Soft Computing Based Iris Recognition Using Contourlet

摘要

Identification of individual is becoming a serious issue now days. Biometric systems play a vital role in auto identification of individual based on the unique feature or characteristic possessed by the individuals. Iris is unique for every person so it can be consider as a most reliable and accurate system of identification. The serious issue in iris recognition process is the choosing of transform for feature extraction and the selection of subset in the feature space for classification. In this paper the method of contourlet transform for feature extraction and gray level cooccurrence matrix (GLCM) technique were used to select feature vector subset for classification. Contourlet transform have the excellent property of capturing the intrinsic geometrical structure of the image which is used for feature extraction of the iris image. The multidirectional filter bank decomposes the iris image into different subband levels and provides best resolution of iris in various orientations along with the reduced feature vector dimension due to the characteristic of multiresolution property. The feature vector subset selections were done efficiently using GLCM technique. Texture classification was also done efficiently with the subset vector of feature obtained while in GLCM value. To obtain the accurate classification and result in less processing time with increase performance Support Vector Machine (SVM) is used. Support Vector Machine (SVM) classifier is used among various soft computing technique for classification of iris.SVM provide unique and global solution by whistling Artificial Neural Network (ANN) without depending on the dimensionality of the input. The proposed innovative method is efficient as well as reliable for personal identification which yields less computational running time both in the training and recognition phase. This type of recognition can be used for authentication purpose and security related areas such in airport, banking, and secret missions. Iris Recognition is more efficient than using username and password technique and prevents the malicious action by the intruders. The above recognition experiment can be simulated using MATLAB, and LABVIEW.