On-Line Signature Verification by Exploiting Inter-Feature Dependencies

The traditional on-line signature verification process involves use of various dynamic features such as velocity, pressure, acceleration, angles, etc. The idea is to device a composite vector structure combining more than one feature where each feature is treated independently. Our proposed research work is an attempt to exploit the inter-feature dependencies by employing a higher dimensional vector approach. The strategy adopted here is to obtain pressure strokes with respect to various velocity bands. The strokes thus obtained are found to portray a reasonably accurate basis for discriminating genuine vs forgery class. The simulation results validate our assumptions and show improvements in the discriminating index