JPH033083A - Drawing recognizing device - Google Patents

Abstract

PURPOSE:To recognize roads expressed in a map by analyzing a road non- intersection part and a road intersection part to find out feature values and tracing the connection between the non-intersection part and the intersection part while deciding the continuity of the road based upon the feature values. CONSTITUTION:The road non-intersection part and the road intersection part extracted from map data inputted from a map data inputting means 1 are analyzed. The feature values of the analyzed result are found out by a road tracking/recognizing means 4 and the connection between the non-intersection part and the intersection part is traced while deciding the continuity of the road based upon the feature values to recognize the whole road. Consequently, the road can be recognized while deciding the continuity of the road and the road can be distinguished from other similar patterns.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a drawing recognition device that converts roads marked on a map into data and recognizes them.

[Conventional technology]

FIG. 6 is a block connection diagram showing a conventional drawing recognition device. In the figure, 1 is a map data input means that inputs and stores map data, and 2 is a map data input means that analyzes the map data to identify road intersections and intersections. 3 is a road non-intersection extraction means for extracting road non-intersection parts between the road non-intersection parts; 3 is a road non-intersection extraction means for extracting road intersections based on the map data and the road non-intersection extraction results; 5 is a road non-intersection extraction means for extracting road non-intersection parts; This road graph structure recognition means analyzes the connection relationship between non-intersection areas and intersection areas, and recognizes the road as a graph structure with intersection areas as nodes and non-intersection areas as arcs. Next, the operation will be explained according to the flowchart shown in FIG. First, map data is input to the map data input means 1, and line segments forming the map are vectorized and stored (step 5TI). In this case, the vector should be in a format that represents the road contour rather than the road center line.For maps where roads are marked with center lines, it is sufficient to create a contour vector of an image of the center line. Next, the road non-intersection portion extracting means 2 extracts road non-intersection portions. That is, the longest one is extracted from among the vectors, and this is designated as vector (2) (step 5T2). Furthermore, a vector that is lined up almost parallel to the right side of vector V and lined up closest to it is determined, and this is set as vector W. Similarly, on the left side, the lines are parallel to each other,
Moreover, the 0th order to find the vectors that are lined up closest to each other,
If the interval between vectors ■ and W is about the width of the road, then vector v
, W are paired and tracked to generate a road center line (corresponding to the road center line). Similarly, the pair of vectors ■ and
). In this step ST4, as shown in FIG. 8, vectors V, , W. To the right, first the vectors V, , W, , then the vectors V3. We are tracking with W3. The pair is tracked in this way, and when the pair reaches a branch in the road and can no longer be tracked, the tracking is stopped. At this time, the terminal point of the skeleton line that has been created and the vector v that has just returned
, W or the tracking end points of vectors V and X, vectors A and B as shown in FIG. 9 are generated. This is called a "glue vector" (step 5T5). Here, the tracked vector is erased (step 5T6). Next, the presence or absence of a vector whose length is equal to or greater than the reference value is checked (step 5T7), and if found, the process returns to step ST2 to continue extracting road non-intersection areas; if not, the process returns to step ST81.
Go to step c and extract the road intersection. The road intersection referred to here includes road branch points such as T-junctions. In this step ST8, glue vector A is used. Examine the surroundings of B and, for example, if there is a closed loop consisting of a set of remaining vectors and glue vectors that did not become a non-intersection area, by detecting this closed loop, the road intersection area is extracted and the road intersection area is extracted. Connect the core lines that connect to the intersection within the intersection as shown in . Based on the connections of the road core lines obtained in this way, a graph structure is obtained in which non-intersection parts of the road are made into arcs and intersection parts are made into nodes, and the process is completed. In order to recognize such roads on the map, the process of step STI is performed by the map data input means 1, the process of steps ST2 to ST7 is performed by the road non-intersection part extraction means 2, and the process of step ST8 is performed by the road non-intersection part extraction means 2. The process of step ST9 is performed by the section extraction means 3 and the road graph structure P2 recognition means 5, respectively.

