JPS63292278A - Pattern recognizing method - Google Patents

Abstract

PURPOSE:To effectively decide the relation between a symbol image and a segment image connected to said symbol image, by recognizing those two images in order to delete an overflowed part due to handwriting in terms of line image data excluding short segment data serving as arrow candidates. CONSTITUTION:Both the branch and end points of a line image are first recognized for the line image data. Then the short line image data having the distance between the branch and end points smaller than a prescribed value is specified for the line image data. Then a symbol image and a segment image connected to said symbol image are recognized for the line image data excluding the short line image data. Thus an arrow image is recognized based on the short line image data on a prescribed area set near the end point of the segment image. In this case, it is desirable to count the number of image elements of the short line image data contained in each prescribed area set so as to sandwich the segment image to recognize the arrow images.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a graphic recognition method, and more particularly to a graphic recognition method for effectively recognizing arrows included in line drawings such as circuit diagrams, block diagrams, and maps.

[Prior Art] Conventionally, arrows in line drawings are recognized by first detecting two short line segments that are close to each other, and then determining whether the angle between the two short line segments and the long line segment sandwiched between them has a predetermined value. I went there depending on whether there was someone there or not.

However, in general, when line thinning processing is performed on a line image, shapes such as whiskers and loops are likely to be formed particularly in areas where multiple lines are gathered. For example, at least three lines gather in the arrow part, but in this case, a loop shape is formed at the angle formed by the short line segment and the long line segment of the arrowhead part, causing the angle to change or making it difficult to judge the angle. It happens. Furthermore, the short line segment may become short, making it difficult to determine the angle.

Additionally, in hand-drawn drawings, lines are often broken or lines at branch points protrude. Therefore, in the above-mentioned conventional technology, it is difficult to distinguish between an original arrowhead portion and a whisker existing at a branch point, and arrows are often misidentified.

[Problems to be Solved by the Invention] The present invention is intended to eliminate the above-mentioned drawbacks of the prior art, and its purpose is to provide a figure recognition method that effectively recognizes arrows in line drawings. be.

[Means for Solving the Problems] In order to achieve the above object, the figure recognition method according to the present invention includes a step of recognizing branch points and end points of a line drawing with respect to line image data, and a step of recognizing branch points and end points of a line drawing with respect to line image data; a step of identifying short line image data in which a distance between a branch point and an end point is equal to or less than a predetermined value; and a step of recognizing a symbol image and a line segment image connected to the symbol image for line image data other than the short line image data; The outline of the method includes the step of recognizing an arrow image based on short line image data of a predetermined area set near the end point of the line segment image.

Preferably, the step of recognizing the arrow image includes recognizing the arrow image by counting the number of pixels of the short line image data included in each predetermined area set to sandwich the line segment image.

[Operation] In this configuration, first, branch points and end points of a line drawing are recognized for line image data. Next, regarding the line image data, short line image data in which the distance between the branch point and the end point is equal to or less than a predetermined value is identified. Next, symbol images and line segment images connected to the symbol images are recognized for line image data excluding short line image data. Then, an arrow image is recognized based on short line image data of a predetermined area set near the end point of the line segment image.

Preferably, the arrow image is recognized by counting the number of pixels of the short line image data included in each predetermined area set to sandwich the line segment image.

[Embodiments] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram of a figure recognition apparatus using a figure recognition method according to an embodiment of the present invention.

In the figure, 1 is an image human-powered device equipped with an image scanner, etc., 2 is a preprocessing unit that performs image processing such as binarization processing and thinning processing on input image signals, and 3 is preprocessing in the preprocessing unit 2. The image storage unit 4 that stores the image data stored in the image storage unit 3 recognizes branch points and end points of line drawings in the image data stored in the image storage unit 3, and stores short line segment data in which the distance between the branch point and the end point is less than or equal to a predetermined value. 5 is an image storage unit that stores image data obtained by removing short line segment data from the image data in the image storage unit 3; 6 is a short line segment buffer that stores the separated short line segment data; 7 is a short line segment buffer that stores the separated short line segment data; A symbol connection line extraction unit 8 extracts symbol figure data and connection line segment figure data from the image data in the image storage unit 5; symbol connection line extraction unit 7;
9 is a symbol connection line data storage unit that stores the extracted symbol figure data and connection line segment figure data; 9 is a line segment integration unit that integrates the image data in the image storage unit 5 and the short line segment data in the short line segment buffer 6; An image storage unit 11 stores image data integrated by the line segment integration unit 9;
This is an arrow extracting section that recognizes and extracts arrows based on the image data of the symbol connection line data storage section 8 and the image data of the symbol connection line data storage section 8.

