ANovelMethodofIris Feature Extraction Basedonthe1CM
Zhaobin Wang
2006-01-01
Abstract:Thefeature ofiris imagewouldbeexactlywherepolice discriminated criminal byinspecting their andefficiently extracted difficultly bycommonmethods.irises color. Until 1987, FlomandSafir proposed anidea Considering that theIntersecting Cortical Model(ICM)can ofautomated iris recognition (1,2,3).Recently, the beusedtodirectly extract pertinent information froma popular methodoffeature extraction isdeveloped by variety ofimages forthepurpose ofrecognition, wefirstlyDaugmanbased on2DGaborfilter (1,3,4),while other propose anovel method offeature extraction foriris image methods arealsoproposed. Andmostexisting iris basedontheICM.Theresults showthatourmethod is recognition methods obtain either themultiresolutional feasible, potential andeffective inobtaining feature from orthedirectional features fromaniris pattern using iris image. Inourmethod, aseries ofbinary images are Gaborfilters orwavelet transforms (1,5,6,7).Butin firstly produced fromtheinput iris image through theICM. mostcases thefeature isunable tobewellgained witha Entropy sequence canbegained fromthese binary images. variety ofreasons. Andthenphaseinformation isobtained fromentropy TheIntersecting Cortical Model(ICM)introduced sequence. Thisphase information istaken asfeature vector byJ.Kinser (8), hasmoretremendous advantages. First, withtheuniqueness fordifferent individuals. Finally, we theICM hasan inherent ability to extract the carry outmatching experiment using thephase information. fundamentals oftheimage. Forexample, theICMcan Theadvantages, disadvantages andfurther workare refine thefundamentals ofanimage(edges, textures, and pointed outintheend. segments) andcanpresent farsimpler binary images to IndexTerms-feature extraction, entropy, ICM,iris recognition engines foranalysis. Second, theICMcan recognition, phase simplify theimagetoallowrecognition engines to perform afareasier task thaniswithin their realm. The advantages oftheICM inthisformshould notbea surprise since itisbasedonhowmammalsperform Iris recognition refers toabiometric technology that recognition. So theICM havemany potential identifies anindividual using theannular iris pattern that applications inmanyfields suchasimage analysis, image surrounds thepupil oftheeye. A iris pattern isunique to segmentation, feature extraction, changedetecting, eachperson andnearly immutable through life, plusit coding, pattern recognition, etc. includes richdiscriminatory information, suchas Applying theICMtothefield ofiris recognition in collagenous fibers, contraction furrows, coronas, crypts,this paper, wefirstly propose anovel method offeature colors, freckles, rifts, andpits(1).Therefore, iris extraction foriris imagebasedontheICM.Inour recognition canbeahighly reliable solution forpersonalscheme, theregion ofinterest (ROI) intheiris imageis identification orverification. sent into theICM,andthenaseries ofbinary images are Although there isrich information intheiris pattern,firstly produced fromtheinput though ICM.Andentropy exploiting their geometric feature isnotaseasyasin sequence canbegained fromthese binary images. The fingerprints orfaces. Thefirst useofiris recognition can phaseexacted fromtheentropy sequence canbe trace backtotheParis prison ineighteenth century,considered asthefeature vector ofthestimulant image.