Object Tracking Using Background Subtraction and Motion Estimation in MPEG Videos

We present a fast and robust method for moving object tracking directly in the compressed domain using features available in MPEG videos. DCT domain background subtraction in Y plane is used to locate candidate objects in subsequent I-frames after a user has marked an object of interest in the given frame. DCT domain histogram matching using Cb and Cr planes and motion vectors are used to select the target object from the set of candidate objects. The target object position is finally interpolated in the predicted frames to obtain a smooth tracking across GOPs.

• Type: Conference
• Year: 2006
• Where: ACCV
• Authors: Ashwani Aggarwal, Susmit Biswas, Sandeep Singh, Shamik Sural, Arun K. Majumdar
• Keywords: ACCV 2006, Candidate Objects, Computer Vision, Object Tracking, Target Object