The Iris Recognition Based on Curvelet Transform and Improved SVM

In order to increase the accurate rates of the iris recognition, this paper proposes iris recognition method based on the combination of second generation curvelet transform and the Support Vector Machine Category correction. The images collected from iris image acquisition system are identified. First, rotation correction of iris and spot removal are carried out for collecting iris images. Then, iris images are through rectangular conversion, filter to extract edge point element from which judge the iris image quality. Furthermore, LoG operator is adopted to extract high-frequency energy on both sides of the iris pupil local area, and special judgment is taken out for pupil blocking by eyelids, eyelashes on basis of the energy spectrum. Finally, the cuckoo search algorithm is designed to optimize iris classifier of SVM parameters, and with the second generation curvelet transform algorithm, to complete iris recognition Results show that iris recognition with the CASIA 1.0 database, which contains 756 images of 108 eyes, is an efficient method for iris recognition with high recognition accuracy of iris image.