Font Size: a A A

Research Of Identification Algorithm Based On Iris

Posted on:2015-04-21Degree:MasterType:Thesis
Country:ChinaCandidate:Y LiuFull Text:PDF
GTID:2298330431995130Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
With the big environment of rapid development of science and technology and veryconvenient transportation and communication in modern society, identity certification isfrequently encountered problems in our daily life. How to realize the identification fast,efficiently and accurately is a hot and potential research topic. In numerous biologicalrecognition algorithms, iris recognition is an emerging biometric Identify technology. theidentities recognized technology based on the iris has advantages as following several aspects:uniqueness, stability, non-invasion and carrying easily, which can make the iris recognitionoccupy important positions in the field of identifications. A large number of research scholarstake a lot of time and energy to the study of the iris identifications, which makes the study ofthe iris identifications develop rapidly in recent years.This article describes the research status based on iris recognition technology in the fieldof iris recognition at home and abroad boundary point determination, the normalizedprocessing of iris images, iris image feature extraction and coding and feature encodingmatching. Finally, identity authentications are realized by determining feature codecompatibility and threshold size relations. Meanwhile in the process of iris localization, thispaper adopted the inner and outer edge of the separation of the iris localization algorithm,which using the least square circle fitting method. Considering the problem that the edgelocation accuracy is not high, the iris on the circumference of a circle on the edge of theaccurate positioning is realized by using the difference operation for the iris inside edge,while the outer edge of the iris localization is used classical hough transform combined withedge detection algorithm. This paper in the normalized phase adopts an algorithm of arectangular coordinate system to the polar coordinate transformation, which is convenient forthe extraction and recognition of the iris feature in the next stage. The feature extraction basedon2-d Gabor filter is applied the most in iris feature extraction algorithms at present. Thetwo-dimensional Log Gabor filter applied is proposed to conduct the iris feature extraction inthis paper, considering that the widths of the2-d Gabor filter are restricted in consequence,and that the one dimensional Log Gabor filter can overcome well the shortcoming andshortage of2-d Gabor, but meanwhile a Log Gabor can lost in the global information wheniris images such as two-dimensional images are processes. The Log Gabor filter groups areconstructed as multi-channel structures with four dimensions and four directions, which areused to make feature extraction of iris texture. And images which can express iris texture wellin filtering results are encoded in the phase position. In phase matching the matching methodof best shift and standard deviation distance is proposed in this paper, which is compared withthe traditional hamming distance matching algorithm. And then the MATLAB GUI module isapplied to design the iris identification interface. The proposed algorithm is used to carriedout the class and class experiment with the CASIA database on iris image. The experimentalresults show there is a lower standard deviation error rejection rate and false accept rate in theproposed optimal shift difference and distance matching algorithms, that is the proposed algorithm in the paper has better iris recognition rate, which can obtain a better identificationeffect.
Keywords/Search Tags:Iris Recognition, Iris location, feature extraction, Hamming distancematching
PDF Full Text Request
Related items