Iris Recognition Based On Wavelet Transform

Iris recognition, a kind of personal identification method based on biometrics, is very important in information and security fields. Iris recognition technique is regarded as one kind of the best promising techniques because of iris's exclusive, steady, un-impinged characteristics.This paper expatiated on the advantage of using iris recognition in personal identification field, summarized the key techniques of iris recognition system, mainly researched on the feature extraction and match process in iris recognition based on the zero-crossing of the wavelet transform first proposed by Boles. This method used the positions of zero-crossing of the wavelet transform and the average integral between the two adjoining zero-crossing as the features to encode and match. Doing some research and experiments on the details in the above method to improve it.The effect of db3 wavelet is preferably after experiments when using all kings of wavelet function. Attempted using the threshold value to wipe off the jerkwater zero-crossing, sampling respectively using the row of the image and the column of the image, using different decompose levels of wavelet transform to feature match, using the correlative information of different rows with accumulating the adjoining rows and using different distance functions to feature match. The problem of variable-length code match and the infection of integral information on recognition effect has been embedded researched. Finally, an iris recognition algorithm based on zero-crossing position of wavelet transform and matching using variable-length code is proposed, After the zero-crossing positions of the wavelet transform at each level from the 4th decompose level to the 8th decompose level are calculated, a sort of shift-matching algorithm is used to directly calculate the distance of the variable-length code. The experimental results show that the capability of the method in this paper proposed is exceed than the traditional method.