Research On The Methods Of Iris Recognition Based On Gabor Transform

Information security is becoming more and more important in the current society and being different with the traditional identity verification methods such as password, ID etc., the biometric recognition technology has risen in the world. It includes: voice recognition, fingerprint recognition, iris recognition and face recognition etc. Among these technologies, iris recognition is an identity recognition method rising in the recent years and it is also considered as the most promising and safest method. Since the iris excels in its uniqueness, stability, collection and it can't be infringed, the iris based identity recognition has been widely used in the security areas like CIQ etc.This paper, firstly, analyzes the unique advantage of iris to the identity recognition, discusses the current developing state of iris recognition technology domestically, simply introduces the composition of iris recognition system as well as the main functions of its each part and analyzes the iris location method based gray-scale as well as the normalization method.Then, the paper mainly studies the theories of 2D-Gabor filter and analyzes the features of 2D-Gabor filter in detail. The 2D-Gabor filters theory was proposed by J.G.Daugman in the year 1980, the function of 2D-Gabor is a short-time Fourier Transform and it is the only function that can get value limitation under the time-frequency combining uncertainty principle circumstance which means that the 2D-Gabor filter can get the local optimization both in the time and frequency domain. Meanwhile, 2D-Gabor filter has a strong local selectivity, frequency selectivity and direction selectivity. Taking advantages of these features, the iris texture information can be easily acquired. Currently, the 2D-Gabor filter based algorithm has become the main stream technology in iris recognition.After that, this paper proposes an iris noise detection algorithm based on the 2D-Gabor filter. The method takes advantage of 2D-Gabor filter to detect the noises caused by eyelid and eyelash in the normalized iris images, takes notes down the disturbed areas and eliminates the disturbing bite in the iris codes when matching to try its best to prevent the noises from affecting the recognition. Compared with the conventional methods, the proposed method can accurately detect and mark the noise area which helps to improve the stability and accuracy of recognition.At last, the paper proposes two feature extraction methods based on the 2D-Gabor filter. One is the study on the iris feature extraction method based on the single frequency 2D-Gabor filter; the other is an iris recognition method using 2D-Gabor filter to extract the texture direction features. The validity of the proposed methods is verified by the experiments on the current iris database.