Iris localization via intensity gradient and recognition through bit planes

Iris recognition is very hot topic in both research and practical applications. In this paper, a robust algorithm is proposed for iris localization and a very simple method is employed for feature extraction. Iris localization is the key step in iris recognition systems because all subsequent steps depend highly on its accuracy. The proposed algorithm utilizes important property of gradient of intensity level in the greyscale images (after converting the images into greyscale if not). Then iris is normalized into a dimensionless rectangular strip of size 128*512 pixels and different features are extracted based upon bit plane slicing of the strip to get binary iris code. ROC curves are also drawn for different features. Matching decision is based on accumulative sum of bitwise XOR of different iris codes. Experiments show that proposed localization algorithm is very effective. Results have been tabulated by evaluating the developed algorithm with 1000 eye images and recognition accuracy has reached up to 99.6%.