High Empirical Study of IRIS Region Recognition Using RED Algorithm Based Discrete wavelet Transform

摘要

In this paper, a biometric recognition system based on the iris of a human eye using discrete wavelet transform (DWT) and multi support vector machine classifier is proposed. The classifier is used to classify the iris zoom into three categories: cut transition, gradual transition and normal sequences. A compact low pass filter with an exponential weighted moving average in a RED algorithm is extensively used. The algorithm can be used to calculate the average queue size as well as maintain high throughput and low delay. Before extracting the features of an iris, the input image is preprocessed to localize, segment and enhance the region of interest, blurring, and artifacts especially those associated with edges. The block textural information of a person's iris is exploited to categorize the input image into a number of vectors (blocks). The categorizing of these blocks were based on extracting the edge in different direction from wavelet transition coefficients in a given image block. Furthermore, computing of block variance and directional variances of neighboring blocks greatly facilitated the reconstructing of the strong diagonal edges. Best extraction of vertical, horizontal, and diagonal image gradients were computed. The largest numbers of these gradients were obtained after changing both the magnitude of the threshold, which were extracted from histograms of those gradients, and the boundary match vector angles. Experimental results show that the proposed method is quite fast efficient and accurate method employs a minimum distance classifier according to hamming distance.