Fast Long−term Motion Estimation for High Definition Video Sequences Based on Spatio−temporal Tubes and Using the Nelder−mead Simplex Algorithm

Multi−frame motion estimation is a new specificity of the video coding standard H.264/MPEG−4 Advanced Video Coding (AVC) to improve compression performances. These reference frames can be short or long−term. However, a long−term motion estimation needs to have a good initial search point to be effective. In this paper, we propose a new method for a fast long−term motion estimation with high definition (HD) sequences. We use an implicit motion model. First we describe the multi−resolution motion estimation based on spatio−temporal tubes. These tubes permits to obtain a good initial search point for the long−term motion estimation that follows. Then, the search is refined using the Nelder−Mead Simplex method. The global approach allows to reduce computational cost and improves the accuracy of motion estimation.