Iris classification/verification is one of the most promising biometric recognition methods. However, its recognition performance can be degraded by specular reflections, occlusion, eye gaze direction and segmentation error. To alleviate these problems, the area of interest is expanded to include not only an iris but also a whole ocular region including an eyelash and an eyebrow. Since sparse representation classification(SRC) has been proven as a potential technique for face and iris recognition, this project aims to first develop an ocular recognition technique based on SRC, which is robust to illumination and pose changes. In addition, the developed method should be able to handle images with incomplete ocular region when these kinds of images are present in either a training set or a test set. The second objective is to apply sparse representation in ocular verification, and the technique should be also robust in all conditions mentioned above. To achieve the first objective, this project first reviews literature on occlusion handling techniques which work well with other types of biometric data. Then these techniques will be tested with ocular images from BDCP and FOCS data sets and they will be studied in detail -- what kind of occlusions they can handle and why they do or don't work with ocular data. Finally, these techniques will be modified to improve recognition rates. To achieve the second objective, this project explores how sparsity concentration index(SCI) can be used in a verification problem. The SCI based technique will be developed so that it can handle both the condition when the system has seen at least one of the query images and the condition when the system has never seen any of the query images. The anticipated result of this project is to show that the developed techniques can attain higher accuracy in both classification and verification problems.