Face Occlusion Detection Using Skin Color Ratio and LBP Features for Intelligent Video Surveillance Systems
Pengfei Ji, Yonghwa Kim, Yong Yang, Yoo-Sung Kim · Annals of Computer Science and Information Systems · 2016
A face occlusion detection scheme which is based on both skin color ratio (SCR) and Local Binary Pattern (LBP) feature, is proposed.The proposed method mainly consists of four steps: foreground extraction, head detection, feature extraction, and occlusion detection.First, foreground is extracted by codebook background subtraction algorithm.Then, the head region is located using HOG head detector.After that, the skin-color ratio and LBP feature are extracted.Finally, SVM is trained based on LBP feature.The recognition result of SVM and the result of skin-color ratio feature are merged by weighted voting strategy, and then occluded faces are classified as three categories: concealed, partially concealed, and visible.Experimental results show that the proposed detection system can achieve desirable results in intelligent video surveillance systems.