JP2001117487A - Device and method for generating polygon road network data - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device and a method to generate road network data in which road elements between crossings are represented by polygons in an electronic map data while minimizing human work. SOLUTION: First, one view-pont is specified within a road on a segment map. Then, visible segment group that are seen from the view-point are detected and a node ploygon 91 which represents one crossing is generated using the groups. Then, a link side 95 which corresponds to an entrance to other road portion is selected among the sides of the node polygon, the visible segment groups being seen from a point 96 on the side 95 and located outside the polygon 91 are detected. Using the groups, a node polygon 119 which is an adjacent crossing and a link polygon 123 which corresponds to the road elements connecting the two node polygons are generated. Similarly, the processes are repeated to generate a new node polygon and a link polygon to produce polygon road network data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for converting a road portion into a polygon (polygon) in a form that minimizes human intervention when the road portion is not identified as a road in electronic map data in a form that can be handled by a computer. ) To create road network data.

2. Description of the Related Art Main electronic maps have a scale of 1/25000.
Based on the Geographical Survey Institute's National Land Map, based on road maps used for car navigation, PC maps, etc., and based on the local government's city planning basic map of 1/2500 scale, this was reduced to 1/1500 scale. A typical example is an enlarged house map or a city map with a scale of about 1/10000. Among them, in the road map of the scale 1/25000, the road width is expressed by several stages of "two-line" different from the actual scale, and the scale is 1/1000.
In the 0 to 1/2500 map, the scale is the actual size.
In particular, in a house map of 1/2500 scale, since "buildings" are mainly represented, "roads" are classified as "others", and a database is not created in a form in which roads are recognized as roads. There are many. However, in recent years, detailed maps around the destination have become necessary for car navigation, and in the case of welfare and nursing care, route guidance using detailed maps, so-called walk navigation, has been required. It has become to.

[0002]

However, the scale 1
There is no explicit type of "road" in the / 2500 level house map database, and considering the future expansion of the range of use, road recognition from house map data and creation of road network data have become major issues. I have.

[0003] Road recognition from house map data has conventionally been approached in various ways.
A method with road extraction and recognition accuracy up to the practical level has not yet been established. The main method of the traditional approach is based on the fact that a road is composed of parallel line pairs, and therefore first detects parallel line pairs and connects them under various conditions. To the road area. However, in this method, since the intersection area is detected only as a break area of the parallel line segment pair, the shape of the actual intersection cannot be accurately represented. In addition, it is not possible to accurately represent the shape of an actual road that does not always satisfy the parallel condition or the shape of a junction between an intersection and a road.

Accordingly, an object of the present invention is to determine the shape of an intersection as a polygon (polygon) in a form that minimizes human intervention when electronic map data consisting of a set of line segments is given. It is to polygonize the shape of the road element part between the intersection and the intersection, and to describe the relationship between the intersection and the road element as a network.

[0005]

According to the polygon road network creating method according to the first aspect of the present invention, first, a point existing on a road on the line segment map data is initially designated as a viewpoint. Then, a first visible line segment group seen from the viewpoint is detected, and a polygon representing a part of the road is created using the first visible line segment group. After that, one starting polygon is selected from the already created polygons, and a side corresponding to an entrance of another road portion is extracted as a link side from a plurality of sides of the starting polygon, and one side is extracted from those link sides. One starting link side is selected, and a second visible line segment group that is visible outside the starting polygon from points other than the end points on the starting link is detected. Then, at least one new polygon is created using the second visible line segment group, which means another road portion that is continuous with the starting polygon. After that, by repeating the process of selecting a new starting link side and creating a new polygon, a large number of continuous polygons can be created one after another from the first created polygon, thereby creating a polygon road network data Is created.

According to this method, starting from a viewpoint initially designated in the road as a starting point, the viewpoints are picked up from the viewpoints while sequentially moving to new places in the road. From the line segments representing various objects such as buildings and roads on the map, it is possible to selectively extract the line segments that constitute the outline of the road, and use these extracted line segments to display the shape of the road. Can be created. Even if the roads are not parallel and the intersection has a complicated shape, it is possible to create a polygon that faithfully represents those shapes. Then, when a polygon of a part of the road is created, then, starting from the polygon, a polygon of another road part connected to the polygon is created, and so on.
Polygons of road portions that are continuous in a potable manner are created, and polygon road network data is created.

According to the polygon road network creating method according to the second aspect of the present invention, first, a point existing on a road on the line segment map data is initially designated as a viewpoint, and the point can be viewed from the viewpoint. A first visible line segment group is detected, and a node polygon representing one intersection is created using the detected first visible line segment group. After that, one starting node polygon is selected from the already created node polygons, and a side corresponding to the entrance of another road portion is extracted as a link side from among a plurality of sides of the starting node polygon,
One starting link side is selected from those link sides, and a second visible line segment group that is visible outside the starting polygon from points other than the end points on the starting link side is detected. Then, based on the second visible line segment group, a new node polygon which means one intersection adjacent to the starting node polygon is created, and this new node polygon is set as an end node polygon. Further, based on the second visible line segment group,
A link polygon representing a road element connecting the departure node polygon and the end node polygon is created. The second above
By repeatedly performing the detection of the visible line segment group, the creation of the terminal node polygon, and the creation of the link polygon, a large number of continuous node polygons and link polygons can be created one after another. Can be built up.

According to this method, polygon road network data composed of node polygons faithfully representing the shape of an intersection in a road and link polygons faithfully representing the shape of a road element connecting the intersections is created. can do.

In a preferred embodiment, a node point is set in each node polygon, and a link segment passing through each link polygon is set so as to connect each pair of adjacent node points. Thus, in addition to polygon road network data, road network data of points and lines composed of a number of node points and link segments can be created together.

In a preferred embodiment, when a node polygon is created, an end point group of the first visible line segment group is mapped so that an end point closer to the viewpoint is located farther from a start point. Then, a convex hull of the mapped end point group is created, and then the convex hull is inversely mapped by returning the mapped end point group to the original arrangement. Then, the polygon corresponding to the inversely mapped convex hull is defined as a node polygon. As a result, the node polygon has a shape that accurately represents the shape of the intersection, with a small portion protruding to the road element extending from the intersection.

In a preferred embodiment, when creating the final node polygon, first, based on the second visible line segment group, one point having a possibility of being within an intersection adjacent to the starting node polygon is detected. One detected point is designated as a new viewpoint. Then, for this new viewpoint, a final node polygon is created in the same procedure as when the first node polygon was created. As a result, starting from one point designated in the road as a starting point, locations of nearby intersections are successively detected, and node polygons are created at the locations of each intersection. Ideally, all intersections existing on the road leading to the first designated point are detected, and node polygons are created at all of these intersections, so simply specifying the first one point will connect to that designated point A polygonal road network covering all roads is automatically created.

