JPH10134144A - Recognition method for drawing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain the method for recognition which surely relates a figure to a character string at a high speed. SOLUTION: A figure having a frame larger than a character is extracted from a drawing image, a polyline having end points at both its ends among the polylines constituting the figure is separated as a leader line (51), whose end-point positions are estimated (S2), thereby searching for a character string at its periphery (S3). Further, the searched character string is cut out of the drawing (S4), and the figure frame that the leader line indicates is made to correspond to the retrieved character string (S5). Then the character string is separated as a character string which is independent of the figure and other character strings (S6), and a character string having no leader line is regarded as a character string relating to the figure frame where the character string is positioned.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for recognizing a drawing read by a drawing input device or a drawing created by CAD, and more particularly to a method for recognizing a drawing composed of graphic frames, characters, and leader lines.

[0002]

2. Description of the Related Art As shown in FIG. 3, a drawing such as a lot number map or a public map is composed of a site boundary line having a complicated mesh shape, and numerals, character strings, and leader lines such as addresses. .

Such a drawing is read by an image scanner in a drawing input device, and line segments (polylines) and characters constituting a figure are recognized and edited as drawing data so that the drawing data can be used as CAD data and the like. . In CAD, a lot number map and the like are created using the drawing function, and the drawing data is stored.

To save these drawings as data,
It is necessary to store the related information between the figure and the character string, so that it is desirable that the operator can automatically recognize the association between the figure and the character string in the drawing recognition instead of instructing this by editing.

In the conventional drawing recognition method, a figure and a character are separated in advance by the size of the frame, and each figure is recognized by a set of closed polylines for the figure recognition, and a character or a character string constitutes the figure. Is recognized by pattern matching of the polyline.

The association between a character string and a graphic is such that a search is made around the character string on drawing coordinates to find a leader line.

[0007]

In the conventional drawing recognition method, in order to automatically recognize the association between a character string and a figure, a search is performed over the entire circumference of the character string, so that the amount of calculation for the search becomes large. , Delaying the recognition process.

For example, in FIG. 3, a character string "591"
After recognizing, a search is made around the area to search for a leader line. However, since it is not known from the character string "591" which direction to focus on, a search is performed over the entire circumference.
The amount of calculation increases.

In addition, when searching around a character string, it is necessary to determine whether or not there is a leader line even for a character string "589" which should originally have no leader line. Will increase.

Further, when the character string is in contact with the graphic frame, such as the character string "592", the character string is not cut out when separating from the graphic. In this case, processing for separating the figure and the character and editing of the operator are required.

SUMMARY OF THE INVENTION An object of the present invention is to provide a drawing recognizing method capable of surely and rapidly associating a figure with a character string.

Another object of the present invention is to provide a drawing recognition method which can surely and easily separate a figure, a character string, and a leader line.

[0013]

The present invention utilizes the fact that both ends of a leader line are end points, first separates the leader line from the drawing by searching for the end point, and searches only around the end point of the leader line. In the following method, when recognizing a drawing composed of a graphic frame, a character string, and a leader line, the drawing image is used. A graphic with a frame larger than a character is extracted, and those having endpoints on both sides of the polyline constituting the graphic are separated as leader lines, and a character string is searched for around the two endpoint positions of the separated leader line, The character string searched by the search is cut out from the drawing, the graphic frame indicated by the separated leader line is associated with the searched character string, and the searched character string is converted into a graphic and another character string. Separated as a separate string from the string, characterized in that the string leading lines does not have the character string described above separated the character string to a character string related to the graphic frame is located.

According to the present invention, in recognizing a drawing composed of a graphic frame, a character string, and a leader line, a drawing image is searched for a cluster of images larger than the character size, and a polyline constituting the image is searched for. The end point is searched, and if there is an end point, the polyline is also searched in the reverse direction, and the portion having the end points at both ends is separated from the image as a lead line in the search, and the end point search and the separation of the lead line are performed by the end point. Iteratively separates the leader lines until it is no longer found, and separates the sections as a unit of one image, and repeats the steps until the separation of the sections until there are no clusters of images. As a feature.

[0015]

FIG. 1 is a flowchart of a process of associating a figure with a character string according to an embodiment of the present invention. Each process will be described in detail below.

(S1) Leader Line Separation: In the drawing as shown in FIG. 3, separation / extraction of leader lines is performed prior to separation of a character string and a figure. To separate and extract the leader lines,
Utilizing the fact that a leader line is a polyline having endpoints on both sides, a figure with a frame larger than one character is first extracted, and when this figure has endpoints on both sides, it is separated from the figure as a leader line.

(S2) Direction position estimation: For each of the leader lines cut out in step S1, the angle and the positions of two end points are determined.

(S3) Search: Searching for a character string around the positions of the two end points of the leader line, utilizing the fact that a character string is predicted to be at one of the two end points of the leader line. I do.

