Research And Implementation Of Iris Feature Extraction And Matching

This paper firstly compares several kinds of biometric technology, and describes theadvantages of iris recognition compared to other biometric. Then some key technologies ofiris recognition systems are studied, and an iris recognition system platform was designed. Inthis paper, the main work focuses as follows:(1)For iris segmentation, we propose a method of super-pixel segmentation of irissegmentation. First, we apply the super-pixel SLIC algorithm for iris segmentation, followedby the extraction of each super-pixel normalized histogram of the region as the feature of thesuper-pixel, then we combine the adjacent superpixels that meet the determination function,which is based on the correlation distance. In the merger process, we first use the averagegray value to find the pupil area, then make use of the pupil super pixel as a starting point tobegin with, then every superpixel will be mergered with the surrounding pixels. This processwill be iterated until convergence merge, and then we extracted the largest superpixelconnected to pupil as the iris region. Experiments shows that the proposed method cansegment the iris more efficient and completion.(2)In the process of iris feature extraction and matching, we make use of an elastic modelof iris. The iris is considered as the superposition of two inner and outer circle of the extractediris region, and is normalized to a rectangular iris to expand. Then we apply Gabor featureextraction to the normalized iris image, then the extracted features can be calculated withHamming distance to determine whether the compared iris are the same person or not.(3)In iris recognition system platform, we introduce the components of the systemsoftware features iris recognition platform, and discusses the background database softwaredesign methods, design a perfect iris offline/online recognition software platform.