A DPM based object detector using HOG-LBP features

In this paper, we combine Histogram of Oriented Gradients (HOG) and Local Binary Pattern (LBP) in order to detect the cranium and its components; namely, the brain, eyes and mouth. Furthermore, Deformable Part Model (DPM) algorithm is paired with the AdaBoost for training and classification. We use a CT/PET database acquired from the National Biomedical Imaging Archive (NBIA) in order to train and test our solution.