(S4) Extraction of character string: A character string extracted from the drawing is extracted from a character string searched by searching for the end point position of the leader line. At this time, even if the character string overlaps the graphic frame, such as the character string “592” in FIG.

(S5) Correspondence between figure and character string: The figure frame indicated by the separated leader line is correlated with the character string. In this process, among the end points on both sides of the leader line, it is determined that the frame including the end point where the character string does not exist corresponds to the character string at the position of the other end point.

For example, in FIG. 3, when the character string "591" is located at one end point of the leader line, the frame (hatched portion) where the other end point of the leader line is located is related to the character string "591". Judge as a frame.

(S6) Character separation: A character string is separated as a character string independent of figures and other character strings in the same manner as in the prior art.

(S7) Correspondence between frame and character in frame: When a plurality of character strings are located in the frame for the separated character strings, one character string is associated with the frame. A graphic frame in which a character string without a line is located is determined as a graphic frame related to the character string.

For example, in FIG. 3, in addition to the character string "559", a character string "560"
Although “561” and “562” are located, the character string “55”
Since only “9” is not associated with the leader line, this character string “559” is determined to be a character string related to the frame.

As described above, according to the present embodiment, the following effects are obtained because the leader line is used as a polyline for associating a figure with a character string.

(A) Since the position of a character string can be predicted in the process of extracting a character string corresponding to a leader line, only the surrounding area needs to be searched, and the search time can be reduced.

(B) For a character string corresponding to a leader line, it is not necessary to recalculate the correspondence between the leader line and the character string, and the calculation time can be reduced.

(C) When a frame and a character string overlap, it is easier to separate a leader line having both ends as end points than to separate a character string. The calculation time can be reduced as compared with the method of performing the calculation.

(D) For a plurality of character strings located inside the frame, the character string associated with the leader line is separated from the frame first, so that the character string associated with the frame can be automatically determined, and the calculation time can be reduced. Can be shortened.

(E) Compared with the conventional method of searching for a leader line whose position and angle are unknown for a character string, only the vicinity of the end point of the drawer is searched, so that the search is accurate.

FIG. 2 shows another embodiment of the present invention, and shows a processing flow for separating a character string, a graphic frame, and a leader line from the drawing.

First, a character string is removed from an image by searching for a cluster of images larger than the size of one character (S1).
1).

Since this remaining image is composed of one or more sections and zero or more lead lines, it is searched for whether or not there is an end point with respect to the polyline constituting the image. Since it is a part, this polyline is searched in the reverse direction (S12).

In this search, a portion having end points at both ends is separated from the image as a leader line (S13).

These processes S12 and S13 are repeated until no end point is found, and the separation is performed by separating the leader lines and setting the remaining portion as one image lump (S14).

These processes S11 to S14 are repeated until there are no more clusters of images, and the remaining clusters of small images are separated as character strings.

Therefore, according to the present embodiment, a character, a figure, and a leader line are separated by utilizing the fact that there are end points on both sides of the leader line. Since it can be realized with various processing techniques, it becomes easy to incorporate correction functions and the like, and customization is also possible.

[0038]

As described above, according to the present invention, by utilizing the fact that both ends of the leader line are end points, the leader line is first separated from the drawing by searching for the end point, and only the periphery of the end point of the leader line is detected. Since the character string is searched for and the end point without the character string is recognized as the graphic frame, the effect of separating and recognizing the graphic from the character string and the lead line reliably and at high speed is obtained.

[Brief description of the drawings]

FIG. 1 is a drawing recognition processing flow showing an embodiment of the present invention.

FIG. 2 is a flow of separating a frame, a character string, and a leader line according to another embodiment of the present invention.

FIG. 3 is an example of a lot number drawing.

Claims (2)

[Claims] In recognizing a drawing composed of a graphic frame, a character string, and a leader line, a graphic of a frame larger than a character is extracted from a drawing image, and polylines constituting the graphic have end points on both sides. Is separated as a leader line, a character string is searched for around the two end positions of the separated leader line, the character string searched by the search is cut out from the drawing, and the graphic frame indicated by the separated leader line is The searched character strings are associated with each other, and the searched character strings are separated as independent character strings from figures and other character strings. A drawing recognition method, wherein a character string related to a graphic frame in which a column is located is used. 2. In recognizing a drawing composed of a graphic frame, a character string, and a leader line, a drawing image is searched for a cluster of images larger than a character size, and an end point is searched for a polyline constituting the image. ,
When there is an end point, the polyline is also searched in the reverse direction, the part having the end points at both ends is separated from the image as a lead line in the search, and the search for the end point and the separation of the lead line are repeated until no end point is found. Separating sections as a lump of one image using the remaining portion after separating the leader lines, repeating the steps until the separation of the sections until there is no cluster of images, and separating the lump of remaining small images as a character string. Characteristic drawing recognition method.