Study On The Key Technologies And Implementation Of Frame Rate Up Conversion

Frame Rate up Conversion(FRUC) technology not only can be used to eliminate the inherent ghosting and tailing phenomenon of liquid crystal display, but also has great significance in the field of video format conversion, low bit video encoding and 3D technology, which has made high speed of development and application in recent years. Simple FRUC algorithm such as frame repetition, frame averaging and so on, just adopts the linear combination of previous and next frames without considering the scene motion, so its interpolating result is poor. While FRUC algorithm based on Motion Estimation(ME) and Motion Compensation(MC) fully considers the motion track of image pixels, and makes the appropriate motion compensation, which greatly improves quality of interpolated images and has been widely used.The purpose of this dissertation is to improve the result of interpolated frames and enhance the image quality of frequency-doubling video sequence by researching on motion estimation, motion vector correction and motion compensation algorithms, which will provide theoretical and technical support for the application of FRUC in the field of video processing. This dissertation aims at high quality interpolated frames and applicability of the algorithm in order to be easily achieved.The dissertation firstly introduces the theoretical basis and several classical algorithms of FRUC based on ME/MC, and then focus on technologies of motion estimation, motion vector correction and motion compensation on this basis, including their applications in FRUC and the implementation of FRUC system. The main research results are summarized as follows:① To improve the reliability of estimated motion vectors, a double matching criterion based and hierarchical motion estimation algorithm is proposed. Bidirectional ME is selected to avoid the problem of overlaps and holes caused by unidirectional ME. The double matching criterion based on SAD and matching confidence criteria can improve accuracy of the block matching algorithm on the premise of low computation. Besides hierarchical ME is proposed in the case of inconsistent object motions within the same block to obtain more precise block motion vector.② To make sure the quality of interpolated frames, we propose a motion vector correction method based on three-dimensional 1-norm Vector Median Filtering(VMF) and a motion compensation algorithm combined with high/low associated region division. Three-dimensional 1-norm VMF algorithm can remove the particular motion vectors in a very efficient way and avoid over-smoothing phenomenon. The proposed MC algorithm makes average compensation for high-related region and makes overlapped block motion compensation for low-related region to reduce the blocking effects.③ The whole FRUC system is implemented in software platform according to the proposed FRUC algorithm. And furthermore a FRUC contrast software is designed and developed to achieve the function of playing and comparing two video sequences at the same time, which is used to display video processing effect of FRUC algorithms.The dissertation tries to make a breakthrough on research methods, and the above research results are experimented, analyzed and compared, which proves the availability of the proposed algorithm.