Vision based moving object tracking through enhanced color image segmentation using Haar classifiers

In this paper we implement a vision based moving Object Tracking system with Wireless Surveillance Camera which uses a color image segmentation and color histogram with background subtraction for tracking any objects in non-ideal environment. The implementation of the moving video objects can be based on any one of the tracking algorithms such as Template matching, Continuously Adaptive Mean Shift (CAMSHIFT), SIFT, Mean Shift, SIFT, Cross correlation algorithm is presented by optimizing the kernel variants by adjusting the HSV value for various environmental conditions. The object occlusions are also removed by calculating the minimal distance between the two objects using Bhattacharya coefficients and it is robust to changes in shape with complete occlusion. The object to be tracked can also be classified using HAAR classifier through machine learning. A software approach for real time implementation of moving object tracking is done through MATLAB.