A novel method for robust object tracking with K-means clustering using histogram back-projection technique

This paper presents a novel and fast method for k-means clustering based object tracking for coloured frames, based on histogram back-projection method. The proposed method uses histogram equalization for finding centroid of the object for each frame. As from the information transfer aspect, this study improves the tracking performance using Bhattacharya Coefficient with the method of histogram back-projection including k-means clustering. Histogram back-projection computes the probability of the object of interest and the clustering process classifies high-performance regions. In addition, a mean shift tracking method is used to monitor the object after the histogram back-projection process, which provides better tracking for fast-moving objects. The proposed mean shift algorithm also provides gradient ascent. In addition, this method is invariant to clutter and camera motion; pose changes and faster tracking. Simulated results show that the proposed method gives better tracking results than other conventional methods while having a lower computational demand. Therefore it is highlighted that the proposed method would have a significant contribution in the field of object tracking.