JPS6322350B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (1) Technical Field of the Invention The present invention relates to recognition processing of handwritten line figures, and particularly to a method for recognizing circular symbols and determining the meaning of the symbols on drawings.

(2) Prior art and problems For example, a logic circuit diagram is drawn by hand on a graph paper with grid axes drawn along the grid axes as much as possible, and logic symbols, wiring patterns, etc. are drawn on the handwritten logic circuit diagram. Automatic reading, determination, and input into a data processing device are being considered and studied.

In such conventional handwritten line figure recognition methods, the algorithm was designed to allow circular symbols to be misaligned to the same extent as line figures of other symbols.
When handwritten by a person, circular symbols are more likely to be misaligned than line segments, so such a method is likely to cause misrecognition of circular symbols.

(3) Purpose of the Invention In view of these conventional drawbacks, the present invention aims to provide a method for recognizing circular symbols present in handwritten line figures such as logic circuit diagrams by absorbing their positional deviations. It is something to do.

(4) Structure of the invention According to the present invention, the purpose is to form line figures (line segments, symbols) drawn along predetermined lattice axes.
In this method, a drawing containing a mixture of letters and characters is read by an input device, and a certain size that encompasses the circular symbol is recognized based on the input video and the recognition result of symbols other than the circular symbol recognized by some means. means for scanning a rectangular field of view (processing region) along the grid axes at a constant pitch based on the grid axis spacing; circular symbol feature extraction means for determining whether or not the circular symbol exists within the field of view; and when it is determined by the circular symbol feature extraction means that the circular symbol exists within the field of view of the rectangular area,
circular symbol candidate point determining means for determining the positions of lattice points (points where the lattice axes intersect) included in the field of view of the rectangular area as candidate points where the circular symbol exists; and the circular symbol candidate point determining means. To provide a circular symbol extraction method characterized in that the position of the circular symbol is finally determined by referring to the determined position candidate points of the circular symbol and the recognition results of symbols other than the circular symbol. achieved by.

(5) Embodiments of the invention Hereinafter, embodiments of the system according to the present invention will be described in detail with reference to the drawings.

In drawings such as logic circuit diagrams, line figures made up of straight line components and line figures made up of curved components coexist. Among these, a recognition method for line figures consisting of components that can be approximated in four directions has been almost established. This invention relates to a recognition system for circular symbols in particular among linear figures consisting of curved components.This system has been patented by the same inventors in Japanese Patent Application No. 56-212936 (see Japanese Patent Laid-Open No. 58-114180). .)
is being carried out.

In other words, the logic symbols in Figure 1 a, b,
There is a circular figure like c, and we are trying to extract it and recognize what kind of mark it is.

Circular shapes in logic circuit diagrams include those that mean inversion as shown in FIG. 1 a and b, and those that mean combinations as shown in FIG. 1 c. In the case of FIG. 1a, if the circle figure M does not exist, it is an AND gate, and if it does, it is a NAND gate, so it is important that the figure M is a circle and what kind of surrounding figures it is connected to. In the present invention, the surrounding figure N of the circular figure M, the first
In Figure A, the AND symbol is recognized separately by other means, and it is necessary to recognize whether the figure M is a circle, what kind of line segments there are in its vicinity, and to learn the meaning of the circle figure. It is something to do.

To extract circular shapes, an n x n bit (usually 3 x 3 bit) window, which is used in general character and figure recognition, is used.
Since it is described in detail in the specification of No.-212936, its explanation will be omitted here.

However, in this method, the circular symbols drawn on the drawings are more likely to be misaligned than other line figures. In order to deal with this, we will introduce a circular symbol extraction algorithm according to the present invention that is flexible against positional deviation so that its presence can be recognized no matter where it is written. The extraction method will be described using a circular symbol of 2 mmφ as an example.

(1) Field of view When the resolution is 8 lines per 1 mmφ and a rectangular area of 25 × 25 meshes is used as the field of view, there is a margin of 4 meshes on one side for a circular symbol of 2 mmφ, so this allows for a positional deviation of up to 0.5 mm. will enter the rectangular area of the visual field. FIG. 2 shows the relative relationship between the rectangular area 1 and the circular symbol 2 in this visual field. FIG. 2 shows the case where the circular symbol 2 is located in the center of the area 1. The field of view is cut out using such a rectangular area.

(2) Feature extraction Next, features are extracted to determine whether a circular symbol exists in this field of view. Feature extraction of this circular symbol is detailed in the specification of Japanese Patent Application No. 56-212936, so it will not be described here.

(3) Scanning method and determination of candidate points for existing positions The interrelationship between the scanning method of the image memory and the candidate points for existing positions is shown in FIG. In FIG. 3, reference numeral 3 indicates each grid point, and the grid points exist at a distance interval of 2 mm, and a rectangular area of 25×25 meshes, that is, a field of view, is scanned along these grid points. 1
Since the tolerance for displacement of the circular symbol within one field of view is 0.5 mm, the field of view is scanned at a pitch of 1 mm. By doing this, it becomes possible to extract the circular symbol no matter where it is placed. Further, when it is determined that a circular symbol exists within the above-mentioned rectangular area, that is, within the field of view, a candidate position for the position of the circular symbol is determined according to the scanning position state of the rectangular area, that is, the field of view. That is,
As mentioned above, since the scanning of the rectangular area, that is, the field of view, is performed at a pitch of 1 mm, the relative positional relationship between the rectangular area, that is, the field of view, and the grid points is as shown in 4a and 4 in Fig. 3.
There are four states: b, 4c, and 4d. Therefore, if it is determined that a circular symbol exists within the rectangular area, that is, within the field of view at the scanning position 4a in FIG. In the same way, 4b, 4c, 4 in Fig. 3 are determined as candidate points.
If it is determined that a circular symbol exists in the rectangular area, that is, within the field of view at the scanning position d, two grid points shown in the rectangular area 4b and two grid points shown in the rectangular area 4c are placed in the rectangular area 4c. point,
One grid point shown in the rectangular area 4d is determined as a candidate point for the position of the circular symbol,
Store in candidate point table.

