGS. In each figure, the original feature shape is indicated by a broken line, a selected side is indicated by a solid line, and a portion where each side is extended to the intersection is indicated by a one-dot chain line. In FIG. 10, a candidate shape SS9 is generated as a combination of the longest sides S1, S4, S5, and S8 in each classification. In FIG. 11, basically, as in FIG. 10, long sides are extracted from each classification to generate candidate shapes, but in this example, only two sides S <b> 2 are selected for the sides belonging to the third quadrant. , S4, the candidate shape SS10 is generated using the second longest side S2. Comparing the two, in the candidate shape SS9, the region BB9 shown in the figure is a region that protrudes from the original feature shape, whereas in the candidate shape SS10, such a region that protrudes does not exist.

8 to 11 show a case where a candidate shape can be generated, but depending on the shape, there are cases where only a shape that cannot be handled as a candidate shape can be configured. Therefore, in this embodiment, the following condition is defined as the condition. That is,
(1) When the length of each side S taken out from each classification is LL,
(1-1) LL> L3.
(1-2) LL / LA> T4
(Where LA is the perimeter of the feature shape),
(2) Each interior angle α in the candidate shape is α ≦ (2π−T5),
Only when this condition is satisfied, it is determined that the candidate shape can be generated. In the embodiment, T5 = 0.5 [rad], but this size is also affected by the size of the output map.

  The above condition (1-1) is to prevent the candidate shape from being generated by forcibly expanding a side whose absolute value is too short, and the condition (1-2) is equal to or greater than a predetermined length L3 as the absolute value. This is because even if the original feature shape is large, the candidate shape is not generated by expanding a relatively small side, and condition (2) is a shape that is cut too deep inside as shown in FIG. Is a condition for not to be a candidate shape. When the sides are classified into four and combined, there are usually few candidate shapes excluded by this condition (2), but this may occur when the number of classification is five or more.

  If a shape that satisfies all of the above conditions cannot be obtained even after trying all the combinations of the sides, candidate shapes cannot be generated (step S135), and the process proceeds to step S125. A process of copying the data of the feature shape is performed. On the other hand, when all the combinations of sides are tried and a shape that satisfies all the above conditions (1-1), (1-2), and (2) is found, the candidate shape is It is determined that it can be generated, and then processing for generating a candidate shape is performed (step S140). In many cases, a plurality of candidate shapes can be generated as illustrated in FIGS. 8 to 11. In this embodiment, all shapes that can be generated while satisfying the above conditions are generated as candidate shapes.

After the candidate shape is generated, next, a process of calculating the score SC for each candidate shape is performed (step S145). In this embodiment, the score SC is calculated by the following equation (1), where A is the area of the original feature shape and B is the area of the candidate shape.
SC = (AB−λ · BB) / A (1)
Here, AB is the product area of the feature shape and the candidate shape, and BB is BB = B−AB, that is, the area where the candidate shape protrudes from the overlapping shape with the feature shape. Also, λ is a weighting coefficient (penalty) for the amount of protrusion.

In the above formula (1), if AA is defined as AA = A−AB, that is, the area where the original feature shape protrudes from the overlapping shape with the candidate shape,
SC = (A−AA−λBB) / A
Therefore, the score SC eventually evaluates both the area AA in which the focused candidate shape is too small and the area BB in which the feature shape is excessive. Therefore, if λ = 1, the excessive portion and the excessive portion are equally weighted, and if λ = 0, the excessive portion is allowed, and λ = If it is set to ∞, no protrusion will be allowed. In this embodiment, λ = 100 is set with priority given to the fact that the simplified shapes after simplification of adjacent buildings do not overlap each other. The reason for dividing by the whole A is to make the score SC a value that does not depend on the size of the feature shape (normalization).

  After calculating the score SC, it is next determined whether or not the candidate shape score SC generated in step S140 is larger than the previously read threshold value TS (step S110) (step S150). When a plurality of candidate shapes are generated, all candidate shapes are determined. Here, if there is a candidate shape having a score SC exceeding the threshold value TS, a candidate shape having the highest score among candidate shapes having a score SC exceeding the threshold value TS is output as a simplified shape (step) S160). As described in step S125, the output format is the same format as the feature shape. Thereafter, it is determined whether or not the simplification process has been completed for all the buildings stored in the feature shape database BDB (step S170), and if not completed, the process returns to step S100 to return from the feature shape database BDB. Repeat the following process to obtain data.

