Font Size: a A A

Study On Iris Recognition Based On Gabor Filtering

Posted on:2011-09-05Degree:MasterType:Thesis
Country:ChinaCandidate:Q F ZhengFull Text:PDF
GTID:2178360308983822Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Nowadays, information security is becoming more and more important, owing to the rapid development of information technology. The biology recognition technology owns some advantage which the traditional identity verification methods have not, and will have longer development in the society. Among all the existing kinds of biological recognition technology, iris recognition technology gets more attention with such as high reliability, uniqueness, non-forgery and non-infringing etc. Iris recognition technology has been regarded as one most development biology recognition technology in the future.According to the characteristics of iris image in the iris database adopted, and study on the iris recognition principle previously, an improved principle is proposed for the purpose of classing the irises precisely and quickly. The main works for the research have been accomplished as following:Firstly, eyelashes can interfere with the iris location on a certain extent, and make some disbennifit for the accuracy in the iris segmentation. Morphological closing with a right structuring element is used to remove eyelashes through analyzing the features of the morphology operation.Secondly, for the purpose of localizing iris precisely and quickly, an improved localization is proposed while based on introducing some prevailing localization algorithms. First, some typical problems appeared in practice are considered, such as eyelids, eyelashes and their shadow. To obtain the real boundaries of the pupil, the boundaries must be corrected by removing the conglutination between the typical problems and pupil region in the binaried image. Second, the boundaries of the pupil can't be modeled by circle accurately. To improve the quality of segmentation, the direct least square fitting of ellipse could be used for detecting the boundaries of the pupil. Finally, the morphology method is introduced to fill up the gaps among the outer edges that are detected by Canny operator and makes the edge detected more easily and fitted more nicely . Experiments have been shown the applicability and effectivity of the algorithm. And the algorithm is faster and while has a better location accuracy .Thirdly, the paper analyzes the features of the Gabor filter and studies the encoding arithmetic of Daugman based on the phases. 2D-Gabor filters with different frequencies and orientations filter the every sub-block in the iris normalized image , and encodes the iris texture by plus or minus to the coefficient of the imaginary unit in the 2D-Gabor filter. In the feature extraction, the method projected overcomes the limitation of the non- integrity feature extraction by Gabor filters with different frequencies on the single orientation, and finally can class the iris more accurately.Experiments have been shown that the method for identity recognition can work well.
Keywords/Search Tags:iris recognition, iris location, edge correcting for the pupil, least square principle, ellipse fitting, edge connecting, 2D Gabor filter, feature extraction
PDF Full Text Request
Related items