Image analysis – Pattern recognition – On-line recognition of handwritten characters
Reexamination Certificate
1999-05-07
2001-04-03
Mehta, Bhavesh (Department: 2621)
Image analysis
Pattern recognition
On-line recognition of handwritten characters
C382S119000
Reexamination Certificate
active
06212295
ABSTRACT:
COPYRIGHT NOTICE
A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
BACKGROUND OF THE INVENTION
The present invention relates generally to data verification applications, and more specifically, to the dynamic reconstruction of a signature generated on a digitizer tablet.
The process of visually comparing signatures to check for forgeries is well-established. For example, banks have signature cards for their customers on file, and tellers can compare the signature on a check or other document with the signature card if there is any question as to the authenticity of the signature. In other instances it may be necessary to compare electronically generated signatures. This form of signature can be authenticated using either an automated, computer based verification system or by relying on human examiners.
In general, signature verification techniques rely on the static or “as written” image of the signature. Unfortunately, reliance on this form of the signature only provides a modicum of security since a skilled forger can often duplicate a static signature with sufficient accuracy to circumvent detection. Furthermore, it is normally quite easy to obtain static images of a person's signature, thus providing the would-be forger with all of the necessary information.
Digitizer tablets are known for data input applications. Typically, digitizer tablets are used for capturing primarily spatial and sequential information. When something is written on a digitizer tablet, the output is a parametric representation of the writing trajectory; that is, the writing is represented as a series of x, y coordinate values as a function of time. A typical digitizer tablet samples the position of the writing pen 100 times per second, at fixed intervals of 0.01 seconds. The tablet continuously transmits the resulting data to the receiving computer. Thus every 0.01 seconds, the tablet samples the pen position and transmits an x-coordinate value, a y-coordinate value, and an additional value which indicates whether the pen is up or down. Since the sampling occurs at uniform time intervals, the physical or spatial distance the pen travels between samples is proportionally larger where the writing is fast and proportionally smaller where the writing is slow. The velocity of writing is proportional to the distance travelled between sampled points. This feature may be applied to other data gathering functions.
What is needed is a method of dynamically reconstructing handwritten information, thereby providing a highly reliable technique for detecting a forgery.
SUMMARY OF THE INVENTION
According to the invention, a method is provided for reconstructing from stored data generated on a digitizer tablet a representation of the dynamics of the input based on the dynamics of the data generation. In a specific embodiment, values useful for dynamic reconstruction are created by accumulating increments of values that are some function of the original data values. To enhance the dynamics, a nonlinear function is preferred. For example, a suitable increment is the square of the spatial difference between samples in the original data. The resulting dynamic reconstruction would show the portions of the signature that are written relatively quickly to be proportionally stretched out, while the areas that are written relatively slowly would be compressed. A second example having the opposite effect is a reconstruction in which the increment is the reciprocal of the difference between the original data points. This function results in compression of fast writing and expansion or stretching of slow writing.
Although the choices for the processing function are limitless, the basis of a first specific embodiment of this invention is to accentuate the velocity component of the original handwritten data. Different functions simply result in different degrees of dynamic enhancement.
The dynamic reconstruction of data may be difficult to read or analyze since some areas of a signature will be expanded while other areas will be compressed, often resulting in an overlap. In another embodiment of the invention, space is added between the individual strokes of the handwritten data, thereby simplifying analysis. (Note: a “stroke” is defined as the continuous segment which begins when the pen touches the digitizer tablet's surface and ends when the pen is lifted from the surface.) By spacing out the strokes, overlapping can be prevented. Spacing out the strokes has the added advantage of clearly displaying the stroke sequence with which the original signature was written. Therefore a forgery can be detected by noting an improper stroke sequence; for —example, at what point in a signature a “t” is crossed or an “i” is dotted.
In application, dynamic reconstruction can either be used singularly, or in conjunction with conventional forgery detection techniques. It is a highly effective signature verification technique since a would-be forger must not only reproduce the static image of a signature, but also must simultaneously reproduce the dynamic image of the signature. The forger's job is further complicated by the difficulty associated with obtaining accurate information regarding signature dynamics. Thus, unless a forger is able to reproduce the forces and motions used to generate a signature, even a highly accurate static tracing would be detectable.
In a typical application, the verification process would be performed by comparing a specimen signature generated by the person desiring to be authenticated with a stored reference signature known to belong to the person. Reference signatures can be stored in hard copy or in computer storage. In the preferred embodiment, the static image and the dynamic reconstruction of the specimen signature would be displayed on a CRT computer screen next to the appropriate reference signatures. The examiner would then compare the images on the CRT screen and determine if the specimen signature is authentic.
The term “signature,” as used herein in the specification and the claims, is not meant to be limited to the name of a person. Any sequence of words, characters, or symbols can be processed by this invention and are intended to be within the meaning of the term.
REFERENCES:
patent: 4344135 (1982-08-01), Crane et al.
patent: 4646351 (1987-02-01), Asbo et al.
patent: 5347589 (1994-09-01), Meeks et al.
patent: 5933514 (1999-08-01), Ostrem et al.
Austin Norman A.
Crane Hewitt D.
Ostrem John S.
Allen Kenneth R.
Communication Intelligence Corporation
Mehta Bhavesh
Townsend and Townsend / and Crew LLP
LandOfFree
Method for dynamic reconstruction of handwritten data does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for dynamic reconstruction of handwritten data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for dynamic reconstruction of handwritten data will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2467239