In a preferred embodiment, as a method of designating the new viewpoint, the second visible line segment group is divided into a right line segment sequence and a left line segment sequence, and the second visible line segment group is selected from the right line segment sequence and the left line segment sequence. From the line segment sequence on one side, one gap corresponding to the entrance of the side road closest to the departure link side is detected, one point is selected between this gap and the line segment sequence on the other side, and that one point is renewed. Specify as a viewpoint. According to this method, a new viewpoint can be accurately set in an intersection adjacent to the departure node polygon.

When a new viewpoint is specified by the above-described method, and a gap corresponding to an entrance of a side road cannot be detected from any of the right and left line segments,
Within a predetermined range, a gap between an end point farthest from the departure link side of the right line segment sequence and an end point farthest from the departure link side of the left line segment sequence is detected, and a new viewpoint is designated near this gap. . Thereby, even if the road element connected to the departure node polygon extends long without having an intersection, a dummy node polygon representing a virtual two-way intersection can be formed in the middle of the road element. Then, using this dummy node polygon as a new starting point,
Polygon road network data can be created so as to extend further.

Other features of the present invention will be specifically described in the following description of the embodiments.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The method of the present invention can be typically carried out by using a programmed computer, but it is needless to say that at least a part of the processing can be carried out by a dedicated hardware circuit. It is possible.

An embodiment of the present invention will be described below. This embodiment is implemented as a computer program for a general-purpose computer in which source code is described in C language. The structure of a general-purpose computer itself is
Since it is well known to those skilled in the art, no particular description will be given. Since the C language is also well known to those skilled in the art, there are several explanations that use the description of the C language, but it should give those skilled in the art a more specific understanding.

1. First, in the present embodiment, in the present embodiment, a computer recognizes a road from line segment set data of a map, polygonizes the road, and creates a road network (hereinafter referred to as a road network recognition process). The general procedure is as follows (1)-
It is shown in (5).

(1) It is difficult to make a computer 100% surely recognize where a road is from a line segment set. Therefore, the line segment set data is displayed on a display screen, and the pattern recognition ability of a human (operator) is utilized. , A point such as "This is definitely one point on the road."

(2) An area near the point specified in (1) is regarded as an “in-road area”, and an area connected to this area is detected as a road candidate. Where "road"
Is the "intersection" and the "road element" that connects the intersections.
It is composed of different types of elements.

(3) The specific procedure of the process (2) is as follows. First, assuming that the designated point exists in a certain intersection, a convex polygon (convex closure) indicating the area of the intersection is determined. At this time, if the specified point is really within the intersection, the convex polygon of the intersection (hereinafter referred to as “node polygon”) is determined, and the specified point is not within the intersection but is a road element connecting the intersections. In the case where the rectangular area of the minimum road width including the designated point
Crossroads virtual intersection ". Next, a side corresponding to the road element is extracted from the intersection polygon determined in this way, and the line segments constituting the road element are sequentially traced from the extracted side, and the nearest intersection is detected and the node polygon is detected. Register as In addition, a road element between both intersections is extracted and registered as a road element polygon (hereinafter, referred to as a “link polygon”). Next, from the detected node polygons, a road network recognition process is performed by a “potato-type method” in which the subsequent node polygons and link polygons are sequentially detected in a similar manner.

(4) If the road network recognition processing is ideally performed, all roads connected to the first designated point should be able to be detected. If the part "a certain point on the road" is additionally specified, the detection is continued.

(5) By regarding the center of gravity of the detected node polygon as a node and the center line of the link polygon as a link, basic road network data is automatically created when the road network recognition processing is completed. Have been.
Thus, road detection, polygonization and road network creation from the line segment data set can be achieved with minimal human intervention.

2. Input data: line segment map data In the present embodiment, the following line segment map data is input.

(1) The coordinate system is a mathematical plane orthogonal coordinate system having an origin at the lower left, extending the x positive axis rightward, and extending the y positive axis upward. Feret coordinates (= lower left corner coordinates, upper right corner coordinates) of the range of one map are (XS, YS), (XE, YE)
And XS is the minimum x coordinate of the processing target range, and YS is the minimum y.
The coordinates, XE, are the maximum x coordinates, and YE is the maximum y coordinate.

(2) FIG. 1 shows an example of line segment map data. As shown in FIG. 1, the line segment map data includes data 1 indicating the number of line segments, and coordinate pair data 3 of a start point and an end point of each line segment. Each line segment need not be a polygon or polyline.

3. Output data: polygon road network data In the present embodiment, the following polygon road network data is created by performing a road network recognition process on the line segment map data.

(1) FIG. 2 shows an image example of a polygon road network represented by polygon road network data. As shown in FIG. 2, the polygon road network data includes a set of node polygons 5, 5,... Representing intersections and a set of link polygons 7, 7,.

(2) The node polygon 5 representing an intersection is a specified radius having a center at the specified point (for example, about twice the maximum road width) when one point is specified inside the intersection (as close to the center as possible). In the end point set having the end points in the area of the circle described below or consisting of the points of the perpendicular legs of the line passing through the circle area, only the above designated points are included as true internal points. "Closed". Here, the “inner convex hull” is a term specific to this specification, and simply means a minimum convex hull composed of end points near the designated point. Where "convex closure"
Is a polygon such that a line segment connecting any two points on the inside or on its side is completely included in the area inside the polygon,
In simple terms, it refers to a convex polygon whose vertices are all outward (ie, have no concave vertices).

(3) FIG. 3 shows the types of node polygons. As illustrated, the node polygons 5 can be classified into the following four types (dc is a type code).

Dc = 0: As shown in FIG. 3A, a road 17a passing through a frame (data boundary) 15 of the map 13
2 set at a position in contact with the data boundary 15
Crossroads 5a.

Dc = 1: As shown in FIG. 3 (b), a road 17b or a blind alley 17b leading to an open area such as a plaza.
A dummy fork intersection 5b representing the end of the intersection.

Dc = 2: As shown in FIG. 3 (c), a dummy two-junction intersection 5c is set in the middle of the curved road 17c in recognition processing.

Dc ≧ 3: As shown in FIGS. 4 (a) to 4 (c), intersections 5d to 5g at or above a normal three-way intersection.

The two-forked intersections 5a to 5 shown in FIG.
c is represented by a rectangular polygon whose intersection length in the road length direction is equal to the minimum road width WMIN.

(4) The node polygon 5 has the following information.

[Node polygon] int Node; / * Number of node polygons registered up to now * / struct NODE {int sts; / * Work variable * / int dc; / * Node polygon type code * / int x ; / * X coordinate of the center of gravity of node polygon * / int y; / * same as above y coordinate * / int z; / * same as above z coordinate (reserved) * / int np; / * number of vertices of node polygon * / int xp [NDPSIZE ]; / * X-coordinate sequence of node polygon vertices * / int yp [NDPSIZE]; / * Same as above y-coordinate sequence * / int nl; / * number of link sides (entrance side to road element) * / int link [NDLSIZE ]; / * Identification number of the link polygon connected to this link side * / int node [NDLSIZE]; / * Number of the tip node polygon connected to this link side * / int lpvs [NDLSIZE]; / * Start point of the link side Node polygon vertex number of * / int lpve [NDLSIZE]; / * Node polygon vertex number of link end point * / int ptr; / * Reserved: Link pointer with external related information * /} nodef [N ODESIZE]; / * Node polygon data file name * / Here, NDPSISE is the maximum number of end points of the node polygon, NDLSIZE is the maximum number of link sides of the node polygon (the maximum number of roads exiting from one intersection), NODESIZE is the maximum expected number of node polygon data.

