A palmprint based biometric authentication system using dual tree complex wavelet transform

Palmprint identification has emerged as one of the most popular and promising biometric modalities for personal identity verification due to its ease of acquisition, non-invasive procedure, high user acceptance and reliability. This paper proposes the development of a new method for palmprint based biometric authentication which utilizes the textural information available on the palmprint by employing the Dual Tree Complex Wavelet Transform (DTCWT). The method proposes to construct a region of interest (ROI) for the scanned color images of the palm, and then determine a histogram of the two dimensional image. This enables to utilize a feature extraction module, implemented using the one-dimensional (1D) Dual Tree Complex Wavelet Transform (DTCWT) on the histogram signal. The DTCWT is an improvement over the discrete wavelet transform (DWT) as it provides nearly shift invariant performance, reduced aliasing and directional wavelets in higher dimensions. Backpropagation neural-network (BPNN) based binary classifiers are developed for authentication utilizing the features extracted. The system is developed on the basis of several scanned color images of palms of individuals in real-life, in our laboratory. The experimental results obtained from the data have demonstrated the utility of the proposed system, by exhibiting an overall mean accuracy as high as 98.35%.