Research On Frame Rate Up-Conversion And Its Application On Video Coding

Frame Rate Up-Conversion(FRUC)is a hot issue in video processing technology.The technology can increase the frame rate of the video sequence from low to high,which can increase the fluency of video and improve the visual experience.However,although the quality of the interpolated frames can be ensured by the traditional algorithms,these methods often consume a large amount of calculation time and can't target both the image quality and the time complexity.The FRUC technology also has similarities with the reconstruction of the missing frame in video coding,and there is a lack of research and discussion between the two technology.For this reason,this paper proposes two kinds of FRUC algorithms,and explores the application of FRUC technology in the field of video coding:(1)A spatio-temporal based multi-resolution FRUC algorithm is proposed.The algorithm utilizes the image pyramid to construct hierarchical structure.Then,a motion estimation algorithm is performed on the low-resolution image based on spatio-temporal correlation at the purpose of reducing the time complexity.In order to correct the low-precision motion vectors,which exist in the initial motion vectors field,a motion vector correction technique is introduced to detect and correct the outliers after the motion vectors field is transferred to the high-resolution image layer.In the experiments,this algorithm has been proved with the advantages of less time complexity and accuracy.(2)A spatial prediction based FRUC algorithm is proposed.A block classification is used to divide the interpolated frame into two types of blocks.Then,full search algorithm with successive eliminate algorithm is utilized on the first kind blocks to track the motion vectors at a low time complexity.According to the spatial correlation,which can optimize the computation efficiency,the motion vectors of remaining blocks can be calculated.This method effectively reduces the number of motion estimation algorithm execution times.At the same time,based on spatial prediction,relative real motion vectors can be obtained without using motion estimation algorithm.Through the experimental comparison,the proposed method provides less time in processing single frame at average.(3)An application example of FRUC technology in the field of video coding is provided,namely,Compressive-Sensing Video Encoding(CSVC),which combines FRUC technology with compressive-sensing theory.Compared with traditional video coding schemes,CSVC has the features of low sampling rate at the encoder and fast reconstruction of video frames at the decoder,which can reduce the transmission bit rate and improve decoding efficiency.Through experimental comparison,the proposed encoder has a lower rate-energy-distortion performance,making it more advantageous in a low-energy environment.