JPH033082A - Drawing recognizing device - Google Patents

Abstract

PURPOSE:To automatically and quickly decide and extract a main road, a side road, a footpath from a map including a road similar pattern by analyzing inputted map data, extracting a town district including annexed footpaths and executing pair vector extraction by a hierarchical vector extracting means based upon the extracted information. CONSTITUTION:Map data are read in by a map data inputting means 1 and analyzed by a town district extracting means 2 to extract a town district including annexed footpaths. Then pair vectors constituting a road non-intersection part are hierarchically extracted by a hierarchical pair vector extracting means 3 based upon the analyzed town district information. Since the pair vectors are hierarchically extracted, the main road can be extracted from the map including the road similar pattern so as to be distinguished from side road and footpaths.

Description

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

The present invention relates to a drawing recognition device that automatically recognizes roads marked on a map, and more particularly to a drawing recognition device that extracts non-intersection portions of roads.

[Conventional technology]

Figure 5 shows, for example, "Image Database 11" by Itaniku Osawa.
29 to 133, a block diagram of a drawing recognition device in which the conventional digital image vectorization method disclosed in the September 1987 issue of Akirido is applied to road recognition. FIG. In the figure, 1 is a map data input means for inputting and storing map data, and 4 is a pair vector extraction means for extracting from the map data all pairs of line segments lined up substantially in parallel with an interval about the width of a road. Next, based on the above device configuration, the operation of the conventional device will be described below with reference to the road recognition flowchart shown in FIG. In the figure, step STI and steps ST22 to 5T26
are each a processing step. Hereinafter, for the sake of simplicity, this will be simply referred to as a step. The conventional device extracts a pair of parallel line segments from map data as a pair vector, and uses this as a road non-intersection portion between road intersections. First, in step STI, map data is input, line segments forming the map are vectorized, and stored in a storage device (not shown). At that time, the vector should be in a format that represents the road contour rather than the road center. 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, in step 5T22, a set of vectors having a length equal to or greater than the reference length is found among the human vectors, and this is set as S. Next, in step 5T23, the set S
Find the longest vector among them and remove it from the set S as vector V. Next, Stella 7”5T24,5
At T25, a pair of vectors W of the vector ■ is determined, and this pair of vectors VW is determined as a road non-intersection portion. First, in step ST24, from the set S, the vector ■
Find vectors on both the left and right sides of the vector ■ that are lined up almost parallel to each other, the distance between them is the closest, and there are no other vectors in between, and calculate this as the vector W.
shall be. Next, in step 5T25, the vectors W and V
If the interval is within the standard value of the road width, the vector ■・W
Let be a pair vector. Also, if it is outside the standard range, it cannot be considered a road, so it is not treated as bare. Next, in step 5T26, the set S is checked, and if there is an unprocessed one, the process returns to step 5T23 to continue extraction of bare vectors, and if there is no unprocessed one, the process ends. The above step STI is processed by the map data input means 1, and steps ST22 to ST26 are processed by the pair vector extraction means 4. The roads on the map are recognized in the manner described above.

[Problem to be solved by the invention]

Conventional drawing recognition devices are configured as described above, so when searching for paired vectors, almost all vectors in the map are considered as paired vector candidates to determine whether or not they can form a pair. Not only does it take a long time to process,
There were issues such as patterns that resembled roads, such as sidewalks, hatched areas (areas shaded with diagonal lines), long and narrow city blocks, and rivers, could be mistakenly identified as roads and extracted. This invention was made to solve the above-mentioned problems, and it analyzes the map data input by the map data input means, extracts the blocks including the attached sidewalks, and creates a hierarchy based on that information. By manually extracting pair vectors, main roads are extracted from maps containing similar road patterns by
The purpose of this invention is to obtain a drawing recognition device that can distinguish and extract sidewalks.

[Means for Solving the Problems 1] A drawing recognition device according to the present invention analyzes map data inputted by a map data manual means using a block extracting means to extract a block (including attached sidewalks); Pair vectors forming road non-intersection areas are hierarchically extracted by a hierarchical bare vector extracting means based on the information on the city block. [Operation] When map data is read by the map data input means, the drawing recognition device according to the present invention analyzes the map data by the block extraction means and extracts the block including the attached sidewalk. Next, based on the analyzed block information, a hierarchical bare vector extraction means hierarchically extracts pair vectors forming road non-intersection areas.

