JP2000251071A - On-line sign recognizing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable recognition with a high recognition rate by using stroke pressure by detecting the maximum value of the stroke pressure, representing extracted stroke pressure as a ratio to the maximum value, and comparing it with the stroke pressure of a registered reference pattern. SOLUTION: An X coordinate, a Y coordinate, and stroke pressure which are inputted are normalized (proportionally distributed) by a number N. Namely, when characters are inputted to tablets 6 and 7, coordinate points are recorded at time intervals and the total time of respective strokes is stored. The movement distances of the respective strokes are summed up to find the total stroke length. The number N for normalization is calculated from the total stroke. The total time is normalized into N vectors, which are each given a stroke trace and stroke pressure to be both converted into fuzzy data. The maximum value PMAX of the stroke pressure of a written pattern is detected and extracted stroke pressure P(n) is represented as a ratio P'(n) to the maximum value PMAX and compared with the stroke pressure P'TXT(n) of the registered reference pattern.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for performing on-line signature recognition using pen pressure.

[0002]

2. Description of the Related Art Conventionally, there has been proposed a method of inputting a signature using a tablet and an electronic pen and recognizing an individual based on data registered in advance. JP-A-4-35
Japanese Patent Publication No. 2080 discloses that in order to increase the accuracy of personal authentication, the subjective feature points of known writing time-series data are fuzzified and registered in a storage unit in advance, and an unknown writing time-series input from an input device is input. A method for performing personal authentication by fuzzy inference using fuzzy rules generated from data on known feature points to dissimilarity between fuzzified data on data feature points and data registered in this storage unit is disclosed. Have been.

[0003]

However, in the conventional method of extracting pen pressure as a feature in online sign recognition and performing recognition by comparing it with dictionary characters, the pen pressure differs depending on the writer and the condition of the input device. In some cases, accurate character recognition could not be performed. An object of the present invention is to provide a recognition method capable of performing recognition with a high recognition rate using pen pressure in online signature recognition.

[0004]

Means for Solving the Problems] To solve the above problems, the online sign recognition method of the present invention is a method of performing online sign recognition using pen pressure, detects a maximum value P _MAX of the writing pressure , The extracted pen pressure P (n) is set to the maximum value P
It is represented by a ratio P ′ (n) to _MAX and compared with the pen pressure P ′ _TXT (n) of the registered reference pattern.

[0005]

Embodiments of the present invention will be described below. FIG. 1 is a block diagram of a system for implementing the online signature recognition method of the present invention. In the figure, 1 is a server having a personal signature database 2, 3 and 4 are personal computers connected to the server 1 via a LAN 5, 6 and 7 are tablets connected to the personal computers 3 and 4, respectively, and 8 is signature authentication software. . In this system, signature data registered in advance is used as a database 2 and stored in the server 1.
Is installed in the personal computers 3 and 4, and the terminal personal computers 4 and 5 in which the signature input tablets 6 and 7 are prepared, input the signature on the tablets 6 and 7 as shown in FIG. When the input is completed, the signature data is read from the server 1 and authentication is performed.

In the present invention, the X coordinate, the Y coordinate, and the pen pressure of the taken signature are normalized (proportionally distributed) by the N number. That is, when characters are input to the tablet, coordinate points are recorded for each time, and the total time Tn for each stroke is stored. Next, a total stroke length is obtained by summing the moving distances of each stroke. From this total stroke, the number N for normalization is calculated.

Next, a sampling interval (time) is calculated. That is, the total time Tn is divided by the number N, the coordinates of the sampling points are calculated, and the movement amount of each sampling point is calculated.

The X and Y coordinates proportionally distributed in this manner are converted into fuzzy angle information. First, it is converted into regular angle information.

As shown in FIG. 2, Δx and Δy are obtained, and the angle is obtained by the following equation. θ = arctan | Δy / Δx | (Δx> 0, Δ
y> 0) θ = 180−arctan | Δy / Δx | (Δx <0, Δ
y> 0) θ = 180 + arctan | Δy / Δx | (Δx <0, Δ
y <0) θ = 360−arctan | Δy / Δx | (Δx> 0, Δ
y <0)

Next, the angle information is fuzzy. This fuzzy angle information is represented by μ = (μ ₀ , μ ₁ , μ ₂ , μ ₃ ) and is obtained by the following equation and Table 1. _{μ A = 1-θ / 90} ° μ B = θ / 90 °

[0011]

[Table 1]

The reason why the proportionally distributed pen pressure is expressed as a ratio is that the strength of the force is almost the same for each person, although the degree of power input varies depending on the day and time.
As a result, a vector having fuzzy angle information and fuzzy pen pressure information is generated.

The angle information of the vector of the input sine is given by μ _T0 ,
Assuming that μ _T1 , μ _T2 , μ _T3 and the angle information of the vector of the reference data are μ _D0 , μ _D1 , μ _D2 , μ _D3 , the method of calculating the similarity Sn between certain vectors is expressed by the following equation.

[0014]

(Equation 1) In the above equation, max is the larger value of the two equations,
min is the smaller value.

Using this equation and DP matching, the similarity between the respective vectors is obtained, and the final similarity is obtained by the following equation. S = (S ₀ + S ₁ + S ₂ + S ₃ ... + S _n ) / n Similarly, the similarity of the pen pressure is obtained.

