Fast optical flow estimation

This paper presents a fast, accurate and reliable approach for robust boundary preserving estimation of optical flow. Several studies have addressed this topic and proposed methods that account for velocity boundaries at the cost of significant computational complexity, which makes them inadequate for current real-time applications. The proposed method is derived from the benchmark algorithm of Horn & Schnuck and Simoncelli's matched-pair 5 tap filters, such that it produces robust, fast and exact detection of motion boundaries and it is very simple to implement. Experimental results using synthetic and real optical flows are presented to demonstrate the effectiveness of our method in comparison to selected methods.