Block-matching Algorithms Of Two-dimensional Motion Estimation

In recent years, digital video technology has been made great advances in information science domain, combined with TV technology, computer technology and communication technology, and widely used in the TV system, computer application, communication domain. Motion estimation that is important in the digital video technology, is one of the basic problems of digital video processing, involved in image plane motion(two-dimensional motion) or object motion(three-dimensional motion).And it is the main task studied by people. Two-Dimensional motion estimation not only plays a very important role in motion-compensated filtering and video compression, but also is the first step to three-dimensional motion analysis. So research of two-dimensional motion estimation is very significant.In the two-dimensional motion estimation, block-matching algorithms have lesser hardware complexity, and this made it widely used in VLSI. Nearly all coding and decoding circuits of H.261 and MPEG1-2 use block-matching algorithms for motion estimation. So block-matching motion estimation has important value.In this dissertation, the following contents are mainly illustrated in details:1. Essence of two-dimensional motion is studied . The dissertation summarized the basic principle of two-dimensional motion, and introduced kinds of motion algorithms based on the principle.2. The dissertation made deep research in block-matching algorithms. Firstly, deep analyses are made to the aspects of block-matching, such as search pattern, matching rule, and so on. Secondly, The dissertation made research in traditional block-matching algorithms, and analyzed their merits and flaws. Thirdly, according as block motion vector is large or small, we proposed an efficient algorithm and made Experiments compare and validate kinds of algorithms.