Identifying Down Syndrome Cases by Combined Use of Face Recognition Methods

Down syndrome is a chromosomal disorder, people affected by this disease having very specific facial characteristics. In this paper we adapt some well-known face recognition methods (Local Binary Patterns, Eigenfaces) and test their ability to distinguish between a Down syndrome face and a normal one in digital images. For classification kNN (k Nearest Neighbor) and Support Vector Machines (SVM) with polynomial kernel of third degree and Radial Basis Function (RBF) are employed. We test the methods using FERET, CAS-PEAL, LFW, AT&T for normal face images and a collection of Down faces gathered from the Internet. Accuracy, precision, recall and specificity are computed in order to evaluate the classification results.