(5) The link polygon 7 represents a road element (link) between adjacent intersections. Both ends of the link polygon 7 are associated with the corresponding node polygon 5 and have the following information.

[Link polygon] int Nlink; / * Number of link polygons registered up to now * / struct LINK {int sts; / * Work variable * / int dc; / * Link polygon type code * / int xF [* 2]; / * Minimum / maximum x coordinate of this link polygon existing rectangle * / int yF [2]; / * Same as above Minimum / maximum y coordinate * / int node [2]; / * Node polygon at both ends of link polygon Number * / int prev [2]; / * Link polygon number one before in the order of angles connected at both ends * / int next [2]; / * Same as above Link polygon number after one * / int len; / * Length of link polygon center line * / int w; / * Minimum value of link polygon inner width * / int dir; / * Directed / non-directional code dir = 0,1,2,3 * / int nc; / * Number of vertices in the center point sequence * / int xc [NLCSIZE]; / * x coordinate sequence in the center point sequence * / int yc [NLCSIZE]; / * y coordinate sequence in the center point sequence * / int wc [NLCSIZE]; / * Half width of (road) at center point sequence point * / int np; / * Number of vertices of link polygon * / int xp [NLPSIZE]; / * x coordinate sequence of link polygon vertices * / int yp [NLPSIZE]; / * Same as above y coordinate sequence * / int zp [NLPSIZE]; / * Same as above z coordinate sequence (reserved) * / int vsp [2]; / * Link polygon start vertex number corresponding to link side * / int vep [2]; / * Same as above End point number * / int parm [4]; / * Reserved: Parameter storage area * / int ptr ; / * Reserved: Link pointer to external information * /} linkdf [LINKSIZE]; / * Link polygon data file name * / Here, NCLSIZE is the maximum number of end points of the center line of one link polygon, NLPSIZE is 1 The maximum number of link polygons, LINKSIZE, is the maximum expected number of link polygons.

(6) A set of node polygons 5 and a set of link polygons 7 represent a road network.
That is, an arbitrary node polygon 5 has information on the number of link polygons 7 connected to the node polygon 5 and the identification numbers of these link polygons 7.
An arbitrary link polygon 7 is assigned identification numbers of the node polygons 5 at both ends thereof. Therefore, the connection relationship between the node polygon 5 and the link polygon 7 represents the relationship between nodes and links in graph theory, and can be said to represent a "road network". As shown in FIG. 2, not only the road network consisting of nodes and links polygonized in this way, but also the point “point” using only the information of the center of gravity 9 of the node polygon 5 and the center point sequence 11 of the link polygon 7. A conventional road network using the nodes and links of "lines" is also obtained.

(7) In the present embodiment, the link between the intersections is a non-directional link having no direction (information indicating which intersection to go to), but this is changed to a directional link as follows. It can be extended. In other words, the direction opposite to the forward direction is defined, one non-directional link is divided into a forward link and a reverse link, and the link polygon is divided into a forward link polygon and a reverse link polygon by, for example, the center line thereof. .

(8) When it is necessary to integrate a plurality of map data, the integration of adjacent map data is performed using the virtual node polygon 5a at the data boundary as shown in FIG. 3A as an interface. It is possible to do.
In addition, by dividing one wide-area map data into a plurality of narrow-area map data and setting a virtual node polygon 5a at a data boundary of a road passing through a boundary between the narrow-area map data, the connection information is obtained. Is possible.

4. Detailed Procedure of Road Network Recognition Process FIG. 5 is a flowchart of the road network recognition process. The road network recognition process includes the following steps.

4.1 Step S1: Reading line segment map data.

4.2 Step S2: Designation of one point on the road.

4.3 Step S3: Detect initial node polygon.

4.4 Steps S4 and S6: Processing control.

4.5 Step S5: End processing.

4.6 Step S7: Link polygon detection.

4.7 Step S8: Registered check.

4.8 Steps S9 to S12: Registration of node polygons and link polygons.

4.9 Return to the loop.

The specific processing contents of these steps will be described below.

4.1 Step S1: Reading line segment map data. First, the line segment map data as shown in FIG. 1 is read,
Based on this, a map image is drawn and displayed on the display screen. As described above, the line segment map data includes the following information. Nvec: Number of line segment data (xs (I), ys (I), xe (I), ye (I)) I = 0, 1,..., Nvec-1: start-end point coordinate pair data of each line segment (Xs, ys) is the starting point coordinates of the line segment, (xe,
ye) is the end point coordinates.

(1) The start and end point coordinates of the line segment are integer data representing the position coordinates in the rectangular coordinate system in the present embodiment. However, this is not an essential condition, and real number data may be used. When used for calculation of computational geometry, almost all real numbers are used.

(2) When reading the line segment data, the minimum and maximum values of the x and y coordinates of the end point of the line segment data are calculated. These are set as XS: x minimum value, YS: y minimum value, XE: x maximum value, YE: y maximum value. Therefore, the map range from the minimum coordinate value (XS, YS) to the maximum coordinate value (XE, YE) is the processing target range. These values are used for determining an intersection or a road on a boundary and controlling display.

4.2 Step S2: Designation of one point on the road. The present embodiment does not incorporate a function for a computer to automatically recognize which part is a road or a building from map data of a simple line segment set. As an alternative, the line segment map data is displayed on the screen, and the human operator "points that have been reliably recognized as" one point on the road ""
Is specified by a pointing device or the like. That is,
Initial values are created using the human recognition function. FIG. 6 shows a case where the operator displays “1 on the road” on the map 21 displayed on the screen.
An example in which a point "23" is designated is shown. Computer
"The coordinates (xv, yv) of one point specified by the operator are
Surely on the road. ", The following processing is performed. If the designated point is not on the road, the following processing will stop immediately.

4.3 Step S3: Initial node polygon detection. Based on one point designated by the operator, assuming that the designated point is within the intersection, an intersection polygon including the designated point is detected. At this time, different node polygons are formed depending on whether the designated point is actually inside the intersection as described above or in a road element that is not an intersection unlike the above assumption. The procedure for detecting a node polygon is as follows (1) to (8).