[Problem to be solved by the invention]

Conventional drawing recognition devices are configured as described above, so they only recognize roads as a graph structure, but cannot recognize a single road as a continuum of many intersections and non-intersection parts of the road between the intersections. In addition, sidewalks and hatching areas (
When there are road-like patterns such as areas shaded with diagonal lines or rivers, there are problems such as misidentifying them as roads. In addition, a technology similar to the conventional drawing recognition device was published in Osawa and Tamanai's "Improvement of core line generation method based on contour lines", IW-2 of the 37th National Conference of the Information Processing Society of Japan (late 1986). Are listed. This invention was made to solve the above-mentioned problems, and it is possible to recognize the road while determining the continuity of the road, and also to create a drawing that can distinguish it from other similar patterns. The purpose is to obtain a recognition device.

[Means to solve the problem]

The drawing recognition device according to the present invention analyzes road non-intersection parts and road intersection parts extracted from map data from a map data input means, calculates their feature quantities using a road tracking/recognition means, and also calculates the feature quantities. This system recognizes the entire road by tracking the connections between non-intersection areas and intersection areas while determining the continuity of the road.

[For use]

The road tracking/recognition means in this invention analyzes the previously extracted road non-intersection areas and road intersection areas, obtains their feature quantities, and determines the continuity of the road based on the feature quantities. The connection between a non-intersection area and a road intersection area is tracked to form a road, and the road marked on the map is recognized.

[Embodiments of the invention]

Embodiments of the present invention will be described below with reference to the drawings. In FIG. 1, 1 is a map data input means for inputting and storing map data, and 2 is a road non-intersection part extraction means for analyzing the map data and extracting road non-intersection parts between road intersections. , 3 analyzes the above map data,
A road intersection extracting means for extracting a road intersection; 4 indicates the connection between the extracted road non-intersection and the road intersection;
The road is tracked while determining its continuity as a road.
It is a road tracking/recognition method that recognizes two things. Next, the operation will be explained according to the flowchart shown in FIG. Note that the processes in steps STI to ST8 are the same as those in the conventional example described above, so a redundant explanation thereof will be omitted. That is, after extracting road non-intersection parts and intersection parts from the map data in steps STI to ST8, the road tracking/recognition means extracts the width, connection direction, connection position, and curve of the road non-intersection part connected to each intersection part. (step 5T9), then step 5
Track the road in TIO-3T17. That is, first, the widest non-intersection part of the road is
Step 5T10), then trace the road to the connecting intersection (Step 5TI).
I). Here, the initial tracking direction may be arbitrarily determined. Furthermore, check whether the current tracking was successful (step 5T)
12) If it cannot be tracked, it is a dead end of the road, so proceed to step 5T13. On the other hand, if it can be tracked, check the non-intersection part of the road that is connected to the intersection and has not been tracked as a road yet. The road is tracked in the direction that best maintains the continuity of the road (step 5T13). FIG. 3 shows the processing method of step 5T13. That is, FIG. 3(A) shows the most common four-way intersection, 8a to 8e are extracted road non-intersection parts, 8f is a road intersection part, and now from the road non-intersection part 8a to the right It is assumed that the vehicle is tracked and reaches a road intersection 8f. Next, the road non-intersection portion 8b connected to the road intersection portion 8f
~8e, the road direction with respect to the road non-intersection part 8a,
A road non-intersection portion 8d that is substantially continuous in width and connected to the road non-intersection portion 8a at a position in front of the road non-intersection portion 8a in the straight-ahead direction.
track towards. At this time, it is mistakenly extracted as a road non-intersection area, but it is actually a sidewalk and a road non-intersection area 8e.
In addition, hatched areas, rivers, long and narrow city blocks, etc. that are mistakenly extracted as road non-intersections are not tracked as roads. FIG. 3(B) shows road junctions, where 8g to 8. i
is the extracted road non-intersection area, 8j is the road intersection area, and if the tracking is now reaching the road intersection area 8j from the road non-intersection area 8g, the road non-intersection area 8j is the extracted road non-intersection area.
h has high continuity, but road non-intersection portion 81 has high continuity in terms of width. The tracking direction is determined based on predetermined criteria, taking these factors into consideration comprehensively. Figure 3 (C) shows a road pattern in which the way the curve is considered to be significant. Next, in the direction of the intersection 81, step 5T14 checks to see if the current tracking was successful. If the tracking has not been completed, the process proceeds to step 5T15. If tracking is successful, return to step 5TII and continue tracking. On the other hand, in step 5T15, it is checked whether the non-intersection part of the road where tracking has been started is tracked in both directions, and if tracking has been completed in only one direction, tracking is started in the other direction (step 5T16). If tracking in both directions has been completed, the tracked road non-intersection portions and intersection portions are stored as one continuous road, and the process proceeds to step 5T17. Furthermore, if all the roads have not been tracked, the process returns to step 5TIO to start the next tracking, and if all the roads have been tracked, the process ends (step ST1?). In the manner described above, it is possible to recognize continuous roads, which was impossible with conventional drawing recognition devices. In the above embodiment, the road non-intersection extraction means 2 tracks the pair of vectors v, W and vector ■9X, but the vectors v, W and vectors ■, 9a, as shown in FIG. 4, even if a glue vector is immediately generated between the end points of both vectors after detection. 9c and 9d are extracted separately as road non-intersection parts, and 9
Since point b is extracted as an intersection, if road tracking is performed using intersection 9b as a road intersection in an expanded sense,
Similar to the above embodiment, not only can roads be recognized, but
This has the advantage of being able to stably recognize road bends and clearly distinguishing them. At the same time, plazas and roundabouts located along roads have the advantage of being able to be distinguished and extracted as intersections in an expanded sense. Furthermore, in the above embodiment, the road non-intersection extracting means 2 generates a road core line, and the road intersection extracting means 3 performs a process of joining the edges. The same effect as in the above embodiment can be obtained even if only the above-mentioned components are used. Furthermore, in the above embodiment, when tracking roads, only the connecting road non-intersection parts and road intersection parts are tracked, but as shown in FIGS. 5(A) and 5(B), the glue vectors are parallel to each other. Even when the roads are lined up, if the gaps are used for tracking, not only can the same effect as in the above embodiment be obtained, but also it is possible to mark a part of the road as shown in Fig. 5(A). This method has the advantage of being able to cope with cases where there is a traffic jam or where the road is divided by an overpass or the like as shown in FIG. 5(B). Further, the road non-intersection extraction method and the road intersection extraction method are merely examples, and other methods may be used. It is also conceivable to extract intersections and non-intersections in this order, contrary to this embodiment. In this embodiment, the recognition of roads has been described, but the present invention can be similarly applied to the recognition of waterways, etc. [Effect of the invention 1] As described above, according to the present invention, road non-intersection parts and road intersection parts are extracted, and the connection between them is tracked by the road tracking/recognition means while determining continuity as a road. With this configuration, even if there are patterns such as rivers that are similar to roads in the map, a drawing recognition device can be obtained that can correctly and selectively recognize continuous roads, excluding these patterns, with high accuracy. effective.

