Research On Motion Estimation Method And Study On Correlation Of Image Sequences

To reduce time redundancy between two frames makes video compression and coding technique as possible. Being an effective method on reducing time redundancy, motion estimation technique is widely used in many video compression and coding standards and it is a hot pot in video compression field for a long time. The more characteristics that we get from image sequences and are used in search method, the higher search efficiency and precision can be got. These two subject matter are discussed and studied in this article. The necessary & possibility and development of video coding technique are discussed at first. Video compression &coding international standards and the function of motion estimation in video compression are introduced. Because motion estimation is a "ill-problem", supposed models are needed to add on motion fields. Parameter model and non-parameter model are the two kinds of supposed models. The "ill-problem"is introduced and the hot studied methods in two supposed models are discussed. Block matching method is widely used in video compression &coding international standards. Search efficiency and precision is different under different matching functions and search algorithms. After some useful matching functions and their calculating complication & matching performance are discussed, the search algorithms and search performance are studied. The study of correlation between standard image sequences is the emphasis of the article. Based on the study, Full Search in small range and Three Step Search being combined together as a new motion estimation method is put forward .The results are as follows: under PSNR measure, this method can reduce search step and improve search precision at the same time. In one words, this method has some use in motion estimation.