(1) As shown in FIG.
, A rectangular search area 25 having a width twice as large as WMAX (maximum road width) in the x direction and a length twice as large as WMAX in the y direction is defined. Then, a line segment having at least one end point in the rectangular search region 25 or passing through the rectangular search region 25 (that is, a line segment at least partially existing in the rectangular search region 25) is listed. In the case of the example of FIG. 6, for example, as shown in FIG.
Is listed. In the following processing, among the listed line segments 27, 27,... Having a portion outside the search region 25, a new intersection point between the line segment and the outer frame of the search region 25 is set. End points (indicated by double circles in FIG. 7) 29, 29,.
29, the outside part is regarded as another line segment. Accordingly, the following processing is performed only on the line segment completely existing in the search area 25.

(2) Next, from the line segments 27, 27,... In the search area 25 listed in (1), the designated point 23 is set as a viewpoint, and a fixed rotation direction ( For example, while rotating in a counterclockwise direction and looking around the entire circumference, a visible line segment at least partially visible from the viewpoint 23 (that is, a straight line extending from the starting point to any point on the line segment, Only the line segments that can be drawn without touching the minute lines) are extracted, and the end point coordinates of these visible line segments are registered in the work area in the order of the rotation direction. In the case of the example of FIG. 7, for example, as shown in FIG.
Are registered, and the end point coordinates of the visible line segments 31, 31,... Are registered in the order of the rotation direction, thereby defining the visible line segment polygon 33 as shown by the thick line.

(3) Further, a perpendicular line is drawn from each end point of the visible line segment to another visible line segment, and if the leg of the perpendicular line is on the visible line segment (that is, not on an extension of the visible line segment, , If it is on the visible line itself), the coordinates of the foot of the perpendicular are also registered as additional end points in the work area. For example, as shown in FIG. 8, when a perpendicular line is dropped from a certain end point 35 to another visible line segment, perpendicular legs 37 and 39 are obtained on another visible line segment, and the perpendicular legs 37 and 39 are obtained. Are additionally registered in the work area (the same applies to the leg of the perpendicular drawn from another end point). Thus, in addition to the original endpoint of the visible line segment,
If the legs of the above-mentioned perpendicular line are also registered as additional end points, when the node polygon is created in the next process (4), the node polygon may have an excessively large size (that is, the node polygon may not extend further from the intersection to the road element). (A natural shape), and a node polygon having a minimum size and a natural shape that does not include a road element but expresses only an intersection is easily obtained.

(4) With respect to the set of end points registered in these work areas, an “internal convex hull” having the designated point (viewpoint) 23 as the only internal point is obtained. Here, the “inner convex hull” is a term specific to this specification, as described above, and simply means a minimum convex hull composed of end points near the designated point 23. . Here, a “convex closure” is a polygon in which a line segment connecting any two points on the inside or on a side is completely included in an area within the polygon, in short, all vertices Refers to a convex polygon that is outward (that is, has no concave vertex).

The internal convex hull having the designated point 23 as the only internal point can be mathematically obtained as follows.

4a) Among the points registered in the work area, the point 4 closest to the designated point 23 as shown in FIG.
1 is found, and a distance r0 from the designated point 23 to its nearest point 41 is obtained. A circle having the specified point 23 as the center and the radius r0 as the radius to the end point 41 closest to the specified point 23 is defined.

4b) A circle having the specified point 23 as the center and the radius r0 from the specified point 23 to the nearest point 41 as a radius is defined. Then, each of the points registered in the work area is
The image is mapped inside the circle according to the reciprocal of the distance from the designated point 23. For example, as illustrated in FIG. 9A, it is assumed that, of the points P0 to P15 registered in the work area, the point P0 is the nearest point, and the distance between the nearest point P0 and the designated point 23 is r0. Each point Pi (i = 0 to 15) is designated by
When mapping is performed in accordance with the reciprocal of the distance ri to 3, FIG. 9B
As shown in the figure, the mapping point P′i of each point Pi is
(R is a constant, for example, r0). Point Pi is designated point 2
The further away from 3, the closer the mapping point P′i is to the designated point 23. From these mapping points P′i, a convex hull 51 that is a polygon having the minimum number of short points and including all the mapping points P′i is determined.

4c) Next, this convex hull 51
= R × r0 / r′i and inversely mapped to the original coordinates as shown in FIG. 9C. The polygon 53 formed by inversely mapping the convex hull 51 in this manner is an internal convex hull having the designated point 23 as the only internal point, that is, a node polygon including the designated point 23 (hereinafter, referred to as an initial node polygon).

The processes 4a) to 4c) are intuitively similar to the operation of selecting the smallest convex polygon from points near the designated point 23. In the case of the example shown in FIG. 8, by performing the above-described processing on the end points of the visible line segment polygon 33, the initial node polygon 61 accurately representing the shape of the intersection is obtained as shown by the thick line in FIG. Can be

The processing for obtaining the initial node polygon described above is implemented as a function GetNodePolygon (...) in the present embodiment. The input / output data of this function GetNodePolygon (...) is as follows.

[Input data] (xv, yv): coordinates of specified point (viewpoint) WMIN: minimum road width WMAX: maximum road width [output data] resp: response code resp <0: no node polygon was obtained ( Resp ≧ 0: node polygon is found np: number of end points of node polygon elist []: line segment number to which end point belongs (xp “”, yp “”): end point coordinates of node polygon Column (xG, yG): Coordinate of the center of gravity of the node polygon nlink: Number of link sides of the node polygon (number of entrances of road candidates) (lpvs [], lpve []): End point number pair of link side lpvs [i]: Link Start point end point number of side i lpve [i]: End point end point number of link side i However, the link side (described later) is equal to or smaller than the minimum road width value WMIN. In a condition to have a road width maximum value WMAX * 1.1 or less in length.

(5) Next, as exemplified in FIG.
The center of gravity 63 of the initial node polygon 61 is calculated. next,
Attention is paid to the sides 65a to 65g of the initial node polygon 61 in the counterclockwise order, and when the starting point of each of the noted sides 65a to 65g is 0 and the ending point is 1, the ratio becomes 0.4 to 0.6. (In short, in a range excluding the vicinity of both ends of each side), a rectangular area 6 having an appropriate vertical width (for example, 1 to 2 times the minimum road width WMIN) on each side 65a to 65g.
7a to 67g are set, and it is checked whether or not there is a solid line segment (that is, a line segment of map data) in each of the rectangular regions 67a to 67g. In the case of the example of FIG. 10, as can be seen in comparison with FIG.
There are no solid line segments in the rectangular regions 67b, 67d, 67f, and 67g, and there are solid line segments in the rectangular regions 67a, 67c, and 67e. The sides 65a, 65c, and 65e corresponding to the rectangular region having at least one solid line segment are determined not to be the entrance of the road element connected to the intersection (node polygon 61), and correspond to the rectangular region having no solid line segment. Side 65b,
It is determined that 65d, 65f, and 65g are candidates for a road element entrance connected to the intersection (node polygon 61). However, even if the side corresponds to a rectangular area having no solid line segment, the road width maximum value WMIN is not less than the road width minimum value WMIN.
Edges that do not satisfy the condition that they have a length of MAX * 1.1 or less are excluded from road element entrance candidates. Sides 65b, 65d, 65 of road element entrance candidates found in this determination
f and 65g are referred to herein as “link edges”.

