지능형 영상감시를 위한 사람의 상체 인식: 성능 평가

초록

The detection of human upper body has attracted much research in the past decade due to the essential role it plays in intelligent video surveillance and vehicle vision systems. The work in this paper applies three well-known methods, including PCA, Viola-Jones framework, and HOG descriptor, for human upper body detection. A comparison between these algorithms in term of accuracy and processing time was carried out using INRIA and PennFudanPed database. This analysis brings out the advantages and disadvantages of each method.