00000ctm a22000003a 4500 UP-99796217608810391 Buklod 20090505093049.0 m |o d | ta 090505s xx d r |||| u| (iLib)UPD-00084796980 DENGII eng LG 993.5 2009 C65 C37 Carampel, Alethea C. TRACE tracking and pose recognition of adults in health care environments Alethea C. Carampel, Joanna Heidi R. Castillo, Maria Franchesca T. Pituk. 2009. viii, 43 leaves ill. With user manual (bound with the main text). Thesis (B.S. Computer Science)--University of the Philippines, Diliman. Automated video surveillance is one of the most important breakthroughs in the field of computer vision. Being able to monitor a person real time is an additional high-light especially in health care environments. A common problem in these intitutions involves the inefficient tracking for manual supervision which may sometimes be problematic due to human errors and inaccuracy. In this paper, an automated surveillance system in health care environment is proposed. It will be used to replaces the one-to-one human intervention and keep track of the activities of all the patients present in the view. It will also enable an alarm every time a real fall or suspicious incident happens to a particular patient. Video analysis requires high quality and highly efficient algorithms for tracking as there is no time to do the costly object recognition each time a new image is captured. This system is divided into three main modules. The first module is Human Detection Module where the input video stream is processed for background subtraction using frame differencing, noise reduction and blob detection. The second module is Human Pose Recognition which analyzes the poses recognized and classifies it according to which pose it belongs. It is further reduced into three stages namely: feature extraction using shape context, machine learning phase using Support Vector Machine, and real time classification. The final module is Human Tracking where optical flow algorithm, which is a type of point tracking technique, was used. These algorithms are proven to have low computational cost which fits the application for real-time monitoring. Human mechanics Computer program. Motion perception (Vision) Computer simulation. Computer vision. TRACE. Fall detection. Castillo, Joanna Heidi R. Pituk, Maria Franchesca T. FI UP UPD DENG-II LG 993.5 2009 C65 C37 Thesis