Study Of Adaptive Motion Estimation In H.264/AVC

With the development of multimedia and network communication technology,the range of applications of video communications becomes more widely. The next-generation Video Coding Standard H.264/AVC is proposed for video coding and transmission.H.264/AVC adopts much new technology and has some bright spots,it not only has high coding efficiency, but also to meet the requirements of low coding rate and high quality of image.These new technologies and bright spots make H.264/AVC have superior coding performance and higher coding efficiency than any previous video coding standard.However, its high coding efficiency and high-precision motion vector estimation is achieved by increasing complexity of the algorithm.Motion estimation is an very important part of the H.264/AVC,because it occupies 80% more time in H.264 video coding.For above reasons, we should optimize motion estimation algorithm and reduce the complexity of the algorithm to attain our goal of enabling motion estimation even more efficient,robust,and faster.This thesis firstly introduces the development status,H.264/AVC's characteristics,relative technology and performance analysis of motion estimation.Secondly,it introduces thinking and fundamentals of motion estimation.Thirdly,this paper provides introductions of some motion estimation algorithms.Finally, the author adds self-adaptive motion estimation in JM10.1.The most suitable motion estimation algorithm is adaptively chosen for different video contents after analyzing it,which is the main idea of this thesis.The author does the following work according the optimization theory of the eighth reference, the original algorithm UMHexagons in JM10.1 platform has been optimized.The optimized algorithm adopts dynamic searching range selection and improves the large hexagon searching pattern and small diamond searching pattern. The optimized UMHexagons algorithm is called IUMHexagons. The author carries out the program of the IUMHexagons and does many tests of it. From the results of the tests we see that the move violent and complex video sequence can get good results if we use IUMHexagons algorithm.However, the results is not good if the video sequence is slow motion and simple.In order to achieve the goal that the most suitable motion estimation algorithm can be chosen for a video sequence adaptatively.The author advances a new theory that adds adapation in part of motion estimation.The ultimate results of tests indicate that optimized algorithm not only saves much motion estimation time,but also reduces the complexity of the algorithm. Furthermore,the motion estimation has self-adaptive performance which improves the coding efficiency and robustness performance greatly.