JPH07200741A - Graphic recognition device - Google Patents

Abstract

PURPOSE:To make the recognition of the feature point of a handwritten graphic immune to noise provided in handwritten data. CONSTITUTION:Concerning the sequence of coordinate points applied from an input device 1 to a point sequence reading means 21, when the distance between two continuous coordinate points is short, one point is removed by a distance thinning means 22. Next, a first angle change calculating means 23 calculates the elevation angles of three continuous points in the remaining point sequence. An angle change thinning means 24 remains just the points for which the sum of elevation angles is larger than a fixed value. The remaining point sequence is inputted to a second angle change calculating means 25. The second angle change calculating means 25 calculates the elevation angles of three continuous points. A deviation value calculating means 26 calculates a deviation value in the angle changes of all the elevation angles. The maximal value, which deviation value is larger than a fixed value, is defined as the coordinate point of an apex candidate.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern recognition device,
In particular, the present invention relates to a figure recognition device that extracts the characteristics of a figure by using thinning processing based on the magnitude of the angle change for feature extraction again.

[0002]

2. Description of the Related Art Conventionally, this type of figure recognition apparatus extracts characteristic points such as vertices of a figure under conditions not affected by noise such as camera shake, and is suitable for the obtained characteristic points. The purpose is to recognize figures. For example, as shown in Japanese Patent Laid-Open No. 62-72087, first, vertices are extracted from the input handwritten figure. Next, the input figure and the dictionary figure are compared to recognize the most suitable figure as a candidate figure.

[0003]

In this conventional pattern recognition apparatus, it is difficult to accurately recognize the vertices due to noise such as camera shake, and therefore, there is a problem of erroneous recognition of a pattern.

[0004]

In order to solve the above-mentioned problems, the graphic recognition apparatus of the present invention has an input device, an output device, a means for reading a coordinate point sequence from the input device, and a fixed coordinate point sequence. Distance thinning means for thinning out, a first angle change calculating means for obtaining an angle change of the remaining coordinate point sequence, a minute angle change thinning means for thinning out the points when the angle change is small, and the remaining coordinates. The second to obtain the angle change of the point sequence
Recognition means for recognizing a figure by using the angle change calculation means, the deviation value calculation means for calculating the deviation value of the angle change, the storage device for storing the coordinate point sequence, and the feature amount calculated by the deviation value calculation means. It is equipped with a device.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a first embodiment of the present invention. Referring to FIG. 1, a first embodiment of the present invention is directed to an input device 1 such as a tablet that is input by a pen, a processing device 2 that operates by program control, and a storage device that stores a point sequence read from the input device 1. 3, an output device 4, and a recognition device 5.

When the processing device 2 connects the point sequence reading means 21 for reading the coordinate point sequence input from the input device 1, the distance thinning means 22 for thinning the point sequence at a constant distance, and the point sequence connected in the order of input. A first angle change calculating means 23 for obtaining a possible angle change, an angle change thinning means 24 for thinning out the point when the angle change obtained by the first angle change calculating means 23 is minute, and the angle change thinning A second angle change calculation means 25 for obtaining an angle change made by connecting the point sequences input from the means 24, and the second angle change calculation means 25.
It comprises a deviation value calculating means 26 for calculating the deviation value of the angle change obtained in (1).

Next, the operation of this embodiment will be described with reference to FIGS. FIG. 2 is a flowchart showing the processing procedure of the embodiment of the present invention. FIG. 3A is an example of a handwritten input figure. FIG. 3B is an example showing an actual sequence of coordinate points of the input graphic. Figure 3
As shown in (b), the actual coordinate point sequence is a scattered point sequence. FIG. 4 is a diagram showing elevation angles formed by three consecutive points. FIG. 5 is an example showing the processing result of the input coordinate points by the angle thinning processing of the present invention. Figure 6
It is an example of a graph showing a deviation value of an angle change when the angle thinning process of the present invention is not performed. FIG. 7 is an example of a graph showing the deviation value of the angle change when the angle thinning process of the present invention is performed. In each of FIGS. 6 and 7, the vertical axis represents the deviation value, and a high deviation value means that the angle change is large.

The coordinate point sequence given to the point sequence reading means 21 from the input device 1 is inputted to the distance thinning means 22.
When the distance between two consecutive coordinate points is shorter than a certain value (for example, a value determined by the size of the circumscribing rectangle of the coordinate point sequence), the distance thinning-out means 22 thins out the last one of the two points. This process is performed in the input order for all coordinate points. At this time, when the thinning-out process is actually performed, the remaining one point and the one next to the thinned-out point are the next processing target points. On the other hand, if it is not actually thinned out, the subsequent one point and the next point become the next processing target points (step A1).

After the processing of step A1, the remaining point sequence is input to the first angle change calculation means 23. The first angle change calculation means 23 obtains the elevation angle from the input sequence of three coordinate points (step A2). Here, the elevation angle, as shown in FIG. 4, connects three consecutive points with two straight lines and extends a straight line connecting the first and second points and a straight line connecting the second and third points. It is the angle formed by. Next, the angle change thinning-out means 24 adds the obtained elevation angle to the sum of the elevation angles obtained up to that point (step A3). When the calculated sum is equal to or larger than a certain value, the angle change thinning-out means 24 leaves the second point among the three points in the storage device 3. At this time, the sum of elevation angles is initialized (step A4,
Step A5). When there are remaining points for which the elevation angle has not been obtained, the elevation angle is obtained again by the first angle change calculation means 23 using the second point, the third point and the next one point (step A6).

