Research On Spatial-Temporal Correlation Oriented Video Frame Rate Up-Conversion

In order to meet the needs of multimedia business and video data interaction,Frame Rate up-conversion(FRUC)technology has become a hotspot in the field of video post-processing.By inserting intermediate frames into two adjacent decoding frames,video can be converted from low frame rate to high frame rate.In high-definition display terminal equipment,motion compensation FRUC is used to improve the accuracy of motion estimation and enhance the smoothness of motion vector field,which can effectively avoid the disadvantages of picture jitter and motion blur caused by low frame rate video and bring more comfortable visual experience to the audience.However,with the improvement of digital video resolution,the requirements for FRUC technology are becoming higher and higher.Inevitably,some technical problems need to be solved urgently.Therefore,aiming at the existing problems of abnormal motion vector and the unbalance between computational accuracy and computational complexity,an in-depth study is conducted and the effective solution is proposed.The main work is summarized as follows:(1)Combining edge preserving filtering with fast motion estimation as a strategy to solve the problem of high computational complexity caused by high precision in FRUC technology,a low-complex frame rate up-conversion with edge-preserved filtering algorithm was proposed.The algorithm first conducts edge-preserved filtering on each frame to preserve the edge and suppress the texture details,then each video frame is subsampled to reduce the computational complexity,and predictive search method is used to reduce redundant search points so as to accelerate the process of motion estimation.This algorithm,which keeps the balance between computational accuracy and computational complexity effectively,provides good objective and subjective quality of up-conversion video sequences.(2)Based on the theory of cellular automata,a motion vector field smoothingalgorithm was proposed,aiming at the problem of unreliable vectors in the initial motion vector field or over-smoothing in the existing smoothing methods.A cellular automaton is constructed to deduce MV outliers according to a defined local evolution rule.By executing CA-based evolution in the loop iteration,we gradually expose MV outliers,and reduce incorrect MVs resulting from over-smoothing as many as possible.This algorithm significantly improves the visual quality of the interpolation frames and enhances the ability of detecting and correcting the unreliable motion vectors.