A Pipelined Approach for Fpga Implementation of Bi Modal Biometric Pattern Recognition

A Biometric system is essentially a pattern recognition system that makes use of biometric traits to recognize individuals. Systems which are built upon multiple sources of information for establishing identity which are known as multimodal biometric systems can overcome some of the limitations like noisy captured data, intra class variations etc... In this paper a Bi modal biometric system of iris and palm print based on Wavelet Packet Transform (WPT), gabor filters and a neural classifier implemented in FPGA is described. Iris is the unique observable visible feature present in the detailed texture of each eye. Palmprint is referred to the textural data like principal lines wrinkles and ridges present in the palm. The visible texture of a person’s iris and palm print is encoded into a compact sequence of 2-D wavelet packet coefficients constituting a biometric signature or a feature vector code. In this paper, a novel multi-resolution approach based on WPT for recognition of iris and palmprint is proposed. With an adaptive threshold, WPT sub image coefficients are quantized into 1, 0 or -1 as biometric signature resulting in the size of biometric signature as 960 bits. The combined pattern vector of palm print features and iris features are formed using fusion at feature level and applied to the pattern classifier. The Learning Vector Quantization neural network is used as pattern classifier and a recognition rate of 97.22% is obtained. A part of the neural network is implemented for input data of 16 dimensions and 12 input classes and 8 output classes, using virtex-4 xc4vlx15 device. This system can complete recognition in 3.25 microseconds thus enabling it being suitable for real time pattern recognition tasks.