Simple features generation method for SVM based iris classification

Iris pattern is the region on human eye that generally used for identifying person. The pattern is unique for each person and must be transformed into a representation that gives meaning to the textures. However, this process could be hampered if the given image has poor contrast of intensity level. This paper suggests an approach to enhance the image in order to obtain abundant iris texture. First, using common method of segmentation, the iris region is localized and transformed to rectangular form. Then, we apply the moving average on the image to reduce random noise. At this stage, an amendment will be imposed to produce uniform gray levels distribution. After that, histogram equalization method will be applied to produce equalized contrast and more embellish iris pattern. Finally, this enhanced image is used to produce one dimensional real value as iris signature. Support Vector Machines (SVM) is used to classify the iris images and the results are promising.