The result data of the initial node polygon obtained as described above is stored in the node polygon data file as the node polygon in the “unprocessed link side” state.
Registration is performed as in the following 1) to 5).

1) A new node polygon number is assigned to the initial node polygon.

2) The barycentric coordinates (xG, yG) are stored as intersection representative points.

3) The number of end points np of the node polygon and the end point coordinate point sequence data (xp "", yp "") are registered.

4) The number of link sides nlink, the end point number (lp
vs [i], lpve [i]).

5) A value -1 meaning unprocessed is registered in the link destination node number and link polygon number of each link side, and the processed number of link sides is initialized to zero.

(6) Next, the number nlink of the link sides of the registered initial node polygon is checked. nlin
If k = 0, it means that the initial node polygon has no road entrance candidate. This is an error return because the designated point is incorrectly designated to a part other than the road (for example, inside a house).

If nlink = 1, the initial node polygon corresponds to, for example, a dead end point of a dead end shown in FIG. 3B or an end (open end) of a road connected to an open land such as a plaza.

If nlink = 2, this corresponds to a case where one point is designated not in an intersection but in a road element.

If nlink ≧ 3, there is a high possibility that the intersection is an ordinary three-way intersection or more as shown in FIG.

When 0 <nlink ≦ 2, as described above, it is estimated that the initial node polygon exists on a road portion other than the intersection. In this case, as shown in FIG. 11, a rectangle (four apexes) polygon 71 as a two-junction dummy intersection having a width equal to the minimum road width WMIN in the road traveling direction centering on the designated point 23 is additionally provided. Generate. The two-junction dummy node polygon 71 is shown in FIG.
A curved road as shown in FIG.
It has an important role as an object for expressing the end point of the road or the open end of the road that contacts the boundary as shown in FIG.

(7) Number of end points np of initial node polygon
Is checked, and if np ≧ 3 and nlink ≧ 3, the above result is determined as the initial node polygon.

In the case of nlink = 2, "turning angle"
Therefore, it is determined whether or not the vehicle is at a corner by the following method. That is, as shown in FIG. 12, from the center of gravity of the initial node polygon 81, each link side 85a, 85b
Calculate the direction angles of the line segments 87a and 87b toward the center of
If the difference 89 in the direction angle (180 degrees or more, the angle obtained by subtracting the angle from 360 degrees) is within a predetermined threshold ACMAX (for example, within 150 degrees), it is determined to be a turning angle. Is determined to be a road close to. If it is determined that the corner is a corner, the initial node polygon 81 is fixed in its original shape. If it is determined that the corner is not a corner, the initial node polygon 81 is converted from the original shape into the two-way node polygon 71 shown in FIG.
Replace with and confirm.

(8) According to the above-described method, any node polygon candidate data can be obtained by designating one point on the road, not at the intersection.

4.4 Steps S4 and S6: Processing control. (1) A node polygon having an unprocessed link side is searched from the node polygons registered so far in step S4. If there is no node polygon having an unprocessed link side, the process exits the loop and jumps to “end processing” in step S5.
If there is no node polygon with unprocessed link side,
Ideally, all intersections and road element polygons included in the road connected to the designated point have been registered.

(2) If there is a node polygon having an unprocessed link side, the flow advances to step S6 to determine the node number node and the unprocessed link side number m of the node polygon, and the flow advances to step S7. At the stage when the initial node polygon has just been registered, the initial link polygon naturally has an unprocessed link side.
Will go to.

4.5 Step S5: End processing. The number of generated node polygons, the number of link polygons,
The number of link sides, the number of link sides that have been processed normally among the number of link sides, the number of link sides for which no solution was found, and the like are totaled, and the road network recognition processing is stopped. In addition, the result of the road network recognition process, that is, the node polygon file and the link polygon file, may be output to the outside before the process is stopped. After the processing is stopped, if a human operator sees an unprocessed road area on the display, the recognition processing is restarted by additionally designating one point in the road area. You can also.

4.6 Step S7: Link polygon detection. (1) The start and end point numbers (lpvs [m], lpve [m]) of the unprocessed link side of the number m of the node polygon of the number node detected in steps S4 and S6 are extracted, and the link side is obtained. End point coordinates (xa, ya)
And (xb, yb). Also, the coordinates of the center of gravity (xG, yG) of the node polygon are taken out, and this is taken as (xs
G, ysG).

(2) In the direction from the center of gravity (xsG, ysG) to the start point (xa, ya) of the link side to the end point (xb, yb), the detection of the adjacent node polygon and the link polygon and A center line candidate is detected. The specific processing procedure is as follows 1) to 11).

1) As shown in FIG. 13, from the center of gravity (xsG, ysG) 93 of the node polygon 91, the midpoint of the unprocessed link side 95 (xv = (xa + xb) / 2, yv
= (Ya + yb) / 2) A rectangular search area 97 having a length DS and predetermined left and right widths WL and WR is set.

2) As shown in FIG. 13, all of the line segments 99 having at least one end point in the rectangular search area 97 or passing through the area 97 are listed.

3) As shown in FIG. 14, using the midpoint (xv, yv) 96 of the unprocessed link side 95 as a viewpoint, look counterclockwise from the start point (xa, ya) to the end point (xb, yb). Set of line segments listed in 2) (Fig. 1
, Are extracted from the midpoint 96, and the end points of the visible line segments 101, 101,... Are rotated counterclockwise. Sort to

4) The visible line segments 101, 1 extracted in 3)
01, ..., or the end point thereof, is set to the center of gravity (xsG, ysG) 93
From the viewpoint (xv, yv) in the 96 direction, as shown in FIG.
.. And left line segment strings 101L, 101L,. Specifically, a line segment sequence corresponding to a part before the end point farthest from the viewpoint (xv, yv) 96 is selected from the end point sequence sorted in the counterclockwise direction by right line segment sequences 101R, 101R,
.., The line segment sequence corresponding to the portion after that is referred to as a left line segment sequence 10.
1L, 101L,...

5) Right line segment sequence 101R, 101R,...
, And the left line segment sequence 101L, 101L,..., The center point (xv, y) is measured from the center of gravity (xsG, ysG) 93.
v) In the direction toward 96, select the closest first endpoint, then select the second endpoint closest to the first endpoint from among the endpoints farther than the first endpoint, and then select , A third end point closest to the second end point is selected from the end points farther than the second end point, and the process of... Is repeated until the end point farthest to the far end is reached. Lines of the first to last endpoints on the right side thus selected are held as “right endpoint columns”, and columns of the first to last endpoints on the left side are held as “left endpoint columns”. Lines belonging to these left and right endpoint columns are retained. The minute number is also held at the same time. By this processing, the end points 103R and 103 entering the side road shown in FIG.
R and 103L are removed from the left and right end point sequences. As a result, at the end points of the left and right contours of one road element extending from the link side 95, and where there is a side road in the middle,
Point sequence 1 on the left and right where end points to be the entrances of the side road are connected in order
05R, 105R,... And 105L, 105L,.