[Brief explanation of the drawing]

FIG. 1 is a block connection diagram showing a drawing recognition device according to an embodiment of the invention, FIG. 2 is a flowchart showing a drawing recognition procedure according to the invention, and FIG. 3 is an explanatory diagram showing a road tracking method according to the invention. , Fig. 4 is an explanatory diagram showing a method of tracking a road intersection, Fig. 5 is an explanatory diagram showing a method of tracking when glue vectors are arranged in parallel, and Fig. 6 is an explanatory diagram showing a method of tracking when glue vectors are arranged in parallel. FIG. 7 is a flowchart showing a conventional drawing recognition procedure; FIG. 8 is an explanatory diagram showing a method for tracking pairs of vectors; FIG. 9 is an explanatory diagram showing how glue vectors are generated; FIG. 10 is an explanatory diagram showing the graph structure of a road intersection. 1 is a map data input means, 2 is a road non-intersection part extraction means, 3 is a road intersection part extraction means, and 4 is a road tracking/recognition means. In addition, in the figures, the same reference numerals indicate the same or equivalent parts. Figure 1 Figure 6 Figure n-1) (A) FC+ Figure (2) (A) +8) Figure (Y-1) Figure (A-T2)

Claims (1)

[Claims] A map data input means that inputs map data and vectorizes and stores the line segments that make up the map; and a road non-intersection section that analyzes the map data and extracts road sections between road intersections. an extraction means; a road intersection extraction means that analyzes the map data and extracts road intersections; and a road intersection extraction means that analyzes the map data to extract intersections of roads; A drawing recognition device comprising road tracking/recognition means for recognizing roads by tracking them.