Video Frame Rate Up-Conversion Method Research Based On Motion Information

With the development of 3D ultra-high clear television,the demand for high frame rate video gradually increased.Frame rate up-conversion(FRUC)is a kind of video post-processing technology,which can obtain the higher frame rate video and realize conversion among videos with various frame rate.It has attracted the attention of scholars and becomes a hot research topic.Frame rate up-conversion is a technique that uses the time-space relativity of original adjacent frames to generate new video frames and inserts them into the original sequence to increase the frame rate.When the frame rate is higher,the video will be more fluent and the texture of frames will be better.FRUC has a wide range of applications in the digital multimedia field due to the increased need of high frame rate video.At the same time the academic community has also developed a variety of FRUC algorithms,among which the most widely used is motion compensation based FRUC algorithm.This algorithm introduces the motion information to ensure that the interpolation frame is on the original motion trajectory,but at the same time it will bring some unavoidable problems such as block effect,blur,occlusion and so on.Therefore,based on the block matching motion compensation FRUC,this paper studies the motion estimation(ME),motion vector post-processing and bidirectional motion compensation interpolation and proposes effective and feasible solutions.The main contents of this paper can be summarized as the following three points:A frame rate up-conversion method with quadruple motion vector post-processing scheme is proposed.By considering occlusion and hole problems,the paper adopts the hybrid motion estimation method which is a combination of bilateral ME and unilateral ME in both directions.In order to produce better quality-interpolated frames,the motion vector(MV)outliers of the both unilateral MV fields are detected and corrected by prior-information-based MV refining method and the bilateral MV field is smoothed by using vector extrapolation and weighted summation.Moreover,the selection and determination criterion,which is composed of matching ratio of blocks and ladder-style analysis,is proposed to select and decide the denser bilateral MV field.The proposed algorithm can get more accurate motion vector and improve the quality of the interpolated frame.An adaptive motion compensation interpolation algorithm based on space-time domain is proposed.The addition of preprocessing and texture information to the motion estimation not only reduces the computational complexity but also improves the accuracy of the motion vector.Then,the modified MV is analyzed and divided into three kinds of block types.Finally,for different types of blocks,different motion compensation algorithms are adopted based on space-time information.The proposed algorithm can obtain high-quality interpolation frame and effectively reduce the block effects and fuzzy phenomenon.Based on double FRUC algorithm,a novel non-integer times frame rate up-conversion(NIFRUC)algorithm using reliable analysis of time-domain motion information is proposed.The proposed method can realize frame rate up-conversion of any fractional multiples based on double FRUC.In addition,in order to eliminate motion jerkiness and motion blurring,the proposed method makes full use of motion information to select the inter-frame where movement is faster as the interpolated position and gets interpolated frames by bilateral motion compensation interpolation.Also,in order to obtain more accuracy interpolated positions,a new scene detection scheme as well as the combination of the inter-frame motion speed and inter-frame correlation degree is applied.Experimental results confirm that the proposed method outperforms previous NIFRUC methods in terms of theory analysis,subjective mean opinion score and objective evaluation.The proposed FRUC methods have been proved by a large number of simulation experiments,both in subjective visual effects and in objective values:Peak Signal to Noise Ratio(PSNR)and Structural Similarity Index Measurement(SSIM)are superior to the classic FRUC methods.