Recognize Algorithm Of Iris Based On Gabor Filtering

Iris is a biometric feature which has many unique advantages, so iris recognition is regarded as one of the most reliable and accurate biometric identification system available. A new improved iris recognition algorithm which based on 2D Gabor transform is brought forward in this thesis.An iris recognition algorithm includes three modules: image pre-processing, feature extraction and classification. In image pre-processing, the Random Hough transform is improved with the geometry features of pupil, gray distribution of images and many prior knowledge. Iris boundary localization is improved by selecting the small block images and reducing the number of edge points, and images are optimized by the means of Mathematical morphology and threshold interception, so many interference points are wiped off. This localization algorithm is as accurate as the Hough transform location algorithm and is more efficiently.In the feature extraction period, an algorithm which uses location information and form information of local regions of iris in different scale is used. In this algorithm, local regions of iris are quantitatively described by using multi-channel 2D Gabor filter. In order to get better filter result, a new strategy to search the optimal filter with the help of the bandwidth parameter of Multi-channel 2D Gabor filters.In feature extraction, the centers of the small block images which are similar with filter are chose as feature points. The weighted average of these feature points is chose as the final feature. Local regions of iris are reflected more veracious by finding optimal weighted algorithm.The presented iris recognition algorithm performs with good recognition rates of 99.754% on the images of CASIA iris database. Therefore, it is proved to be a reliable and accurate recognition algorithm.