【Example】

An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, the same parts as in FIG. 5 are indicated by the same reference numerals. In FIG. This is a hierarchical bare vector extraction means that hierarchically extracts pair vectors forming road non-intersection parts between road intersections. Next, the operation will be explained with reference to the flowchart shown in FIG. First, step STI operates in the same manner as in the conventional example. That is, in step STl, map data is input, vectorized, and stored in a memory (not shown). In this case, the vector should be in a format that represents the road contour rather than the center of the road. This process is executed by the map data manual means l. Next, in step ST2, portions that are considered to be discontinuities in lines representing city blocks (including attached sidewalks) are corrected. A detailed explanatory diagram of step ST2 is shown in FIG. Figure 3 (A) is an example of a break T1 due to an error during input, Figure 3 (B) is an example of a break T2 due to an omission during entry, and Figure 3 (C) is a break T3 due to the representation of a dead end road. This is an example. Next, in step ST3, vectors that are connected to each other are combined into one group, and this is defined as one city block (including the attached sidewalk). Next, in step ST4, the adjacency relationship between the city sections is determined. This is done, for example, by checking the positional relationship of the circumscribed rectangles of each block. The above processing is executed by the block extracting means 2. Next, in steps ST5 to ST7, as the first stage of hierarchical pair vector extraction, pair vectors forming major roads are extracted. An explanatory diagram of the concept of major roads is shown in Figure 4. The property that the paired vectors constituting the main road belong to blocks adjacent to each other is utilized. First, in step ST5, the longest vector among the unprocessed vectors having a reference length L or more is determined, and this vector is set as ■. Next, in step ST6, among the vectors constituting the block next to the block to which the vector ■ belongs, find one that has a length of L or more, is lined up almost parallel to the vector ■, and has the closest interval, and converts this into a vector. Let W be the vector, and let the vector ■ and the vector W be a pair vector of 1 & 11. Next, in step ST7, if there are unprocessed vectors of length or longer remaining, the process returns to step ST5 and continues extraction of paired vectors constituting the main road, and if no vectors remain, the process proceeds to step ST8. In steps ST8 to ST13, as the second stage of hierarchical pair vector extraction, pair vectors forming the sidewalk R, side street P, etc. are extracted. The property that these paired vectors belong to the same city block is utilized. First, in step ST8, attention is paid to one city block. Next, in step ST9, the longest unprocessed vector in the block that is longer than the reference length L is found, and this is designated as vector {circle around (2)}. Next, in step 5TIO, among the vectors constituting the block to which the vector ■ belongs, find the vector that has a length of L or more, is lined up almost parallel to the vector ■, and has the closest interval.
Let this be a vector W, and let the vector 2 and the vector W be a pair of vectors. Next, in step 5TII, it is checked whether there are any unprocessed vectors with a length longer than 1 remaining in the currently focused block, and if so, the process returns to step ST9, and Vector extraction continues, and if no vectors remain, proceed to step 5T12. In step 5T12, it is checked whether all blocks have been processed.
If the process has not been completed, the process proceeds to step 5T13, and after paying attention to the next block, returns to step ST9 to continue extracting pair vectors, and if the process has been completed, the entire process ends. Note that the processing in steps ST5 to ST13 is executed by the hierarchical pair vector extraction means 3. As described above, road recognition can be performed by distinguishing between main roads, side streets P, sidewalks R, etc., which could not be distinguished using conventional devices. In addition, in the above embodiment, the configuration of the drawing recognition device that recognizes only non-intersection parts of roads is shown, but it is also possible to add an intersection extraction means to this to perform more advanced road recognition. . Further, in the above embodiment, the hierarchical pair vector extraction means 3
A processing method for extracting pair vectors one set at a time was described in , but if a processing method is used that examines the vectors connected to both ends each time a pair is extracted and creates a pair vector, the same effect as the above embodiment can be obtained. In addition, there is an advantage that pair vector extraction can be performed faster. In addition, in the above embodiment, a method for correcting line breaks in the block extracting means 2 has been shown, but if the blocks connected by crossover corridors, pedestrian bridges, etc. are separated, median strips are removed, etc. can be recognized correctly. Further, in the above embodiment, the hierarchical pair vector extraction means 3
The adjacency relationship between city blocks is determined by , but if we incorporate the concept of adjacent directions and use it in pair vector extraction, we can not only obtain the same effect as the above example, but also
This has the advantage that pair vector extraction can be performed faster. Further, in the above embodiment, the hierarchical pair vector extraction means 3
In the above, a two-step process of extracting main roads and extracting side streets P and sidewalks R is performed, but only one of them may be extracted depending on the purpose. Furthermore, although this embodiment has been described with respect to road recognition, the same can be applied to waterway recognition and processing that generally requires pair vector extraction, such as center line conversion rings of direct image contour vectors. Alternatively, a configuration may be adopted in which pair vector extraction is performed in order from the widest road width to the narrowest road width, and the same effect as in the above embodiment can be obtained. Furthermore, there is an advantage that more important thick roads can be reliably extracted with priority. Furthermore, when the block extraction means 2 extracts a block, vectors forming the outermost outline of the block are arranged in a fixed direction to form a closed loop. If this process is performed, the pair vector search target can be specified as the vectors forming this closed loop when major roads are extracted, so that the process can be speeded up. Further, by utilizing the property that pair vectors are vectors in opposite directions, more reliable pair vector extraction can be performed.

【Effect of the invention】

As described above, according to the present invention, input map data is analyzed to extract blocks including attached sidewalks, and pair vectors are extracted by hierarchical vector extraction means based on the extracted information. Since the present invention is configured to do this, it is possible to automatically determine and extract main roads, side streets, sidewalks, etc. from a map including road similar patterns in a short time.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a drawing recognition device showing an embodiment of the present invention, FIG. 2 is a flowchart showing the operation of FIG. 1,
Figure 3 is an explanatory diagram of block extraction in Figure 2, Figure 4 is an explanatory diagram of the outline of main roads and side streets in Figure 2, Figure 5 is a block diagram of a conventional drawing recognition device, and Figure 6 is an illustration of the outline of the main roads and side streets in Figure 2. 6 is a flowchart showing the operation of FIG. 5. 1... Map data input means, 2... Block extraction means,
3... Hierarchical pair vector extraction means. In addition, in the figures, the same reference numerals indicate the same or equivalent parts. Figure 6 Patent Applicant: Mitsubishi Electric Corporation (2 others) 5(・2/ Figure YO (Directly generated diagram (1)) Figure 6 (Z of 'f)

Claims (1)

[Claims] A map data input means for inputting and storing map data such as roads described on a map, a block extraction means for analyzing the map data and extracting data corresponding to blocks, and block information obtained by the block extraction means. A drawing recognition device comprising hierarchical pair vector extraction means for hierarchically extracting pair vectors forming road non-intersection parts between road intersections based on the following.