An Optimal Cluster Fusion and Morphological Segmentation Technique for IRIS Recognition

The biometric identification and also eyeball analysis may require designing an efficient Iris Recognition (IR) technique. Existing IR methods based on the FCM and clustering needs an improvement in segmented area determination. This paper has proposed wavelet based cluster Fusion approach to improve the performance of K-mean clustering. This may control issues of multi cluster representation of objects. The thresholding based segmentation is applied on fused cluster image. In order to eliminate the unwanted boundaries, the morphological processing is performed. Finally, circle fitting is applied to track an Iris region. Performance of existing and proposed Iris segmentation is compared. It is observed that the proposed Iris detection is less complex and efficiently works for all kind of images.