Object tracking in deformed patterns

The geometrical modeling and tracking of visual targets is a key subject in several advanced technologies. The precision of the visual tracking is still an open area for research due to the presence of several serious problems. This paper presents a successful technique to benefit from the camera as a real ranging sensor by which the serious errors due to deformations will be eliminated. The visual object tracking is an important subject for precise identification and pursuing of targets (human and objects) in deformed video stream. A three dimensional (3D) geometrical model has been developed to determine the current pose of an object and predict its future location based on FIR model learned by the OLS while the visual extraction relies on the YC_bCr model. We propose a robust ranging technique to track a visual target to replace the traditional expensive ranging sensors. The presented research work is applied to real video stream and achieved high precession results.