Detection and Classification of Moving Objects-Stereo or Time-of-Flight Images

摘要: This paper describes a system for detection and classification of moving objects based on support vector machines (SVM) using 3D data. Two kinds camera systems are used to provide the with range images: time-of-flight (TOF) stereo vision system. While former uses modulated infrared lighting source information in each pixel photonic mixer device (PMD) sensor, latter employs disparity map from images calculate three dimensional The proposed is classify different dynamic environment under varying conditions. first preprocessed then two approaches applied extract their features. approach computer generated method which principal component analysis (PCA) get most relevant projection data over eigenvectors second human extracts features some heuristic techniques. training sets derived image set PCA train multi class SVM classifier. experimental results show that classifier TOF superior