JPH0225985A - Handwriting judging device - Google Patents

Abstract

PURPOSE:To improve identification accuracy by setting a feature area by including a spot, in which the features of respective characters easily appear, executing weighting processing to a judged handwriting concerning this feature area. CONSTITUTION:When a standard character code to correspond to the judged handwriting is outputted from a character recognizing part 1, handwriting data and feature area data are read from a handwriting dictionary 2 by a reading circuit 3 based on this standard character code. The weighting processing is executed for a judged handwriting pattern from the character recognizing part 1 by an emphasizing circuit 4 based on these feature area data. Resemblance degree is obtained by a resemblance degree arithmetic circuit 5 for the judged handwriting pattern, to which the weighting processing is executed, and individual handwriting data which are read from the handwriting dictionary 2. Then, a person to correspond to obtain a maximum value out of the respective obtained resemblance degrees is identified as a writer which is related to the judged handwriting. Thus, when the writer is identified from the handwriting, the identification accuracy is improved.

Description

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

The present invention relates to a handwriting identification device that identifies a writer based on handwriting.

[Conventional technology]

The conventional method for handwriting identification, which identifies the writer from the handwriting, takes an image of the handwriting to be identified, preprocesses the video signal, separates the characters, extracts features, and then analyzes each handwriting dictionary listed in the individual's handwriting dictionary. The handwriting and munching are taken to find the degree of similarity, and the individual whose handwriting has the highest degree of similarity is determined.

[Problem to be solved by the invention]

In general, individual handwriting characteristics appear prominently at specific locations such as the beginning, end, stroke, and stroke of a character. In the conventional techniques as described above, there is a problem in that the identification accuracy is low because the above-mentioned specific points, which have an important meaning in handwriting identification, are not sufficiently taken into account. SUMMARY OF THE INVENTION An object of the present invention is to provide a handwriting identification device that solves the problems of the conventional techniques and improves the accuracy of identification when identifying a writer from handwriting.

[Means to solve the problem]

In order to solve this problem, the handwriting identification device according to the present invention includes a character recognition unit that determines the pattern of the handwriting to be judged and the code of the standard character corresponding to the pattern of the handwriting to be judged based on the captured image signal of the handwriting to be judged. and; a handwriting dictionary in which the standard character code, individual handwriting data related to the standard character code, and data related to a characteristic area set including a place where the characteristics of each standard character are likely to appear are stored; a reading circuit that reads out the individual handwriting data and the feature area data from the handwriting dictionary based on the standard character code from the character recognition unit; and the handwriting to be appraised from the character recognition unit based on the feature area data. an emphasizing circuit that weights patterns; a similarity calculation circuit that calculates the degree of similarity between the output from this emphasis circuit and the individual handwriting data read from the handwriting dictionary; and a determination circuit for determining the individual associated with the handwriting data corresponding to the handwriting pattern to be appraised based on the output of the handwriting pattern to be appraised, and the output from this determination circuit is used to identify the writer associated with the handwriting to be appraised.

[For use]

When the standard character code corresponding to the handwriting to be appraised is output from the character recognition section, the handwriting data and characteristic region data are read out from the handwriting dictionary by the reading circuit based on this standard character code. Based on this characteristic region data, the emphasis circuit weights the handwriting pattern to be appraised from the character recognition section. Each similarity between the weighted handwriting pattern to be appraised and the individual handwriting data read from the handwriting dictionary is determined by a similarity calculation circuit. The individual corresponding to the maximum similarity among the obtained similarities is identified as the writer of the handwriting to be appraised.

