An Iris Recognition Algorithm Based on Weighted KNN and Weighted Majority Voting
Xiaohua Liu · Journal of Chinese Computer Systems · 2010
This paper studied the performance of iris recognition based on the color channels from different color spaces using the color image iris dataset,which is characterized by the fact that many of the images were captured under real conditions so as to incorporate some kinds of noise purposely.Then it proposes a decision level data fusion method for iris recognition,which combines Weighted K-Nearest Neighbor and Weighted Majority Voting method.For each channel,the system finds the first K nearest neighbors according to the distance,and then set weights for them employing an algorithm namely RSWKNN as the output of the channel.After that,the weights are weighted summed according to the performance of each channel.This paper proposed an algorithm to calculate the weights of the K neighbors,and four methods for the weights of each channel.The experiments show that,this method can improve the performance of iris recognition effectively.