The ability to automatically detect and track people is a key component of intelligent surveillance system (ISS). ISS is able to automatically analyze surveillance camera data with limited human intervention. In this research, robust people detection and tracking methods are studied for public surveillance system. The proposed method composed of foreground-background segmentation and people detection using mixture of gaussians (MOG) and histogram of oriented gradients (HOG) and then followed by multi object tracking (MOT) method for people tracking. The proposed method is tested on a closed circuit television (CCTV) system at the engineering College of Sebelas Maret University (UNS) Surakarta. All of the people tracking methods has been tested, and the result is good enough with the success up to 100%.
展开▼