6) As shown in FIG. 16, in each of the left and right end point rows 105R, 105R,... And 105L, 105L,..., Whether the first end point and the second end point are both ends of the same line segment. Check, if yes, go to the next third endpoint, otherwise (ie, if there is a gap between the two endpoints), calculate the distance between these two endpoints (the length of the gap) and If the distance is larger than the minimum road width WMIN, there is a possibility that a side road may come out of the gap between the two end points, so this gap is listed. Similarly, it is checked whether or not a space between two adjacent end points is a space that may correspond to an entrance of a side road, and if so, the space is added to the list. This detection process is performed until the end of the left-right end point sequence,
Upon completion, as shown in FIG. 16, the gaps 107R, 107R, 107L, and 107C of the side street entrance candidates are detected. These detected voids 107R, 107R, 107
L and 107C are created by sorting data in order of distance from the viewpoint (xv, yv) 96.

7) In 6), if no air gap is detected at all, the road element is, for example, a blind lane or the like. Therefore, a two-forked road dummy node polygon with dc = 1 (FIG. Node polygon 5b) shown in b)
Generate

8) In 6), when the gaps 107R, 107R, 107L and 107C are detected as shown in FIG. 16, the gaps 107R, 107R and 107C are detected.
L, 107C in order of decreasing distance from the viewpoint 96,
Test for possible intersections. Specifically, FIG.
As shown in the figure, the gaps 107R and 107R in the right end point row are perpendicular to the left end point row from the midpoint of the gap line segment.
09, 109, and the midpoints 111, 111 of the perpendicular line
Is set as a viewpoint (xu, yu), and a node polygon is detected. Similarly, for the gap 107L in the left end point sequence, the perpendicular line 113 dropped from the midpoint of the gap line segment to the right end point sequence
The viewpoint is set at the midpoint 115, and the node polygon is detected. Regarding the void 107C formed by one point of the right end point sequence and one point of the left end point sequence, the viewpoint is set to the midpoint 117 of the line connecting the two points, and the node polygon is detected.

The detection of the node polygon is performed in the same manner as the detection of the initial node polygon described above, that is, the function GetNod
This is performed using ePolygon (...). That is, the above “4.3.
Step S3: Detection of initial node polygon. May be read as the initial node polygon as the node polygon and the designated point as the viewpoint.

9) In step 8), the node polygon is started to be detected from the sideway entrance candidate gap closest to the link side midpoint 96. If no node polygon can be detected there, the node polygon is detected with respect to the next nearest sideway entrance candidate gap. Try to detect. As a result of continuing this, if no node polygon can be detected in any of the gaps, an error return is made. Usually, some node polygons including the two-junction dummy node polygons are detected. In the case of the example shown in FIG. 16, as shown in FIG. 17, a node polygon 119 indicated by a thick line is detected at the nearest sideway entrance candidate gap 107R. If this road element has no side road at all, a two-way intersection dummy node polygon (not shown) is located at a gap 107C between the farthest point in the right end point sequence and the farthest point in the left end point sequence. ) Will be detected.

The node polygon 119 thus detected
Is called an "end node polygon" in the sense of the end point of one road element. Also, the node polygon 91 and the link side 95, which are the starting points of the above processing, will be referred to as a "departure node polygon" and a "departure link side", respectively.

10) Next, the final node polygon 11
It is determined which of the link sides of the starting node polygon 9 corresponds to the starting link side 95 of the starting node polygon 91 (whether they are connected by one road element). That is, as shown in FIG.
1a, 121b, and 121c are detected, and, for example, the link side 121a having the shortest distance between the midpoint of the link sides 121a, 121b, and 121c and the midpoint (xv, yv) 96 of the starting link side 95 Is determined, and the selected link side 121a is determined as the link side corresponding to the departure link side 95 (referred to as “end link side”). When the end link side 121a of the end node polygon 119 is determined in this way, the midpoint (xv, yv) 96 of the start link side 95 is next determined.
, And sequentially traces the right and left end point columns in a counterclockwise direction. At this time, the end link side 1a is folded back at the end link side 121a, that is, the end link side 1
21a, the tracking is performed in the order of moving to the left end point column through the end link side 121a. Thus, the start point (xa, ya) of the departure link side 95, the end point closest to the departure link side 95 in the right end point sequence, each end point in the right end point sequence, the end point closest to the end link side 121a in the right end point sequence, the end link Right end point (end point) of side 121a, end link side 1
The end points are picked up in the order of the left end point (start point) of 21a, the end point closest to the end link side 121a in the left end point sequence, each end point in the left end point sequence, and the end point (xb, yb) of the departure link side 95. If these end points are connected in FIG.
As shown in (1), a link polygon 123 indicated by a bold line, which represents a road element connecting the start node polygon 91 and the end node polygon 119, is completed.

11) Next, a center point sequence of the completed link polygon is produced. As shown in FIG. 18, the center point sequence is prepared by first forming a straight line segment 1 from the midpoint 133 of the departure link side 131 to the midpoint 137 of the end link side 135.
39, and each end point 143a-
From 143d, 143f, 143g, a perpendicular is drawn to this straight line segment 139 (however, in this example, the end points 143e and
From 3h, the perpendicular cannot be dropped on the straight line segment 139), and the feet 145a to 145d, 145f, and 145g of those perpendiculars
Is stored as the position ratio of the line segment 139 (0 at the start point 133 and 1 at the end point 137). Next, the perpendicular legs 145a to 145d, 145f, and 145g are sorted in ascending order of the position ratio data, that is, in the order closer to the starting point 133, and the perpendicular legs 145a to 145g are sorted in the sorted order.
End points 143a-1 corresponding to 5d, 145f, and 145g
As shown in FIG. 19, the perpendicular lines 147a, 147g, 147 from 43d, 143f, 143g to the other end point row
b, 147f, 147c, and 147d. And
The center of gravity of the starting node polygon 151 is defined as the starting point 153, the center of gravity 157 of the ending node polygon 155 is defined as the ending point, and between the starting point 153 and the ending point 157, the midpoint 1 of the perpendicular in the above order.
49a, 149g, 149b, 149f, 149c, 1
49d, these point sequences 153, 149a, 14
9g, 149b, 149f, 149c, 149d, 15
Let 7 be the center point sequence of the link polygon. Lastly, the sum of the lengths of the distances between the points along the center point sequence is obtained, and the sum is used as the total length of the link polygon 141.
a, 147g, 147b, 147f, 147c, 147
The minimum value of the length of d is obtained, and this is
1 minimum width.

The above processes 1) to 11) are implemented as a function GetLinkPolygon (...) in the present embodiment. The input / output data of this function GetLinkPolygon (...) is as follows.

