An Approach For Iris Recognition Based On Singular Value Decomposition And Hidden Markov Model

Biometric Recognition Technology develops very fastrecently, which identify people by some sole and stable features.The operation of this technology is convenient and rapid.Meanwhile, there doesn't exit problems such as cryptogram gotout or card lost which is present in cryptogram identification andcard identification respectively. So it is highly reliable and safeIn this paper, we discussed the advantages of identificationbased on biometrics comparing with those based on thetraditional ways. The structure, and it is uniqueness and stabilityof human irises are described in detail. less invasive promisingmeans for the to acquire iris images.Firstly,basing on thecharacteristic of the iris images,we put the hidden markov modelin to the system of the iris image Recognition.Know to all, hidden markov model got the extensive applicationin the speech Recognition, not a few scholars have already alsotry to apply hidden markov model method to identify in thecharacter Recognition recent years,especially in person's faceand the bank paper Recognition etc.But is applied it to recognitein the iris little and little.This text not only explained thatpossibility of this method is valid but also prove that thisviewpoint is to go up and theoretically of valid onexperiments .Withdrew the stage to use the Singular Value ofthe iris image matrix in the characteristic conduct and actionsobservation vector.Adopt a value of grey to be directly before theobservation vector contain two very big limitations:First a valueof grey does not mean the steady characteristic,it is verysensitive in the noisy of image,changing of light and circum ofimage;The next in order, the observation vector of the big sizecauses the calculation complications increase, as a result timethat increased to carry on training and Recognition to thesystem.Usage Singular Value conduct and actions observationthe vector overcame these limitations.Train the stage in themodel moreover according to prognosticate parameter of beworth the sequence more heavy estimate the method, more andgoodly resolved the hidden markov modeltraining problem ,improved the training quality of the model, thus raised thesystem's recognization function.