Research On Iris Feature Extraction And Recognition Algorithms

Iris recognition is a biometric technology with the highest security and stability,it is the best identification technology in the future,with broad market prospects and scientific research value.At present,the research work of iris recognition technology is divided into three parts,namely,iris image acquisition,pretreatment and feature extraction.This paper focuses on the iris features extraction.The difficulty of iris feature extraction is how to extract stable and effective feature information,filter out unstable feature information caused by image distortion and unclear texture.In order to solve these problems,this paper proposes a new feature extraction algorithm and image deformation compensation method.The main contents of the article are as follows:Firstly the research status of iris recognition technology at home and abroad is introduced.The technical principle and system framework of iris recognition are summarized.The algorithm of iris image used in preprocessing is briefly described.Secondly,we study the representative feature extraction algorithm and Gabor feature extraction algorithm at home and abroad.Based on the principle and framework of extracting iris features from two algorithms,we design the extraction algorithm based on texture direction energy feature,and use these three algorithms Three feature extraction schemes are designed.Thirdly,it is found that the iris feature extracted by the algorithm becomes unstable when the image rotates and scales,and the traditional method only compensates the deformation caused by the rotation of the image without compensating the deformation caused by the image scaling.In this paper,we propose a method to encode the local extreme points of feature images as effective feature points,which can solve the above problems effectively.Finally,the CASIA iris library of Chinese Academy of Sciences was used as the experimental sample,and the three kinds of feature extraction schemes and image deformation compensation methods were tested.The results of the six experiments show that the iris feature extraction scheme based on the texture direction feature extraction algorithm obtains the best recognition effect in the three schemes,and the recognition performance of the algorithm is improved again after combining the image deformation compensation method.