  After calculating the score SC (step S150), when a candidate shape having a score SC exceeding the threshold TS cannot be found (step S150), a process for generating a union is executed (step S170). The union is generated as follows. Of the candidate shapes (both square) generated in step S140, N2 are extracted in descending order of area, and the ratio of the area B to the area A of the feature shape exceeds the predetermined value T2. A union of two quadrangles is generated for a target, and this is used as a new candidate shape. In the example shown in FIG. 8, as shown in FIG. 13, the union set SSor with the quadrangle dS1 formed by the sides S1-S2-S5-S8 and the quadrangle dS2 formed by the sides S2-S3-S4-S5 is a new one. It becomes a candidate shape. It should be noted that those that do not cause an overlap between the two squares are not considered as simplification candidates. Similarly, a combination in which any one of the inner angles α satisfies α ≦ (2π−T5) even if there is an overlap is excluded. Furthermore, the case where the shape which combined two squares turns into a more complicated shape than the original feature shape is also excluded.

  Thereafter, it is determined whether at least one union has been generated (step S180). If none can be generated, the simplification process cannot be performed, the process proceeds to step S125, and the original data is output as it is. All simplification processes for the feature shape of interest are completed. On the other hand, if at least one union can be generated (step S180), the process returns to step S145, and the above-described processing is repeated from the calculation of the score SC for the shape of the union (steps S145 to S160). If there is a shape of the union that has a score SC higher than the threshold value TS, a process of outputting the union of the highest score among them as a simplified shape is performed (step S160). In any case, since the simplification process has been completed for the feature shape focused on by this, it is determined whether the simplification process has been completed for all the feature shapes in the feature shape database BDB. (Step S170), the above processing is continued until completion.

  According to the present embodiment described above, a large number of complex feature shapes can be simplified to a quadrangle or a shape of two unions thereof by a simple process. Conventionally, it has been extremely difficult to process and simplify a large number of complicated shapes such as feature shapes existing on a map at high speed. What to do when generating a map for a navigation system from a detailed map that is the basis of a map used for car navigation etc. In some cases, it may be necessary to simplify the shape of as many as 10 million features, and a simplification device and a simplification method that perform such processing at high speed are extremely useful. In the present embodiment, the input feature shape data is output as it is without performing unreasonable simplification for those that cannot be simplified, so that it can be directly used for drawing a map or the like. Even if not all feature shapes are simplified, if most of the feature shapes are simplified, the total data capacity can be reduced. This is convenient when recording on a DVD. In addition, shapes that are not simplified can be easily identified by looking at the number of connection points, so only those data are extracted, and other simplified methods that simplify time-consuming but exceptional shapes are adopted. Then, the simplification process may be performed. In this case, since many feature shapes have been simplified by the method of the present embodiment, even if other methods require time for the simplification process, the total time is not so long. You can do it.

  Further, in this embodiment, the threshold value TS for evaluating the simplification is changed as shown in FIG. 6 by the area SS of the feature shape. Therefore, only the feature shape and approximate candidate shape are output as the area increases. The larger the shape, the larger the area that protrudes or is insufficient even if the proportion of matching is equal. For this reason, as the area SS of the feature shape increases, only those having a high matching rate are output as simplified shapes. In this case, the feature shape with a larger area is difficult to be simplified, but in reality, the larger the feature shape, the larger the number of sides constituting the feature shape, so that various candidate shapes can be created. Since the proportion of the feature shape in the feature shape database BDB becomes relatively smaller as the feature shape is increased, even if the data is output as it is, the increase in the amount of data as a whole is not so large. The effect of reducing the capacity of the entire data can be enjoyed.

  In this embodiment, when calculating the score SC, the coefficient λ is used to easily change the evaluation ratio for the protruding portion. Therefore, depending on the nature of the target feature shape, from the condition that the protrusion is not allowed (λ = ∞), the condition that evaluates to the same extent as the part where the protrusion does not overlap (λ = 1), and further, the protrusion Can be easily set up to a condition (λ = 0) that does not give a penalty on the score.