[Input Data] (xsG, ysG): Coordinates of the center of gravity of the starting node polygon (xa, ya), (xb, yb): End point coordinates of the starting link side DS: Length of rectangular search range WS: Rectangular search range Width (= WR + WL) WMIN: Minimum road width to be detected WMAX: Maximum road width to be detected [output data] resp: Processing result code resp <0: No solution was obtained (a negative value indicates an error code ) Resp> = 0: solution is found dc: type code of end node polygon nd: number of end points of end node polygon (xd "", yd ""): end point polygon end point coordinate sequence (xdG, ydG): Coordinates of the center of gravity of the final node polygon nld: Number of link sides of the final node polygon (lpvsd [], lpved []): Starting point / ending point end point number of link side lid: number of ending link side lengthd: total length of link polygon (based on center point sequence) wd: minimum width of link polygon (based on center point sequence) nc: number of points of link polygon center point sequence (Xc [], yc []): Coordinate sequence of center point sequence of link polygon nq: Number of end points of link polygon (xq [], yq []): Coordinate sequence of end point of link polygon (vsp [], vep [] ): End point number pairs of sides corresponding to the start node polygon and the end node polygon of the link polygon (that is, the start link side and the end link side)

(3) Processing when a link polygon cannot be obtained A code “0” indicating “No solution was found” is stored in the departure link side of the departure node polygon, and the number of processing link sides of the departure node polygon is stored. Is incremented by one.

4.7 Step S8: Registered check. When a link polygon is obtained, it is checked whether or not the obtained end node polygon is already registered in the road network data. This processing is implemented as a function CheckNetwork (...). This function CheckNetwork (...) checks whether there is already another node polygon in the obtained link polygon. If so, this function determines that the node polygon is closer to the origin node polygon, corrects the data of the node polygon and the link polygon up to that point, and returns a value. This function CheckNetwork (…)
Are as follows.

[Input data] node: Departure node polygon number (xsG, ysG): Coordinate of center of gravity of departure node polygon (WMIN, WMAX): Road minimum width, maximum width Output data of function GetLinkPolygon (...) [Output data] resp0 : Processing result code resp0 <0: Neither the current end node polygon nor the link polygon is registered in the existing network. It is a new node polygon. resp0 = 0; another node polygon has already been registered at a position closer to the end node polygon. That number is respNode. However, since there is no link polygon up to that node, a link polygon up to that node was created and a value was returned. resp0> 0: Another node polygon already exists at a position closer to the end node polygon, and the link polygons up to that point have already been registered. The node polygon number is respNode, and the link polygon number is respLink.

4.8 Steps S9 to S12: Registration of node polygons and link polygons. (1) Processing of resp0 <0 1) A new node polygon number and a new link polygon number are obtained, and the detected end node polygon data and link polygon data are stored in the storage area. It is assumed that a departure node number and a link polygon number are stored in a link side of the storage destination node polygon that is connected to the departure node polygon, and that one link side has been processed.

2) The end node polygon number and the link polygon number of the departure node polygon corresponding to the currently detected end node polygon and link polygon are entered, and the number of processes on the link side is increased by one. deep.

(2) Processing of resp0 = 0 A new link polygon number is obtained, and the obtained resp
The node polygon data of the Node stores the connection information with the departure node polygon and one set of link polygon data. The node number and link number of the connection destination are stored in the corresponding link side of the departure node polygon.

(3) Processing of resp0> 0 The starting node polygon data is overwritten with the node number and link number obtained as existing nodes.

4.9 Return to the loop.

After the processing of 4.8, “4.4 Step S
4, S6: Processing control. Return to.

By performing the above-described road network recognition, polygon roads represented by node polygons 5, 5,... And link polygons 7, 7,. Network data is obtained. Actually, depending on the connection of the line segments in the line segment road map data and the connection status of the roads, not all roads in the processing target range may be detected in one recognition process. In this case, a person newly designates one point of the undetected road part,
The recognition process is restarted, and the recognition result is added.

The embodiment of the present invention has been described above.
These embodiments are merely examples for describing the present invention, and are not intended to limit the present invention only to these embodiments. Therefore, the present invention can be implemented in various modes other than the above-described embodiment. For example, a computer may automatically designate a point on a road, instead of a person.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of line segment map data.

FIG. 2 is a diagram showing an example of a polygon road network represented by polygon road network data.

FIG. 3 is a diagram showing an example of a node polygon.

FIG. 4 is a diagram showing another example of a node polygon.

FIG. 5 is a flowchart of a road network recognition process.

FIG. 6 is an explanatory diagram showing an example of one designated point in a road and a set search area.

FIG. 7 is an explanatory diagram showing an example of a result of extracting a line segment in a search area.

FIG. 8 is an explanatory diagram showing an example of a result obtained by extracting visible line segment polygons from a designated point as a viewpoint;

FIG. 9 is an explanatory diagram showing a procedure of a mapping for obtaining an “internal convex hull” (node polygon) and a reverse mapping.

FIG. 10 is an explanatory diagram showing a procedure for obtaining a link side of a node polygon.

FIG. 11 is an explanatory diagram showing an example of a two-junction dummy node polygon set at a portion other than an intersection.

FIG. 12 is an explanatory diagram showing a procedure for determining whether or not the vehicle is at a corner;

FIG. 13 is an explanatory diagram showing a procedure of setting a rectangular search area for detecting a link polygon and picking up a line segment in the rectangular search area.

FIG. 14 is an explanatory diagram showing a procedure for extracting a visible line segment for detecting a link polygon.

FIG. 15 is an explanatory diagram showing a procedure for creating left and right end point sequences for detecting link polygons.

FIG. 16 is an explanatory diagram showing a procedure for detecting a gap that is a candidate for a side road entrance from the left and right end point columns and setting a new viewpoint.

FIG. 17 is an explanatory diagram showing a procedure for creating a terminal node polygon in a recent gap.

FIG. 18 is an explanatory diagram showing a procedure for creating a link polygon between a departure node polygon and an end node polygon.

FIG. 19 is an explanatory diagram showing a procedure for sorting end points of a link polygon in order to detect a center point sequence of the link polygon.

FIG. 20 is an explanatory diagram showing a procedure for linking the centers of gravity of the departure node polygon and the end node polygon with a center point sequence of link polygons.

