| Today's digital video technology is developing rapidly. People have produced a lot of new revolutionary technologies. Digital video technology has been applied to the video store, home TV, distance learning, video conferencing, telemedicine, video telephony, television, video products, games and entertainment and other fields, especially been widely used in film special effects production and digital video data compression, etc. The digital representation of the real video needs a large number of bits. The transmission medium bandwidth and storage capacity has been improving, but for most of the transmission and storage, the data of digital video signals is very large, and therefore, data compression in the field of digital video applications has indispensably important role.In this paper, the study is video motion estimation and compensation techniques that play a key role in the field of data compression. As information technology continues to evolve, the capacity of transmission and storage has been growing. However, a simple and effective video compression system in the network will improve the quality of video image greatly. Video compression technology enables users to get high-quality video or image quickly, it solves contradiction between the user's high demand and limited network bandwidth. Digital video motion estimation algorithm is good or bad mainly by the search speed and image quality to measure, usually motion estimation algorithms include search strategy and scope, as well as making comparisons between blocks will be used to matching criteria, these are the three key parts of the composed for improving the overall efficiency of algorithm. The purpose of motion estimation is to determine motion vector of the corresponding pixel in inter-frame, and be able to use the video image content of the corresponding positions in the follow-up steps, and to achieve the purposes of image quality improvement. In the motion estimation stage, the accurate of the sub-pixel motion estimation is a key factor in success of the compression system.In addition, the video motion estimation technology can also be used for the dynamic monitoring in industrial processes , military target tracking, video data compression in commercial areas and many other areas, it has broad application prospects.The basic idea of the Block-Matching Motion Estimation is that moving images of each frame is divided into many mutually non-overlapping blocks, and assume that all the pixels in each block have the same amount of movement displacement, and we need to calculate separately for each macro-block. In the calculation of the optimal matching block of the current block in the current frame, at first we search the block of reference frame in a certain search area, according to certain matching criteria, to find the best matching block that can make the best of the matching criteria, this block is the optimal matching block, and calculate the matching block and the motion vector and pixel difference of current block, and the difference of the pixel determines the quality of estimation algorithm. In the H.261 and MPEG-4 and other international standards, motion estimation has been used for digital video compression. However, these standards do not specify a specific motion estimation algorithm, so look for a good algorithm which is simple and effective, good for real-time become an object of study, so in this case block-based motion estimation algorithm become the natural choice, The motion estimation algorithm which based on block developed rapidly, resulting in six classical algorithms that are a full-search algorithm, three-step search algorithm, diamond search algorithm, cross-search algorithm, the new three-step search algorithm, simple search algorithm, and four-step search algorithm. In this paper, the above six kinds of algorithm is done in detail.The innovation lies in the text proposes a new adaptive block matching based on slanting rood algorithm, the main idea of this method: This method uses two size of the slanting rood search model, mainly including an initial self-adaptive search and fixed pattern search. Comprehensive consideration in video sequences in the time domain and space domain correlation in the initial adaptive search using a large slanting rood patterns and adaptive step search strategy of combining point, greatly reduce the possibility that the algorithm get into a local minimum and the num of check points . For each of the current frame in a block, the initial self-adaptive search is only implement at the beginning of algorithm, and it can get a good search direction for the fixed pattern, and then in a fixed pattern search, a small slanting rood model which is fixed-size will be a repeatedly executed until you find the best matching block of the current block and get the motion vector, the algorithm is end. A large number of video images in experiments proved the simple effectiveness of this algorithm. In this experiment, I used the six classical algorithms for the foreman and the mother-daughter video sequence, and calculated the predicted motion vector and motion compensation image, as well as the PSNR value. In the actual analysis for either vision effect or the PSNR indicate that the use of this motion estimation algorithm achieved very good result, further testify the simple effectiveness of the ASRPS algorithm.In the end of this article, I summarized the whole paper and prospected for the future, pointing out the lack of the video motion estimation technology and the application in the future. In the motion estimation algorithm, the choice of the search strategy plays a key role in the quality. The well-designed search strategies can accurately and quickly find the best matching block position, so find a more realistic application of the search strategy is an interesting research direction. Although now in motion estimation algorithms, block matching motion estimation algorithm has been a research hotspot, but we can not just be limited to the classic block-matching algorithm. We can introduce multiple reference frames, and variable block sizes and other new technologies into the block-matching motion estimation algorithm. Make the best use for these technologies in an algorithm, it would greatly enhance the algorithm's general applicability, but at the same time it also increased the difficulty of achieving. |
PDF Full Text Request