Improvement of iris location accuracy in iris recognition

Locating the pupil circle is an important step in iris recognition. As the light shines on the pupil, a bright reflection spot is formed in the iris image. In the process of iris location using Hough transform, location accuracy will be affected, because the reflection spot cannot be eliminated in practical optical system. Therefore, we proposed an algorithm to eliminate the reflected light spot in the pupil. The average gray value of the iris image can be obtained by using the gray histogram of the iris image. Then convolution of iris image with summation matrix of ten by ten is obtained, and the sum of gray values is less than a hundred times the average gray value and the approximate position of pupil can be found out. The pupil position is used to get the(1) approximate position of the reflected highlight spot. The threshold is set to average threshold plus 128, and the point in the pupil field that is larger than the threshold can be obtained, and the position of the reflection point can be determined. The positions of the spot are set to zero. Using the algorithm, the number of highlight point in the binary iris image is reduced effectively. Thus Hough circle location would get higher location accuracy.