Image Segmentation Based Iris Recognition Using Wavelet Features and Support Vector Machine Classification

Nowadays, biometric recognition is becoming one of the most promising and reliable way to authenticate the identity of a person. In this research, we have presented a robust and fast image segmentation based iris recognition system, which was tested using database of grayscale eye images in order to verify the authorized user of iris recognition technology. The preprocessing schemes appeared to have significant role in the segmentation performance for all the techniques. We described a combination of edge based and clustering based algorithms which means fast segmentation process. We developed a more accurate iris segmentation framework to automatically segment iris image acquired under less constrained imaging environment. We also developed a series of post processing operations to accurately localize limbic boundaries in noisy iris images. Finally features of the iris region were encoded by convolving the normalized iris region with Log-Gabor wavelet and Haar wavelet. The support vector machine was adopted as the basic classifier in order to develop the user model based on iris code data. Based on obtained results, Haar wavelet and SVM classifier seems in a good level of security.

• Supervision