Exploitation of Regression Line Potentiality to Track the Object through Color Optical Flow

Abstract

Normally gray images are less potential with the optical flow, especially to emulate relevant information. Since very long time the color optical flow strategy had been ignored. In this work, we are proposing a strategy which attempts to separate out the optical flow for each channels such as R, G and B through Horn-Schunk with Barren, Fleet and Beuchemin (BFB) kernel. Subsequently, obtained upshots are overlaid to get the rich information of the motion to detach the moving objects. Consequently the histograms of the moving objectss are employed to create the regression lines and extract product-moment correlation of each moving object. This coefficient utilized to match between the template and the candidate templates. Hence the template is updated with the best match. Further the bounding box is enclosed over the object based on spatial information of updated template.