If you examine many line images such as circuit diagrams, block diagrams, process diagrams, etc., you will find that arrows are attached only to the end points of connecting line segments. Here, a connecting line segment refers to a line segment that touches at least one end of a symbol, block, or other connecting line segment,
The line segment may include corner portions, curved portions, and the like. Further, the arrow (arrowhead portion) exists at an end point of such a connecting line segment that makes contact with another, or exists at an end point that does not contact with anything else.

Therefore, on the Kinomi style side, the short line segment data that is thought to form the arrowhead is temporarily removed from the image data, and the symbol figures included in the remaining image data and the connecting line figures that connect the symbol figures are By accurately grasping and clarifying these contact points, anticipating the possibility of the existence of an arrowhead at this contact point, and referring to the removed short line segment data, it is possible to identify the short line segment in a reliable manner. Recognize whether the data part is an arrow or not.

In the above configuration, the image data in the image storage unit 3 includes not only original arrows but also protruding portions of handwritten lines, broken portions of lines, and the like. The short line segment separation unit 4 detects end points and branch points included in this image data, uses this information to find the distance from the end point to the branch point, and recognizes those whose distance is less than a predetermined value as short line segments. do. The endpoint in this case is an endpoint that is not in contact with anything else. Then, the portion of the short line segment data that is recognized as a short line segment is stored in the short line segment buffer 6 as a coordinate string or a coded version of each pixel. The image storage unit 5 stores the remaining image data. Therefore, in the image data stored in the image storage unit 5, the data of the arrowhead portion and the data of the portion protruding from the branch point are removed. Next, the symbol connection line extraction section 7 extracts symbol graphic data and connection line segment graphic data from the image data in the image storage section 5. This is because the presence or absence of an arrow largely depends on the relationship between the symbol figure and the connecting line segment. Furthermore, when extracting this symbol graphic data and connecting line segment graphic data, processing for connecting lines and breaks can be performed if necessary. In this way, the extracted symbol graphic data and connecting line segment graphic data are stored in the symbol connecting line data storage section 8. On the other hand, the line segment integration unit 9 reproduces the original image data by integrating the image data in the image storage unit 5 and the short line segment data in the short line segment buffer 6, and stores this in the image storage unit 10. At that time, for example, the pixel code of the short line segment data is changed and returned.

This is to identify the short line segment data from other image data.

Then, the arrow extraction unit 11 counts the image data (i.e., short line segment data) in the image storage unit 10 that exists within the area determined based on the symbol graphic data and the connecting line segment graphic data in the symbol connecting line data storage unit 8. The arrow is recognized and extracted by comparing the counted value with a predetermined threshold value.

For convenience of explanation, the image storage units 3.5 and 10 are depicted as being different, but in reality they may be the same. That is,
In this embodiment, only when the symbol connection line extraction section 7 extracts symbol figure data and connection line segment figure data,
It is sufficient if the short line segment data is removed from the original image data.

FIG. 2 is a flowchart showing the processing procedure for arrow extraction in the embodiment, and FIG. 3 is a diagram showing the concept of arrow extraction in the embodiment.

In FIG. 3, for example, if we consider a diagram consisting of a rectangular symbol 1oo and a connecting line segment 101 connected to it, the arrowhead portions 102 and 103 are attached only to the end point a of the connecting line segment 101. Therefore, the recognition of the arrow is the connecting line segment 1
It is only necessary to examine the vicinity of the end point of 01. Also, g in Figure 3
The point is a branch point, and the point g is the other end point of the connecting line segment 101. Therefore, the presence or absence of an arrow is also determined for this part.

