Tracking Looming and Receding Objects by Orientation Code Matching

抄録

We use Orientation Code Matching (OCM) based search for finding objects undergoing scale changes as result of looming motion. OCM is robust for object search even in presence of some irregularities like shading, highlighting, occlusion or their combination. We propose the use of robustness property of OCM against occlusion for extending its use to search looming as well as receding objects which appear larger or smaller due to relative motion between camera and the object of interest. A two stage framework is proposed; a scale factor is estimated in the first stage and then the object is located within the image frames by searching the template which is resized by the estimated scale factor. Experimental results with real scenes show the effectiveness of the proposed method for object tracking in image sequences.