By processing all the coordinate point sequences in steps A2 to A6, the remaining point sequence is in a state in which noise is removed, as shown in FIG. Then, it is input to the second angle change calculation means 25. The second angle change calculation means 25 calculates the angle change using the coordinates of the three consecutive input points similarly to the first angle change calculation means 23 (step A7).

When all the coordinate point sequences have been processed in step A7, the deviation value calculation means 26 receives the value of the angle change obtained in step A7 as an input and obtains the deviation value for each angle change. (Step A8).

Among the calculated deviation values, the maximum value having a certain value or more is set as the coordinate point of the vertex candidate (step A).
9). In particular, as shown in FIGS. 6 and 7, the difference in the deviation value between the feature point and the other points can be clarified by the noise removal by the angle thinning processing.

The recognition device 5 recognizes the candidate figure using the feature amount obtained by the above processing.

Next, a second embodiment of the present invention will be described.

Referring to FIG. 8, in the second embodiment of the present invention, in addition to the configuration of the processing device 2 in the first embodiment shown in FIG.
The difference is that the angle change change thinning means 28 for thinning the input coordinate points is provided when the change in the angle change is minute.

The operation of this embodiment will be described with reference to FIGS.

Since the operation of steps A1 to A6 of FIG. 9 is the same as that of the first embodiment, the description thereof will be omitted. Step A
After the processing of 6, the remaining point sequence is the change calculation means 2 of the angle change.
Input to 7. The angle change change calculation means 27 calculates the ratio of two consecutive angle changes. When the value of the ratio obtained by the change calculation means 27 for angle change is less than a certain value, the change thinning means 28 for minute angle change thins out the last point of the second three points of two consecutive angle changes. (Step B1). After the process of step B1, the remaining point sequence is input to step A7, and the same operation as that of the first embodiment is performed thereafter (steps A7 to A9). By the process of step B1, a smooth curved portion can be deleted.

Therefore, the difference in deviation value between the feature point and the other points can be made clearer.

[0019]

As described above, the graphic recognition apparatus according to the present invention is resistant to noise such as camera shake. Further, by making the minute angle change constant, the variance of the flat portion is reduced and the deviation value of the apex portion shows a large value, which has the effect of improving the recognition accuracy of the apex and singular point of the figure. .

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a flowchart showing a processing procedure of the first embodiment of the present invention.

FIG. 3 is a diagram showing an example of a sequence of coordinate points obtained from the input device 1.

FIG. 4 is a diagram showing an elevation angle formed by connecting three coordinate points.

FIG. 5 is a diagram showing a coordinate point sequence after an angle change thinning process.

FIG. 6 is a graph showing a deviation value when an angle change thinning process is not performed.

FIG. 7 is a graph showing a deviation value after the angle change thinning process.

FIG. 8 is a block diagram showing a second embodiment of the present invention.

FIG. 9 is a flowchart showing a processing procedure of a second embodiment of the present invention.

[Explanation of symbols]

 1 Input Device 2, 6 Processing Device 3 Storage Device 4 Output Device 5 Recognition Device 21 Point Sequence Reading Means 22 Distance Thinning Means 23 First Angle Change Calculating Means 24 Angle Change Thinning Means 25 Second Angle Change Calculating Means 26 Deviation Value Calculating means 27 Change in angle change calculating means 28 Change in thinning angle change means

Claims (3)

[Claims] 1. An input device for inputting a handwritten figure, an output device, a unit for reading a coordinate point sequence from the input device, a storage device for storing the coordinate point sequence, and a distance for thinning the coordinate point sequence at a constant distance. Decimation means, first angle change calculation means for obtaining an angle change of the sequence of coordinate points left by the distance thinning means, and a point when the angle change obtained by the first angle change calculation means is minute, Angle change calculation means for thinning out, a second angle change calculation means for calculating the angle change of the coordinate point sequence remaining by the angle change thinning means, and a deviation value of the angle change calculated by the second angle change calculation means. And a recognition device for recognizing a graphic by using the feature quantity obtained by the means for obtaining the deviation value. 2. An input device for inputting handwritten figures, a means for reading a coordinate point sequence from the input device, an output device, a storage device for storing the coordinate point sequence, and a distance thinning for thinning the coordinate point sequence at a constant distance. Means, first angle change calculation means for obtaining an angle change of the coordinate point sequence remaining by the distance thinning means, and when the angle change obtained by the first angle change calculation means is small, Thinning-out angle change thinning-out means, angle-change change calculating means for obtaining a change in angle change of the coordinate point sequence remaining by the angle-change thinning-out means, and magnitude of change in angle change found by the angle-change change calculating means When the distance is very small, a small angle change change thinning means for thinning the point, and a second angle change means for obtaining the angle change of the coordinate point sequence remaining by the minute angle change change thinning means. And a recognition device for recognizing a graphic by using the feature amount obtained by the deviation value obtaining means. A graphic recognition device characterized by being provided. 3. The graphic recognition apparatus according to claim 1, wherein the distance thinning means determines the distance for thinning out according to the size of the circumscribed rectangle of the input coordinate point sequence.