In FIG. 2, in step S21, it is checked based on the image data in the symbol connection line data storage section 8 whether there is any unprocessed connection line segment. If there is no unprocessed connecting line segment, the process ends, and if so, the process advances to step S22 and one unprocessed connecting line segment is selected. In step S23, it is determined whether or not there is an unprocessed end point for the selected connecting line segment.

If there are no unprocessed pane points, the process advances to step S21 to check whether there are any other unprocessed connecting line segments.

If there is an unprocessed endpoint, the process proceeds to step S24, and an area for checking the presence or absence of an arrowhead at the corresponding endpoint is determined in the image data of the image storage unit 10.

FIG. 3 shows an example of a method for determining this area. now,
Note the end point a of the connecting line segment 101.

Then, point b, which is a predetermined distance away from point a, on the connecting line segment 101 is found. This predetermined distance is a distance that sufficiently encompasses the arrowhead portion.

Next, the connecting line segment 10 is placed between the connecting line segments 101.
Rectangular area ab with line segment ab as one side on both sides (top and bottom) of 1
Set cd and abef. The rectangular areas abed and abef are large enough to fully include the arrowhead portion.

In step S25, the area abe obtained in step S24 is
Check whether there is a short line segment within d and abef. For example, the presence or absence of a short line segment is checked as follows. As mentioned above, when the short line segment data is returned to the image storage unit 10, the pixel values are changed.
Only the short line segment data of the arrowhead portions 102 and 103 have different pixel values. Therefore, the areas abed and ab
ef is scanned and the number of pixels of each short line segment data is counted. If this count is greater than or equal to a predetermined value, it is determined that an arrowhead portion exists within this area abcd or abef. Arrows can be expressed in various ways depending on the drawing.

For example, one area abcd or abe of the connecting line segment 101
For example, there is an arrowhead part only in the area f, or there is an arrowhead part in both areas. In step S2B, the conditions of the drawing are also considered, and if it is determined that there is an arrowhead portion in one or both of the areas, it is recognized that there is an arrow at the end point of this connecting line segment 101.

In the above embodiment, connecting line segments, symbol figures, etc. are recognized after removing short line segment data that are arrow candidates, and then adding short line segment data that is arrow candidates again to check whether they are arrows or not. Ta. However, in addition to this, for example, once short line segment data that is an arrow candidate has been determined, that short line segment data can be labeled differently from other connecting line segments or symbol shapes, which is equivalent to the removal described above. It may also have the effect of

[Effects of the Invention] As described above, according to the present invention, symbol images and line segment images connected to the symbol images are recognized for line image data excluding short line segment data that are arrow candidates.
Protruding portions caused by handwriting can be removed, and the relationship between a symbol image and a line segment image connected to the symbol image can be effectively determined.

Further, according to the present invention, since an arrow image is recognized based on the short line image data of a predetermined area set near the end point of the connecting line segment image, it is possible to ensure the recognition result and simplify the recognition process.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a figure recognition device using a figure recognition method according to an embodiment of the present invention, FIG. 2 is a flowchart showing a processing procedure for arrow extraction in an embodiment, and FIG. It is a diagram showing the concept. In the figure, 1...Manual device, 2...Preprocessing unit, 3,5°10...Image storage unit, 4...Short line segment separation unit, 6...
・Short line segment buffer, 7...Symbol connection line extraction unit, 8
Symbol connection line data storage unit, 9...line segment integration unit, 1
1...Arrow extraction section.

Claims (2)

[Claims] (1) A step of recognizing branch points and end points of a line drawing in the line image data; a step of identifying short line image data in which the distance between the branch point and the end point is equal to or less than a predetermined value in the line image data; A step of recognizing a symbol image and a line segment image connected to the symbol image for line image data excluding data, and a step of recognizing an arrow image based on short line image data of a predetermined area set near the end point of the line segment image. A shape recognition method comprising: (2) The step of recognizing the arrow image is characterized in that the arrow image is recognized by counting the number of pixels of the short line image data included in each predetermined area set to sandwich the line segment image. The figure recognition method described in item 1.