- Features a fast, easy to use and good accuracy system
- Using an artificial neural network with a new training technique for verification
- Using a new threshold method where both the neural network output and selected global features are used to obtain an improved false acceptance and false rejection rate
- Can be used for PC log-on and internet log-on
How the system works
During Enrollment
- Signature data such as pen position and pen pressure with respect to time are captured using a digitizing tablet and act as the input to the system
- Important features that can characterize the signature are extracted after data has been pre-processed to ensure uniformity
- Signature data are validates using a correlation technique to decrease the variations among genuine signatures
- Signature data are trained using a novel neuro-template method
- Neuro-template of each person will be stored in database
- User will sign in
- Signature data of user is captured and pre-processed
- Important features of signature data, and the neural network output are input to the threshold module
- If these values are within the threshold levels set by the system, then the person will be authenticated and can gain access otherwise access will be denied
- Network environment needs accurate, fast and automatic authentication of users
- PINs, password, smart card are methods of authentications but can be forgotten, misplaced, or stolen
- Signature, a behavioral biometrics identification is the most accepted method of asserting a person's identity
