SELECTED SINGLE FACE TRACKING IN TECHNICALLY CHALLENGING DIFFERENT BACKGROUND VIDEO SEQUENCES USING COMBINED FEATURES

Abstract

The commonly identified limitations of video face trackers are, the inability to track human face in different background video sequences with the conditions like occlusion, low quality, abrupt motions and failing to track single face when it contain multiple faces. In this paper, we propose a novel algorithm to track human face in different background video sequences with the conditions listed above. The proposed algorithm describes an improved KLT tracker. We collect Eigen, FAST as well as HOG features and combine them together. The combined features are given to the tracker to track the face. The algorithm being proposed is tested on challenging datasets videos and measured for performance using the standard metrics.