(4) Determining its location Due to its nature, the 2mmφ circular symbol has meaning attached to other symbols. On the other hand, regarding the recognition of other symbols, it was filed by the same inventors as Japanese Patent Application No. 57-43954 and is described in detail in the specification of the same application (see Japanese Patent Application Laid-open No. 58-161088). The explanation will be omitted here.
Therefore, the method of determining the position of the circular symbol, that is, the final position of the circular symbol from the candidate points of the circular symbol, is based on the technique filed as the above-mentioned Japanese Patent Application No. 57-43954. A table of recognition results for symbols other than circular symbols created using is stored), estimates the position where a circular symbol may be drawn when creating a drawing, and determines whether that position has been determined as a candidate point where the circular symbol exists. If the position is a candidate point for a circular symbol, it is finally determined that a circular symbol exists at that position. For example, 5a, 5b, and 5c in FIG. 4 indicate candidate points for the positions of circular symbols around the NAND gate. At this time, the shape of the basic part of the NAND gate (the shape of the AND gate) is stored in the symbol recognition result table for symbols other than circular symbols, and is also stored in the circular symbol candidate point table, so the location of the circular symbol can be determined. is executed as follows. First, extract the position where the shape of the basic part of the NAND gate (and gate shape) exists from the symbol recognition result table for symbols other than circular symbols, and find the position where the circular symbol may be written (in this example, the position 5c). presume. Next, it is checked whether a location where a circular symbol could be written is a candidate point for the circular symbol. In this example, the position 5c is the candidate point for the position where the circular symbol exists, so the candidate point for the circular symbol at the position 5c is finally determined as the position where the circular symbol exists. However, since there are no symbols other than circular symbols in the vicinity of the other circular symbol candidate points 5a and 5b, it is determined in the final determination that no circular symbol exists there. In this way, by referring to the symbol recognition result table for symbols other than circular symbols, each of the candidate points for the position of the circular symbol can be finally determined as being the position of the circular symbol or not. is determined, and the location of the circular symbol is determined.

FIG. 5 is a flowchart for explaining the method according to the present invention. First, the contents of the image input device 11 are stored in the image memory 12, and the contents are cut out in a rectangular field of view in block 13. Block 14 is a feature extraction circuit which extracts any circular symbol within the field of view. Its contents are as disclosed in the specification of Japanese Patent Application No. 56-212936. Block 15 determines candidate points for a circular symbol if it is extracted within the field of view. The determined candidate points are written into a candidate point table indicated by block 16. On the other hand, block 17 is a symbol recognition result table whose contents are as follows.
It is disclosed in the specification of No. 43954. Block 18 is a circuit that compares the contents of block 16 and block 17 to determine the position of the circular symbol.

(6) Effects of the Invention As explained in detail above, in the present invention, a field of view of a constant size that includes a circular symbol and takes into account the positional deviation of the circular symbol is set along the lattice axes at lattice axis intervals. Scanning is performed at a certain pitch based on the field of view, and features are extracted to determine whether or not a circular symbol exists within the field of view.If it is determined that a circular symbol exists, multiple grids included in the field of view are extracted. After setting a point as a candidate point for the circular symbol's existence position, the position of the circular symbol is finally determined by narrowing down the candidate points for the circular symbol's existence position while referring to symbol recognition results other than the circular symbol. , can absorb positional deviations that are essential for recognizing handwritten line drawings,
The effects of the present invention are extremely significant in the recognition process of handwritten line figures.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram showing an example of a line figure including a circular symbol, Fig. 2 is a diagram showing the relationship with the circular symbol in a rectangular field of view from which the circular symbol is extracted, and Fig. 3 is a scanning method and existence of the circular symbol. FIG. 4 is a diagram for explaining the method of determining the position candidate point, FIG. 4 is a diagram for explaining the determination of the position of the circular symbol, and FIG. 5 is a flowchart for explaining the operation of the method according to the present invention. Show chart. In the drawing, 1 is a rectangular field of view, 2 is a circular symbol, 3 is a grid point, 4a, 4b, 4c, 4d are candidate points for the circular symbol existence position, 5a, 5b, 5
c is a circular symbol at the existing position candidate point and the determined point, 13 is a field of view cutting circuit block, 15 is a circular symbol candidate point determining circuit block, 16 is a circular symbol candidate point table block, and 18 is an existing position determining circuit block. Each is shown below.

Claims (1)

[Claims] 1 A drawing containing a mixture of linear figures (line segments, symbols) drawn along predetermined grid axes and characters is read with an input device, and the input video and symbols other than circular symbols recognized by some means are recognized. Based on the result, in a method for recognizing circular symbols, means for scanning a field of view (processing area) of a rectangular area of a constant size that includes the circular symbol along the grid axis at a constant pitch based on the grid axis spacing; a circular symbol feature extraction means for performing feature extraction on a video within the field of view of the rectangular area to determine whether or not a circular symbol exists within the field of view of the rectangular area; and feature extraction of the circular symbol. When it is determined by the means that a circular symbol exists within the field of view of the rectangular area, the position of a lattice point (a point where the lattice axes intersect) included within the field of view of the rectangular area is determined as a candidate point where the circular symbol exists. The existence position of the circular symbol is determined by referring to the circular symbol candidate point determination means, the circular symbol existence position candidate point determined by the circular symbol candidate point determination means, and the recognition results of symbols other than the circular symbol. A circular symbol extraction method characterized by comprising means for finally determining.