Iris detection for person identification using multiclass SVM

Abstract

There are critical applications which need high confidentiality and high security. For this, biometric devices are great tools. In the proposed system, the biometric authentication system using iris detection is presented. In this iris image is preprocessed then Hough transform is applied and finally, the image is normalized using Daugman's rubbersheet model. Feature extraction is done using Gray Level Co-occurrence (GLCM) technique and classification is done using multiclass SVM. This system is evaluated on UPOL database and it gives 94.23% accuracy.