Signature recognition system and method

• US 20040049411A1
• Filed: 09/08/2003
• Published: 03/11/2004
• Est. Priority Date: 01/13/1999
• Status: Active Grant

• Current Assignee: Computer Associates Think Inc. (Broadcom, Inc.)
• Original Assignee: Computer Associates Think Inc. (Broadcom, Inc.)
• Inventors: Suchard, Eytan, Avni, Yossi
• Granted Patent: US 6,950,538 B2
• Time in Patent Office: Days
• US Class Current: 705/4
• CPC Class Codes: G06F 18/2135 based on approximation criteria, e.g. principal component analysis
• CPC Class Codes: G06F 18/24137 Distances to cluster centroïds
• CPC Class Codes: G06Q 40/08 Insurance
• CPC Class Codes: G06V 40/30 Writer recognition; Reading and verifying signatures

Abstract

A method of authenticating a signature including the steps of sampling a signature and storing data representative of the signature, converting the data to high dimensions vectors, feeding the high dimension vectors to an unsupervised neural network, performing a high order principal component extraction process on the high dimensions vectors to thereby identifying clusters of high dimension points, and analyzing the clusters of high dimension points to determine, based on previously stored information, the authenticity of the signature. Also an apparatus for such authentication including a sampling device for sampling a signature and storing data representative of the signature, a converting device connected downstream of the sampling device for converting the data to high dimension vectors, an unsupervised neural network for receiving the high dimension and performing a high order principal component extraction process on the high dimensions vectors to thereby identify clusters of high dimension points, and an analyzing device connected to the unsupervised neural network for analyzing the clusters of high dimension points to determine the authenticity of the signature.

Claims

1. A method of electronically learning a signature, comprising the steps of:

sampling a signature and obtaining raw data representative thereof using a recursive sampling process;

translating the raw data into high dimension vectors; and

extracting, via an unsupervised neural network, high order principal components of the high dimension vectors by cumulative ortho-normalization.

2. The method of claim 1, further comprising integrating the high order principal components by generating a value r corresponding to a ratio of the number of vectors within an ellipsoid to the total number of vectors and a value s, the value s corresponding to the average of distances of all vectors within the ellipsoid.

3. The method of claim 2, further comprising:

calculating a value A=(average r−

current signature sample r)²/(variance of r) and B=(average s−

current signature sample s)²/(variance of s); and

multiplying the values A and B together.

4. The method of claim 3, wherein multiplying the values A and B together comprises multiplying the values A and B together in a Pi neuron.

5. Software for electronically learning a signature, the software encoded in media and operable when executed to:

sample a signature and obtaining raw data representative thereof using a recursive sampling process;

translate the raw data into high dimension vectors; and

extract, via an unsupervised neural network, high order principal components of the high dimension vectors by cumulative ortho-normalization.

6. The software of claim 5, further operable to integrate the high order principal components by generating a value r corresponding to a ratio of the number of vectors within an ellipsoid to the total number of vectors and a value s, the value s corresponding to the average of distances of all vectors within the ellipsoid.

7. The software of claim 6, further operable to:

calculate a value A=(average r−

current signature sample r)²/(variance of r) and B=(average s−

current signature sample s)²/(variance of s); and

multiply the values A and B together.

8. The software of claim 7, wherein the software operable to multiply the values A and B together comprises the software operable to multiply the values A and B together in a Pi neuron.

9. A computer for electronically learning a signature, comprising:

memory; and

one or more processors collectively operable to;

sample a signature and obtaining raw data representative thereof using a recursive sampling process;

translate the raw data into high dimension vectors; and

extract, via an unsupervised neural network, high order principal components of the high dimension vectors by cumulative ortho-normalization.

10. The computer of claim 9, the one or more processors further operable to integrate the high order principal components by generating a value r corresponding to a ratio of the number of vectors within an ellipsoid to the total number of vectors and a value s, the value s corresponding to the average of distances of all vectors within the ellipsoid.

11. The computer of claim 10, the one or more processors further operable to:

calculate a value A=(average r−

current signature sample r)²/(variance of r) and B=(average s−

current signature sample s)²/(variance of s); and

multiply the values A and B together.

12. The computer of claim 11, wherein the one or more processors operable to multiply the values A and B together comprise the one or more processors operable to multiply the values A and B together in a Pi neuron.

13. A system for electronically learning a signature comprising:

means for sampling a signature and obtaining raw data representative thereof using a recursive sampling process;

means for translating the raw data into high dimension vectors; and

means for extracting, via an unsupervised neural network, high order principal components of the high dimension vectors by cumulative ortho-normalization.