Face Detection Based on Depth Information Using HOG-LBP
Tiemeng Li, Wenjun Hou, Fei Lyu, Yu Lei, Xiao Chen · 2016
Face detection is one of the key technology for face information recognition. Currently, face detection is mainly based on the RGB image, which might lead unexpected results when dealing with non-real 2-dimensional face, which printed on clothes and papers. This work proposes a face detection approach based on features of depth image. We used depth camera to obtain the raw depth data, and then mapped it to the 2D image combined with smooth image processing method to get the depth image. An improved HOG-LBP algorithm was designed to profiling the features of face depth, and SVM-light was used for machine learning. Finally, an online video face detection system was accomplished.