FIG. 3 is a flowchart showing the flow of a signature authentication program according to the present invention. Step 100: The signer signs the tablet with a pen. At this time, the coordinates are taken from the tablet. Step 110: The end of the signature is detected when the pen is separated from the tablet for a predetermined time or more. Step 120: If it is determined that the signature has been completed, it is determined whether there is a reference pattern registered in advance as the signature of the person. Step 1 if there is a reference pattern
Proceed to step 30, otherwise proceed to step 160. Step 130: Normalize the input pattern. Step 140: Compare with the reference pattern. Step 150: Output the result by collating with the reference pattern. Step 160: If there is no reference pattern in step 120, specify whether to register as a reference. If the reference is to be registered, go to step 170; otherwise, go to step 1.
Return to 00. Step 170: Normalize the reference pattern. Step 180: Register as a reference pattern.

The procedure for normalizing the reference pattern in step 170 is shown in the flowchart of FIG. Step 200: Store the total time of the pattern. Step 210: Detect the total stroke of the pattern. Step 220: Calculate the N number from the total stroke. Step 230: The pen pressure is represented by a ratio. Step 240: Normalize the total time to N pieces (fuzzy handwriting and pen pressure).

The procedure of input pattern normalization in step 130 of FIG. 3 is shown in the flowchart of FIG. Step 300: The ratio with the total time of the reference pattern is obtained, and the time axis is adjusted. Step 310: Handwriting is represented by a ratio. Step 320: Normalize from total time to N (fuzzy handwriting and pen pressure). The time axis always shifts even if the person writes. If the input is normalized as it is, the deviation of the time axis greatly affects the authentication. Therefore, by adjusting the time axis as shown in FIG. 5, the influence on the authentication is reduced.

FIG. 6 shows a procedure for comparison with the reference pattern in step 140 of FIG. Step 400: Extract similarity of the most similar vector using DP matching. Step 410: Determine whether or not N matchings have been completed. When the processing is completed, the process proceeds to step 420; Step 420: Calculate (average) the total similarity. In the DP matching part of step 400, DP matching is basically performed, and the matching calculation is performed by mi between input data and reference data as shown in FIG.
An n-max operation is performed, and the one with the highest similarity is selected.

As an example, S11 = 0.666, S12 = 0.
777, S13 = 0.888, S14 = 0.543, S15 =
If 0.333, then S13 is the most similar, so S
1 = 0.888. The next matching starts from the third place. As described above, DP matching is performed, and a total similarity is calculated by the following equation using the obtained similarities. S = (S1 + S2 +... + Sn) / n

At the time of normalization from the total time to N vectors, each vector is given a handwriting (direction information) and a pen pressure. Here, both the handwriting and the pen pressure are converted into fuzzy data. Handwriting is converted to fuzzy data in the same manner as character recognition (expressed in four directions of fuzzy grades).

As the pen pressure, the maximum value of the pen pressure of the written pattern is detected, and the value is set to 100. The other detected pen pressures are expressed as a ratio using the following equation. P ′ (n) = (P (n) / P _MAX ) × 100 P ′ (n); n-th writing pressure P (n) expressed as a ratio; n-th writing pressure P _MAX ; written pattern Writing pressure maximum

Next, a rule as shown in FIG. 8 is prepared, and the pen pressure obtained by the above equation is applied to the rule, and expressed by three fuzzy grades.

The method of calculating the number N from the total stroke is as follows. If total stroke TotalS
If the stroke is 10,000 or less, N = 100, and if it is more than that, N = (TotalStroke / 1000) × 10

In this way, by expressing the pen pressure as a ratio and comparing it with the pen pressure of a dictionary character, recognition with a high recognition rate becomes possible.

[0026]

As described above, according to the present invention, the maximum value of the pen pressure is detected, each of the extracted pen pressures is represented by the ratio with the maximum value, and the pen pressure is compared with the pen pressure of the registered reference pattern. Thus, even when the writing pressure varies depending on the writer or the condition of the input device, it is possible to perform online signature recognition with a high recognition rate.

[Brief description of the drawings]

FIG. 1 is a block diagram of a system for implementing an online signature recognition method of the present invention.

FIG. 2 is an explanatory diagram showing a method of normalizing an input stroke.

FIG. 3 is a flowchart showing a flow of a signature authentication program according to the present invention.

FIG. 4A is a flowchart illustrating a procedure for normalizing a reference pattern, and FIG. 4B is a flowchart illustrating a procedure for normalizing an input pattern.

FIG. 5 is an explanatory diagram of a method of adjusting a time axis.

FIG. 6 is a flowchart showing a procedure for comparing with a reference pattern.

FIG. 7 is an explanatory diagram of a matching method between input data and reference data.

FIG. 8 is a graph of a fuzzy rule used in the present embodiment.

[Explanation of symbols]

1 server, 2 personal signature database, 3,4 PC, 5 LAN, 6,7 tablet, 8 signature authentication software

Claims (1)

[Claims] 1. A method for performing online sign recognition using pen pressure, wherein a maximum value P _{MAX of} pen pressure is detected, and each extracted pen pressure P (n) is a ratio P ′ to the maximum value P _MAX. An on-line signature recognition method represented by (n) and compared with the pen pressure P ′ _TXT (n) of a registered reference pattern.