Enhanced discrete cosine transformation feature based iris recognition using various scanning techniques

Distinctive and irreversible features of the iris make its most secure and trustworthy biometric modality for person identification. This paper presents the comparison of various scanning techniques for feature extraction in iris recognition. Iris localization and segmentation was performed using Circular Haugh Transformation, which estimates the parameters of iris using edge map. Normalization on the segmented iris was performed using Rubber sheet model, to convert the circular iris in-to a rectangle of fix dimension. Thereafter Discrete Cosine Transformation coefficients were extracted from the normalized iris using different scanning techniques. The scanning techniques used are Zigzag, Raster, and Saw-tooth. Experimental results show the promising performance of Raster Type-II scanning technique with 100 coefficients. The database used for the observations is CASIA iris database version-IV. The analysis and experimental results show that proposed scheme can be used in iris recognition systems for better performance.