Object Tracking by Adaptive Mean Shift With Kernel Based Centroid Method

An approach towards the object tracking, the impactful method object improve localization, is proposed. Object is firstly optimize by the adaptive mean shift, if there is high localization then current frame of video replace by the next frame and adaptive mean shift is use to maximize the Bhattacharya coefficient. If there is low localization by adaptive mean shift then histogram based target representation is regularized by an isotropic kernel and then centroid is calculated by the binary gradient matrix.