Motion Estimation Method for Video Sequences at Different Resolutions Based on Block Matching

Abstract

The present invention is a motion estimation method for video sequences under different resolutions based on block matching. For the comparison under a single resolution in the past, the method is based on three classical algorithms of full search, three-step search, new three-step search and traditional matching. The SAD rule is used to compare the motion estimation performance of fast, medium and slow international standard video sequences at two resolutions, that is, the number of search blocks, search time, PSNR, and reconstructed images. An important conclusion of the present invention is that low- and medium-speed video sequences perform better at low resolutions, whereas fast video sequences have better motion estimation effects at high-resolution. The conclusion of this method makes up for the lack of motion estimation at different resolutions, and solves the problem that video sequences with different characteristics are more suitable for search methods at different resolutions. The results can also be applied to target recognition, motion tracking, video compression, etc. aspect.

Description

technical field [0001] The invention relates to the technical field of image processing, and relates to a motion estimation method of video sequences under different resolutions based on block matching. Background technique [0002] With the rapid development of society, multimedia technology has enriched people's life. Video images are developing rapidly in people's lives, for example: network videophone, digital TV and so on. However, the large amount of information contained in video images requires high bandwidth for transmission. In order to solve this problem, "video compression technology" can improve transmission efficiency under limited bandwidth, so it is necessary to reduce video time redundancy. In video compression, block matching motion estimation is a method commonly used to reduce video redundancy. In fact, the calculation amount of motion estimation is more than 50% of the total compression calculation amount, which means that the quality of motion estima...

AI Technical Summary

Problems solved by technology

Although the block matching algorithm has been widely used, the existing motion estimation methods are usually for the purpose of saving search time and improving search accuracy. In addition, the existing motion estimation methods only compare a single resolution, There is a lack of comparison of motion estimation accuracy at different resolutions, therefore, the differences in motion estimation properties exhibited at different resolutions are not recognized

[0003] So far, there is no relevant target: to supplement the experiments and conclusions of the classic algorithms, and to propose a performance comparison of "three different video sequences" at "different resolutions", that is, three international standard video sequences of high, medium and low Literature Report and Practical Application of Motion Estimation Methods in Two Common Resolutions (640×480, 176×144)

Images