Image analysis – Applications – Personnel identification
Reexamination Certificate
1999-01-14
2002-08-06
Boudreau, Leo (Department: 2621)
Image analysis
Applications
Personnel identification
C382S215000
Reexamination Certificate
active
06430308
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a hand writing verification device for conducting the verification of handwriting in accordance with information of the handwriting. For example, the present invention relates to a hand writing verification device for ascertaining the truth of an inputted signature by verification information of the inputted signature with information of the previously registered signature.
2. Description of the Related Art
FIG. 5
is a view showing an example of information of handwriting obtained when handwriting of “Ogawa” is written on a tablet with an electronic pen. In this case, the point
20
is defined as a sampling point, at which sampling is conducted at regular intervals on information of a position and writing pressure on the tablet with which the electronic pen comes into contact. Also, in this case, the line
21
is defined as a stroke, which is a line of the sampling points obtained while the electronic pen is continuously coming into contact with the tablet. Accordingly, in the example shown in
FIG. 5
, information of the handwriting “Ogawa” inputted as an electronic signal is composed of 21 sampling points and 6 strokes.
As a method of ascertaining the truth of handwriting by verification a pattern of handwriting, which has been inputted as an electric signal, with a pattern of handwriting, which has been previously registered, for example, there is proposed a handwriting verification method which is disclosed in Japanese Unexamined Patent Publication No. 2-268373.
First, the conventional handwriting verification method will be explained as follows.
FIG. 6
is a flow chart showing operation of the conventional handwriting verification device.
In step S
11
, the registered pattern is called. In this step, the registered pattern, which has been previously made of a signature of a person himself, is called. In step S
12
, the signature is inputted, that is, the signature of the writer is taken in as three-dimensional time series information of coordinate information and writing pressure information. In steps S
13
and S
14
, normalization is conducted. In step S
13
, the number of sampling points is made to be the same value of L, and in step S
14
, the position and size of the signature are normalized.
In step S
15
, DP matching (dynamic programming matching) is used, and a time strain function is found so that a cumulative discrepancy between the registered pattern and the input pattern can be minimized.
In step S
16
, time of the input pattern is corrected by the above time strain function, and a discrepancy of the profile from the registered pattern is calculated as a first discrepancy. Formula (1) shows a discrepancy “d” between the sampling point, the order of which is “a” of Z reg of the registered pattern, and the sampling point, the order of which is “b” of Z inp of the inputted pattern. In formula (1), Wp (0≦WP≦1) is a weighted coefficient, P reg is a piece of writing pressure information of the registered pattern, and P inp is a piece of writing pressure information of the input pattern.
d
2
(
a, b
)=(1
−Wp
)|
Z reg
(
a
)−
Z inp
(
b
)|
2
+Wp|P reg
(
a
)−
P inp
(
b
)|
2
(1)
Z reg=X reg(I)+i Y reg(I)
Z inp=X inp(I)+i Y inp(I)
I=1, 2, . . . , L
i={square root over (−1)}.
In step S
17
, a cumulative discrepancy of the time strain function itself is calculated as a second discrepancy dm. The second discrepancy dm is defined by the following formula (2), wherein the time strain function is {a(k), b(k)}.
dm=∥a
(
k
)−
b
(
k
)∥
&LeftDoubleBracketingBar;
a
(
k
)
-
b
(
k
)
&RightDoubleBracketingBar;
2
=
1
K
k
∑
k
=
1
K
&LeftBracketingBar;
a
(
k
)
-
b
(
k
)
&RightBracketingBar;
2
(
2
)
a
(
1
)=
b
(
1
)=1
a
(
K
)=
b
(
K
)=
L.
In step S
18
, the signature is verified using the first discrepancy and the second discrepancy.
As described above, according to the conventional signature verification method, a discrepancy of the profile is found as the first discrepancy by dynamic programming matching of the sampling point of the input pattern with the sampling point of the registered pattern. Further, a cumulative discrepancy of the time strain function itself is found as the second discrepancy, and the signature is verified using them.
However, in the conventional signature verification device described above, the following problems may be encountered. When the conventional signature verification device is used, it is impossible to precisely evaluate a local characteristic of the profile of the signature. That is, in the above method of dynamic programming matching, for example, when a plurality of sampling points are made to correspond to one sampling point, a difference of the coordinate value between the one sampling point and the plurality of sampling points, which are made to correspond to the one sampling point is made to be a local discrepancy, and also a difference of the writing pressure value between the one sampling point and the plurality of sampling points, which are made to correspond to the one sampling point, is made to be a local discrepancy. That is, only information of the above points is used, and information of the peripheries of the above points is not used. Therefore, it is impossible to precisely express a local difference of the profile.
Further, the following problems may be encountered in the conventional signature verification device. In the conventional signature verification device, all characteristics used in the verification are used when dynamic programming matching is conducted. Therefore, when the types of characteristics used in the verification are increased, processing time required for the signature verification is increased. That is, dynamic programming matching is conducted when coordinate information and writing pressure information are used as the characteristics used for the verification. However, in order to enhance the accuracy of the signature verification, it is necessary to add the types of characteristics used for the verification. Therefore, in accordance with the increased number of the characteristics to be added, processing time required for the signature verification is increased.
As described above, evaluation of the characteristics accompanied by a change in time is conducted only by the time strain function in the conventional signature verification device described above. Accordingly, it is impossible to conduct an accurate evaluation on the characteristics which change in accordance with a change in time. For example, it is impossible to conduct an evaluation on the characteristics such as a change in the writing pressure and a change in the writing speed which change in accordance with a change in time.
SUMMARY OF THE INVENTION
The present invention has been accomplished to solve the above problems, and therefore an object of the present invention is to provide a handwriting verification device characterized in that: a local difference in the handwriting can be accurately reflected in the verification of the handwriting; even when the types of characteristics used for the verification are increased, it is possible to suppress an increase in the processing time required for the signature verification; and an accurate evaluation can be conducted even on a characteristic which changes in accordance with a change in time.
According to one aspect of the present invention, there is provided a handwriting verification device comprising: a normalizing means for normalizing an inputted handwriting which has been subjected to sampling at regular intervals; a registered handwriting dictionary in which registered handwriting is accommodated; a correspondence making means for making the inputted handwriting, which has been normalized in accordance with the sampling points in the sampling, correspond to the registered handwriting
Kawamata Takenori
Ogawa Isamu
Yoda Fumio
Boudreau Leo
Mitsubishi Denki & Kabushiki Kaisha
Werner Brian
LandOfFree
Handwriting verification device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Handwriting verification device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Handwriting verification device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2960227