We present an automatic handwritten signature verification system to prevent identity fraud by verifying the authenticity of signatures on Australian passports. In this work, fuzzy modeling has been employed for developing a robust recognition system. The knowledge base consists of unique angle features extracted using the box method. These features are fuzzified by an exponential membership function, consisting of two structural parameters which have been devised to track even the minutest variations in a personu27s signature. The membership functions in turn constitute the weights in the Takagi-Sugeno (TS) model. The optimization of the output of the TS model with respect to the structural parameters yields the solution for the parameters. The efficacy of the proposed system has been tested on a large database of over 1200 signature images obtained from 40 volunteers achieving a recognition rate of more than 99%.
展开▼