【Example】

Embodiments of the handwriting identification device according to the present invention will be described below with reference to the drawings. The configuration of this embodiment will be explained with reference to FIG. FIG. 1 is a block diagram showing the basic configuration of this embodiment.
In the figure, description of memories or buffers for each output data is omitted. In FIG. 1, numeral 1 denotes a character recognition unit, which outputs an authenticated handwriting pattern Ai which has been formatted as well-known and whose position, size, strength, etc. have been normalized, and a standard character code Bi related to the authenticated handwriting. be done. Reference numeral 2 denotes a handwriting dictionary, which stores standard character code BL feature area data D iL personal code Nn and individual handwriting data C4n, which will be described in detail later. FIG. 2 is a configuration diagram of a handwriting dictionary, in which feature area data Dij for standard character codes Bi and handwriting data Cin for each individual code Nn are expressed in a table format. Note that i is a counter related to characters, j is a counter related to feature regions, and n is a counter related to individuals. FIG. 3 is an explanatory diagram of feature region data, in which A
is the handwriting pattern to be appraised for the character “bun”, K is the character frame, R1
, R2, and R3 are three characteristic regions provided in the handwriting pattern A to be appraised. Each characteristic region R1, R2, R3 is
The left center, lower left corner, and lower right corner of the handwriting pattern to be appraised are set to include locations where the characteristics of a "sentence" are likely to appear, that is, the handwritten part, the left-handed part, and the right-handed part. The data related to each feature region R1, R2, R3 are the coordinates of the upper left corner (starting point) HCH3, R3 of each rectangle indicating the region, and the horizontal and vertical side lengths of each rectangle PLQI;R2,Q2;R3
, Q3. Now, returning to FIG. 1, 3 is a readout circuit that reads handwriting data Cin and characteristic area data D based on the standard character code Bi.
ij is read from the handwriting dictionary 2. Reference numeral 4 denotes an emphasizing circuit which generates the handwriting pattern Ai to be appraised based on the feature area data Dij.
are weighted in a well-known manner. Reference numeral 5 denotes a similarity calculation circuit which calculates each similarity Fin between the weighted handwriting pattern Ei of the emphasis circuit 4 and the individual handwriting data Cin. 6 is a determination circuit that determines the personal code N related to the handwriting data Cix that has the maximum value among each similarity Fin.
x is determined, and the individual X associated with this individual code Nx is identified as the author of the handwriting to be appraised. The operation of the embodiment will be explained with reference to the flowchart of FIG. 4 for the sake of organization. In FIG. 4, in step St, the character recognition unit 1 (see FIG. 1, the same applies hereinafter)
From appraised handwriting pattern A + + standard character code Bi
is input, and in step S2, the reading circuit 3 reads out characteristic region data Dij from the handwriting dictionary 2. In step S3, the emphasizing circuit 4 weights Ai based on Dij to obtain a weighted handwriting pattern Ei. In step S4, the reading circuit 3 reads individual handwriting data Cin from the handwriting dictionary 2, and in step S5, the similarity calculating circuit 5 reads E.
The similarity Fin between i and each Cin is calculated. In step S6, the determination circuit 6 performs a determination process, that is, determines n (-x) corresponding to the maximum value of Fin, and performs a process to specify the personal code Nx. Then, the individual X associated with this specific individual code Nx is identified as the writer of the handwriting to be appraised related to the handwriting pattern Ai to be appraised.

【Effect of the invention】

As described above, in the present invention, a characteristic region is set including a portion where the characteristics of each character are likely to appear, and the handwriting to be appraised is weighted with respect to this characteristic region. Therefore, compared to the conventional matching method, the characteristics of each individual's handwriting to be identified are emphasized, so that the accuracy of identification can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the basic configuration of an embodiment according to the present invention, FIG. 2 is a configuration diagram of a handwriting dictionary, FIG. 3 is an explanatory diagram of feature region data, and FIG. 4 is a flowchart showing the operation of the embodiment. be. Character recognition unit in code explanation, 2: Handwriting dictionary, 3: Continuation circuit, 4: Emphasis circuit, 5: Similarity calculation circuit, Ryo 1st diagram

Claims (1)

[Claims] 1) a character recognition unit that obtains a pattern of the handwriting to be appraised and a standard character code corresponding thereto based on a video signal related to the imaged handwriting to be appraised; the standard character code and the individual associated with this standard character code; a handwriting dictionary storing other handwriting data and data related to a feature area set including locations where the characteristics of each standard character are likely to appear; the handwriting based on the standard character code from the character recognition unit; a reading circuit that reads out the individual handwriting data and the feature area data from a dictionary; an emphasizing circuit that weights the handwriting pattern to be appraised from the character recognition unit based on the feature area data; a similarity calculation circuit that calculates each degree of similarity between the output from the handwriting dictionary and the individual handwriting data read from the handwriting dictionary; and the handwriting corresponding to the handwriting pattern to be appraised based on the output from the similarity calculation circuit. a determination circuit that determines the individual related to the data;
A handwriting identification device characterized in that the writer of the handwriting to be judged is identified based on the output from the judgment circuit.