Iris Image Compression using Wavelet Based Directional Transform through DT CNN Dynamics

In this paper, we have presented a compression scheme for iris images using Wavelet Based Directional Transform (WBDT) through Templates of Discrete Time Cellular Neural Network (DTCNN) and Directional Filter Bank (DFB). The complex annular part of the iris portion of the eye image contains many distinctive features such as arching ligaments, furrows and ridges. The compression algorithms developed for iris images have to preserve the details present in the iris part of the image, which are used for subsequent biometric processes. The directionality features can be very well analyzed by means of Directional Filter banks in WBDT than Wavelet decomposition. The decomposed image using WBDT can be coded effectively by using modified SPIHT encoding. The encoder output is further compressed using SOFM based VQ coder. The subjective quality of the reconstructed images obtained is comparable with the 2D wavelet decomposition. It is inferred that atleast 10dB improvement can be seen over wavelet based technique for the same entropy. The results obtained are tabulated and compared with those of the wavelet based ones.