TRACKING MANUALLY SELECTED OBJECT IN VIDEOS USING COLOR HISTOGRAM MATCHING
S. Bhuvaneswari, T. S. Subashini · 2014
Tracking a moving object over time is a challenging task. In this paper object to be tracked is manually selected by the user in one video frame and it is tracked in all subsequent frames of the given input video sequence. The work is carried out in two steps. First the object is detected using 64bin colour histogram matching and the object positions in all the video frames is determined to obtain a temporal sequence of coordinates. Next the object is tracked by calculating Euclidean distance between tracked objects and manually selected reference object. The proposed method is compared with KLT tracker which is a kernel based tracking method and it is found that 64 bin colour histogram matching method can be used to track objects effectively. The results show that the colour histogram features are very efficient in tracking objects, in small and medium length video sequences.