| Motion estimation is a key technology in video image stabilization and videocompression encoding, and how to achieve it faster and better is a hot issue of the imageprocessing technology. Many motion estimation algorithms have been proposed, theedge information is important in image, and motion estimation based on the edgeinformation is being more and more attention. The main content of this thesis isdescribed as follows.The principles of motion estimation、 the problems in it and some kinds ofcommonly used motion estimation algorithms are introduced in this thesis, Blockmatching algorithm is analyzed in details. An improved four-step search algorithm isproposed based on four-step search algorithm, and the simulation results show that theproposed algorithm can effectively reduce the search points and maintain a highefficiency.Some concepts of edge and edge detection and several classical edge detectionoperators are introduced. Theory analysis and experiments are carried out to comparetheir advantages and disadvantages. The Canny algorithm is analyzed in details. Animproved Canny edge detection algorithm based on gradient amplitude and direction isproposed. The simulation images show that the proposed algorithm improves theaccuracy of edge detection and reduces the false edge. |
PDF Full Text Request