[Explanation of symbols]

 5 Node polygon 7 Link polygon 23 Designated point on road 27 Visible line segment (first visible line group) 51 Convex closure 53 Internal convex closure (node polygon) 61 Node polygon 63 Center of gravity of node polygon 71 Two-junction dummy node polygon 81 Node polygon 83 Node polygon centroid 91 Departure node polygon 93 Departure node polygon centroid 101 Visible line segment (second visible line group) 107 Gap between side street entrance candidates 119 End node polygon 123 Link polygon 141 Link polygon 149 Link polygon Center point 151 Departure link polygon 153 Center of gravity of departure link polygon (node) 155 End link polygon 157 Center of gravity of end link polygon (node)

Claims (11)

[Claims] 1. An initial viewpoint specifying means for initially specifying a point existing on a road on a line segment map data as a viewpoint, and a first visible line segment group visible from the viewpoint is detected and detected. A first polygon creating means for creating a polygon representing a part of a road by using one visible line segment group; selecting one starting polygon from the already created polygons; , The side corresponding to the entrance of the other road part is extracted as the link side,
Selecting one departure link side from the link sides, detecting a second visible line segment group visible outside the departure polygon from a point other than an end point on the departure link, and using the second visible line segment group; The second polygon creating means and the second polygon creating means for creating at least one new polygon representing at least one other road portion that is continuous with the departure polygon are repeatedly operated, whereby a large number of continuous polygons are formed. A polygon road network data generating apparatus, comprising: a repetition means for generating and generating polygon road network data. 2. An initial viewpoint designating means for initially designating a point existing on a road on the line segment map data as a viewpoint, a first visible line segment group visible from the viewpoint, and A node polygon creating means for creating a node polygon meaning one intersection by using one visible line segment group; selecting one starting node polygon from already created node polygons; , A side corresponding to the entrance of another road portion is extracted as a link side, one departure link side is selected from the link sides, and the departure polygon is selected from a point other than an end point on the departure link side. A second visible line segment detecting means for detecting a second visible line segment group which is visible outside of the image forming device; and one of the one adjacent to the departure node polygon based on the detected second visible line segment group. End node polygon creating means for creating a new node polygon representing an intersection, and using the new node polygon as an end node polygon; based on the second visible line group, the departure node polygon and the end node polygon Link polygon creating means for creating a link polygon meaning a road element connecting the above, the second visible line segment detecting means, the terminal node polygon creating means and the link polygon creating means are repeatedly operated, whereby A polygon road network data creating apparatus, comprising: a repetition means for creating a large number of continuous node polygons and link polygons to create polygon road network data. 3. A node point is set in each node polygon, and a link line segment passing through each link polygon is set so as to connect each pair of adjacent node points. 3. The apparatus according to claim 2, further comprising: means for setting link segments to create road network data of points and lines. 4. The node polygon creating means maps an end point group of the first visible line segment group such that an end point closer to the viewpoint is located farther from the start point, and the mapped end point A convex hull of a group is created, the convex hull is inversely mapped by returning the mapped end point group to the original arrangement, and a polygon corresponding to the inversely mapped convex hull is defined as the node polygon. An apparatus according to claim 2. 5. The destination node polygon creating means detects one point having a possibility of being within an intersection adjacent to the departure node polygon based on the second visible line segment group, and determines the detected one point. 3. The apparatus according to claim 2, further comprising a new viewpoint designating unit for newly designating the viewpoint, wherein the node polygon creating unit is operated for the newly designated viewpoint to create the final node polygon. 6. The new viewpoint specifying means divides the second visible line segment group into a right line segment sequence and a left line segment sequence, and selects one of the right line segment sequence and the left line segment sequence. From the line segment sequence, one gap corresponding to the sideway entrance closest to the departure link side is detected, one point is selected between the gap and the line segment sequence on the other side, and the selected point is newly set as the viewpoint. 6. The apparatus according to claim 5, wherein 7. When the new viewpoint designating means cannot detect a gap corresponding to the sideway entrance from either the right line segment sequence or the left line segment sequence, the new viewpoint designating means sets the right line segment sequence within a predetermined range. Detects a gap between the farthest end point from the departure link side and the farthest end point from the departure link side in the left line segment sequence, selects one point near the gap, and newly designates the selected point as the viewpoint 7. The device of claim 6, wherein 8. An initial viewpoint specifying step for initially specifying a point existing on the road on the line segment map data as a viewpoint, and detecting a first visible line segment group visible from the viewpoint, A first polygon creation step of creating a polygon representing a part of a road using one visible ray group; selecting one departure polygon from already created polygons; , The side corresponding to the entrance of the other road part is extracted as the link side,
Selecting one departure link side from the link sides, detecting a second visible line segment group visible outside the departure polygon from a point other than an end point on the departure link, and using the second visible line segment group; A second polygon creation step of creating at least one new polygon representing another road portion that is continuous with the departure polygon and the second polygon creation step are repeated, thereby creating a large number of continuous polygons. A polygon road network data creating method comprising the steps of: 9. An initial viewpoint designating step of initially designating a point existing on a road on the line segment map data as a viewpoint, and detecting a first visible line segment group seen from the viewpoint. A node polygon creating step of creating a node polygon meaning one intersection using one visible line segment group; selecting one starting node polygon from the already created node polygons; , A side corresponding to the entrance of another road portion is extracted as a link side, one departure link side is selected from the link sides, and the departure polygon is selected from a point other than an end point on the departure link side. A second visible line group detection step of detecting a second visible line group that is visible outside of the image, and a next visible line group adjacent to the departure node polygon based on the detected second visible line group. Creating a new node polygon meaning one intersecting intersection; creating a new node polygon as the final node polygon; and forming the final node polygon based on the second visible line segment group. A link polygon creating step for creating a link polygon meaning a road element connecting to the end node polygon; and a step of detecting the second visible ray group, the end node polygon creating step, and the link polygon creating step are repeated. A repetitive step of creating a large number of continuous node polygons and link polygons to create polygon road network data. 10. An initial viewpoint specifying step of initially specifying a point existing on a road on the line segment map data as a viewpoint, detecting a first visible line segment group visible from the viewpoint, A first polygon creation step of creating a polygon representing a part of a road using one visible ray group; selecting one departure polygon from already created polygons; , The side corresponding to the entrance of the other road part is extracted as the link side,
Selecting one departure link side from the link sides, detecting a second visible line segment group visible outside the departure polygon from a point other than an end point on the departure link, and using the second visible line segment group; A second polygon creation step of creating at least one new polygon representing another road portion that is continuous with the departure polygon and the second polygon creation step are repeated, thereby creating a large number of continuous polygons. A computer-readable recording medium carrying a program for causing a computer to execute a repetitive step of creating polygon road network data. 11. An initial viewpoint designating step of initially designating a point existing in a road on the line segment map data as a viewpoint, and detecting a first visible line segment group seen from the viewpoint. A node polygon creating step of creating a node polygon meaning one intersection using one visible line segment group; selecting one starting node polygon from the already created node polygons; , A side corresponding to the entrance of another road portion is extracted as a link side, one departure link side is selected from the link sides, and the departure polygon is selected from a point other than an end point on the departure link side. A second visible line group detection step of detecting a second visible line group that is visible outside of the above, and based on the detected second visible line group, Creating a new node polygon meaning one adjacent intersection, and using the new node polygon as a final node polygon, a final node polygon creating step; based on the second visible line segment group, A link polygon creating step of creating a link polygon meaning a road element connecting to the end node polygon; and repeatedly performing the second visible line segment detection step, the end node polygon creating step, and the link polygon creating step. Accordingly, a computer-readable recording medium carrying a program for causing a computer to execute a repetitive step of creating a large number of continuous node polygons and link polygons to create polygon road network data.