A Graph-Based Approach to Optical Flow Estimation

A vector field which describes the apparent motion between two images of a video sequence is commonly known as optical flow. The accurate estimation of these displacement vectors is crucial for several computer vision problems, including video object segmentation and tracking. In this work we propose a new algorithm for computing a dense optical flow field that tackles the inherent problems of optical flow algorithms, namely the estimation of flow vectors in regions of low texture as well as the precise identification of motion boundaries. We try to overcome these problems by taking advantage of color segmentation and robust optimization via graph-cuts. Experimental results show the good performance and robustness of our method