Intensity Separation based Iris Recognition Method using Curvelets and PCA

This paper presents a PCA-based iris recognition method called Intensity Separation Curvelet based PCA (ISCPCA). The proposed method uses Canny Edge detection and the Hough transform to extract and rectangularize the iris from the input eye image. The second generation Fast Digital Curvelet Transform (FDCT) is then applied to the resulting image, dividing it into its subbands. The resulting complex subbands coefficients within the same level are concatenated, generating two single frames. The coefficients in each resulting frame are then normalized and evenly divided into a preselected number of bands. The coefficient matrices within each frame are then vectorized and concatenated, generating a single 2D matrix. Conventional PCA is then performed on the resulting 2D matrix extracting its eigenvectors which are used for iris matching. The Euclidean distance is used as a measure to quantify the closeness of different iris images. Experimental results on images from the CASIA-Iris-Interval benchmark eye image dataset show that the proposed ISC-PCA technique significantly outperforms the state of the art PCA based methods, and achieves competitive results to those of the learning based techniques.