Motion Estimation Using Gradient Orientation Structure Tensors

This paper describes a novel technique for motion estimation in image sequences using gradient orientation structure tensors (GOSTs). The technique differs from conventional motion estimation techniques using gradient structure tensors (GSTs) in that GOSTs are based on only gradient orientation information and independent of the gradient magnitude. Since gradient orientation is invariant to global changes of image intensities, the proposed method performs motion estimation robustly regardless of time-varying image intensities often caused by irregular lighting conditions. The performance of the proposed method is compared with that of the conventional GST-based motion estimation technique under ideal, time-varying lighting, and noisy conditions.