Contact lens detection and classification in iris images through scale invariant descriptor

We propose a new machine-learning technique for detecting the presence and type of contact lenses in iris images. Following the usual paradigm, we extract the regions of interest for classification, compute a feature vector based on local descriptors, and feed it to a properly trained SVM classifier. Major improvements w.r.t. current state of the art concern the design of a more reliable segmentation procedure and the use of a recently proposed dense scale-invariant image descriptor. Experiments on publicly available datasets show the proposed method to outperform significantly all reference techniques.