Detection and Tracking of Multiple Faces in Video using Modified KLT Algorithm

In proposed research, A system for accurate face detection and continual monitoring that makes use of face key characteristics as essential elements of the face that are incorporated in modified KLT approach. A greater degree of data is provided by video frames than by a single image. Critical attribute extractors (points) are used to automate the detection and monitoring of faces in a wide range of applications, such as human behavior identification, automated surveillance and human computer interface. By dividing the monitoring approach into three sections, one can construct a straightforward facial detection and continual monitoring system. Face detection, facial feature identification and tracking of faces are the first three steps. The Viola Jones object detector is used to achieve the face detection and modified KLT method is used to accomplish the continual monitoring. According to the findings of the investigation, our method is more reliable and effective at evaluating video frames to account for different types of occlusions and pose variations. Visual Tracker Benchmark database is used as a standard database. Further results are compared with staff database which was created to verify the outcomes. Average accuracy of 95% is obtained for the proposed method.