  Further, in this embodiment, when sufficient simplification cannot be achieved with a single quadrangle, a union of candidate shapes satisfying a predetermined condition can be obtained and used as a new candidate shape. Even in this case, it is not necessary to change the evaluation method, and the candidate shape of the quadrangle that is the basis of the union is extracted first, so that the handling is easy. Since buildings are often realized by a combination of squares, simplification with a square or a union thereof has the advantage of easily leaving the features of the original shape. For the same reason, the technique for obtaining the union and evaluating it again is simple but has a great effect. In this embodiment, the candidate shape to be obtained first is a quadrangle, but a pentagon or a hexagon may be allowed. Further, the union is not limited to two, and simplification processing may be performed assuming that three or more unions are allowed.

  In the present embodiment described above, the simplification process is performed using the dedicated simplification apparatus 10, and the simplified feature shape database SDB is formed for display and output from the feature shape database BDB. However, the simplification algorithm described with reference to FIGS. 5 and 7 may be incorporated in a display or output device, and simplification processing may be performed on the spot as needed. When displaying a map on the screen of a car navigation system or mobile phone, the number of feature shapes provided for display at one time is not so large, so only the necessary feature shapes are simplified and displayed. It is also possible. For example, when a map is displayed on a terminal device such as a mobile phone connected via a network, the map data itself is prepared on the server side. Moreover, since the display magnification on the terminal device side is various, depending on the magnification, there is a case where it is desired to display the building using the data of the feature shape database BDB stored in the server as it is. There is also a case where it is desired to display the building while simplifying the shape of the building by reducing the scale. In the latter case, the simplification process can be performed only on the feature shape included in the region to be displayed while adjusting the threshold value TS according to the scale. The simplification process in this case may be performed on the server side or the terminal device side.

  As mentioned above, although one Example of this invention was described, this invention is not limited to such an Example at all, Of course, in the range which does not change the summary of this invention, it can implement in a various aspect.

It is a schematic block diagram of the simplification apparatus 10 as one Example of this invention. It is explanatory drawing explaining the data structure of the feature shape database BDB in the simplification apparatus 10. FIG. It is explanatory drawing which illustrated one of the feature shapes used as the object of simplification. It is explanatory drawing which illustrates the simplified shape after being simplified. It is a flowchart which shows the simplification process routine in a present Example. It is a graph which shows the setting method of threshold value TS. It is a flowchart which shows a part of determination process routine of simplification. It is explanatory drawing which shows a mode that a candidate shape is produced | generated based on the edge classified into four. It is explanatory drawing which shows the mode of the classification | category of each edge | side S1 thru | or S8. It is explanatory drawing which illustrates the relationship between a candidate shape and a feature shape. It is explanatory drawing which illustrates the other relationship of a candidate shape and a feature shape. It is explanatory drawing explaining the restriction | limiting by the magnitude | size of an interior angle. It is explanatory drawing which shows a mode that a new candidate shape is produced | generated from two squares.

Explanation of symbols

10 ... Feature shape simplification device 21 ... CPU
22 ... RAM
23 ... ROM
25 ... Keyboard 27 ... Display 30 ... Hard disk

Claims (14)

