Video Surveillance
Surveillance system is important in providing security and constant observation of locations to obtain information when needed. Over the years, various techniques in this field has been developed, such as recognition, identification, tracking, counting, authentication and so on. The challenges arose from developing these techniques are related to the quality of the video and issues that might affect how computer views the video, such as low resolution, obtrusion, moving objects and feature identifications. In this field, our research focuses on obtaining higher accuracy in video surveillance to achieve multiple purposes, such as identifying while tracking moving objects without being affected by obtrusion. Additionally, we aim to achieve alternative practical applications of these techniques in other fields, such as human relationship mapping and unmanned store maintenance for robots.
Pedestrian Detection
Occlusions in pedestrian detection represents a great challenge in real-world applications, including urban autonomous driving and surveillance systems. However, the head-shoulder feature of pedestrians, which is more stable and less likely to be occluded than other areas of the body, can be used as a complement to full body prediction to boost pedestrian detection accuracy. In this research, we investigate the unique full body and head-shoulder features which improves pedestrian detection.
Moving Object Detection
To reduce the operation loads and focus on specific purposes, it is possible to only detect moving objects or any objects which are static and begins to move. This method is useful in detecting intruder or any unexpected movements which could provide early notification or warning signals for necessary measures.
Speed Detection
By calculating the change of distance between video frames, automatically calculating the velocity of a moving object with accuracy is challenging. Aside from being a possible alternative for expensive speed camera, our developed system can be realized into portable devices, allowing early signals of any incoming objects to people with impaired vision or to other applications.