Frame Rate Up Conversion Method Based On Motion Compensation

With the development of Ultra High Definition Television (UDTV), people have obtained more delicate, more realistic, and more fluent video visual effect. Ultra high definition not only means the spatial resolution achieve4K*2K,more important is to achieve the higher frame rate, the greater dynamic range and the richer color depth. In practical applications, the production of high frame rate video needs expensive cost of the shooting, and also faces enormous pressure of transmission and storage. So how to obtain high quality videos which meet the users’needs, from the existing low resolution videos is a major issue to be solved. Frame rate up conversion technology (FRUC) as an effective means of video format conversion, has became a research focus in the domain of multimedia information processing.Motion compensation frame interpolation (MCFI) has been the most effective FRUC method, which gets the interpolated frame taking motion information into accounted. However, MCFI also brings many inevitable problems, such as block effect, hole problems, occlusion problems and so on. Naturally, these problems have affected the video quality and users’ visual experience seriously. Based on this, the improved FRUC methods are proposed to solve the above challenges in this paper. Finally, the main contents of this paper are divided into the following points:An improved FRUC scheme based on motion vector refinement is proposed to solve hole problems in this paper, the contributions are including the two aspects:to get the correct motion vectors and to fill the hole using motion refinement method. Firstly, a texture based adaptive motion estimation is used to acquire the good motion vectors, then the multi-directional enlarged motion vector post-processing method is exploited to further guarantee the dependability of motion vectors. After that, the range of the hole regions is reduced greatly. Secondly, the motion vectors of the hole regions are got from the motion information around them, then the hole is filled according to the motion compensation theory. This low-complexity solution effectively solves the annoying visual problems caused by hole problems, and has gained the high quality frame.An improved FRUC scheme based on motion vector angle is proposed to solve occlusion problems in this paper. The method analyzes the cause of occlusion problem, and an adaptive handing method is proposed to solve the different types of occlusion problems. For the exposed regions owing to scene switching, the interpolated frame is got using frame repeat method; For occluded and exposed regions owing to the motion transformation, the motion compensation method combined with the forward and backward motion estimation is to get the correct occluded regions. At the same time, the motion vector angle and the information obtained by forward and backward motion compensation are introduced to judge the exposed and occluded areas, which reduce the computational complexity effectively. In addition, in order to avoid error prediction of motion estimation, luminance information based improved Unsymmetrical Cross Multi Hexagon grid Search (UMHexagonS) is able to estimate the high accuracy motion vectors, which introduces the RGB vector distance color difference formula, so this makes the processing of occlusion problems simpler.In this paper, two kinds of application software system are designed based on the above proposed FRUC methods, which make the theoretical research into the practical productive forces. Finally, a FRUC comparison system that is integrated with some classic methods is used for the different selections to different video sequences by users.The proposed FRUC methods are better than the classical methods based on a large number of experimental tests, not only in subjective visual effect, but also in objective value:Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index Measurement (SSIM). At last, the system implementation is received.