This is a feature shape simplification device for converting a feature shape expressed as a polygon having a vertex number N (N is a natural number of 5 or more) into a figure having a vertex number M (M is a natural number of 3 or more and less than N). And
Edge information input means for inputting information on consecutive N edges constituting the feature shape;
Edge classification means for classifying the N edges to which the information is input into M types based on a direction from the start point to the end point of the edge;
Edge extraction means for extracting one edge for each classification from the classified edges;
Candidate shape forming means for configuring, as a candidate shape, a polygon having at least three or more vertices by the extracted side or a side obtained by extending the side;
An evaluation value calculating means for calculating an evaluation value by comparing the feature shape as the structure made candidate shape,
A feature shape simplification device comprising: simplified shape determination means for obtaining a simplified shape obtained by simplifying the feature shape based on the evaluation value.   2. The feature shape simplification device according to claim 1, wherein the side classification unit is a unit that classifies the side according to a range in which the direction of the side belongs to M divided in all directions.   The feature shape simplification device according to claim 2, wherein the M divisions in all directions are performed based on a direction of a longest side among the sides to be classified.   2. The feature shape according to claim 1, wherein the edge extracting unit extracts a predetermined number from the longest side among the edges belonging to each of the M categories, and uses the evaluation value calculating unit for the calculation. Simplification device.   The edge extraction means is an edge whose absolute value is larger than a predetermined value and has a ratio larger than a predetermined ratio with respect to the perimeter of the feature shape, among the edges belonging to each of the M types. The feature shape simplification device according to claim 1, wherein the object is to be extracted.   2. The feature shape simplification device according to claim 1, wherein, when using a side obtained by extending the extracted side, the candidate shape forming unit configures only a shape that does not protrude from the feature shape as the candidate shape.   When the candidate shape forming means configures the candidate shape by the sides, only the shapes in which all the angles formed by the two adjacent sides inside the candidate shape are smaller than a predetermined angle of at least 270 degrees or more. The feature shape simplification device according to claim 1 configured as the candidate shape.   The feature shape simplification device according to claim 1, wherein the evaluation value calculation means calculates an evaluation value using each area of the feature shape and the candidate shape and an area based on an overlap of both shapes. The feature shape simplification device according to claim 8,
The evaluation value calculation means, as an area based on the overlap of the two shapes, at least one of the area of the overlapping portion of both shapes and the area of the portion where the candidate shape does not overlap the feature shape A feature shape simplification apparatus that obtains the evaluation value using a ratio of the area to the area of the feature shape. The feature shape simplification device according to claim 1, further comprising:
In the case where a plurality of shapes are obtained as the candidate shapes, and when any of the candidate shapes does not exceed the predetermined threshold value, a predetermined condition among the plurality of shapes is satisfied. Means for obtaining the evaluation value as a new candidate shape obtained by synthesizing two or more shapes satisfying, and obtaining the synthesized candidate shape as the simplified shape when the evaluation value satisfies a predetermined synthesis condition With
Feature shape simplification device. The feature shape simplification device according to claim 10,
The feature shape simplification device including a condition that two or more shapes to be combined satisfy a condition that an area of each shape is a predetermined ratio or more with respect to an area of the feature shape. The feature shape simplification device according to claim 10,
The predetermined synthesis condition is:
The two or more shapes overlap, or
All the angles formed by the two adjacent sides in the candidate shape after synthesis inside the candidate shape are smaller than a predetermined angle of at least 270 degrees or more.
Is a feature shape simplification device. A computer having a feature shape database in which shape information relating to the shape of a feature is stored is used to represent a feature shape expressed as a polygon having a vertex number N (N is a natural number of 5 or more). A natural number of 3 or more and less than N),
The computer acquiring information of consecutive N sides constituting the feature shape from the feature shape database;
The computer classifying the N sides from which the information has been acquired into M types based on a direction from the start point to the end point of the side;
The computer extracting one edge for each classification from the classified edges;
The computer constructs a polygon having at least three vertices as candidate shapes by the extracted side or a side obtained by extending the side; and
The computer calculates the evaluation value by comparing the configured candidate shape and the feature shape;
The computer obtains a simplified shape obtained by simplifying the feature shape based on the evaluation value;
Perform feature shape simplification method. A process for converting a feature shape expressed as a polygon having a vertex number N (N is a natural number of 5 or more) into a figure having a vertex number M (M is a natural number of 3 or more and less than N) is related to the shape of the feature. A program to be executed by a computer having a feature shape database in which shape information is stored,
A function of acquiring information on consecutive N sides constituting the feature shape from the feature shape database;
A function of classifying the N sides from which the information has been acquired into M types based on the direction from the start point to the end point of the side;
A function of extracting one edge for each classification from the classified edges;
A function of configuring a polygon having at least three vertices as candidate shapes by the extracted side or a side obtained by extending the side;
A function for obtaining an evaluation value by comparing the configured candidate shape and the feature shape;
A function for obtaining a simplified shape obtained by simplifying the feature shape based on the evaluation value;
Program to realize.

Images