Algorithm Of Frame Rate Up-conversion

Frame rate up-conversion (FRUC), proposed in 1980s, aims at getting videos with higher temporal resolution. In FRUC, a series of video frames are produced and then inserted into the original video sequence. Rich original data and high correlation among neighboring frames in sequence images and videos makes it possible to implement FRUC. With the development of digital television (DTV), FRUC has been widely applied in all kinds of fields.Implementation of FRUC algorithm mainly includes two parts, motion estimation (ME) and motion compensation (MC). Motion estimation aims at getting the motion vector (MV) of each pixel in the current frame using motion information among neighboring frames. Motion vector filed is composed of all motion vectors. According to the motion information in the current frame, motion compensation provides an intermediate frame using estimated motion vectors. Accuracy of motion vectors and reasonability of motion compensation interpolation is critical to FRUC algorithm.This thesis proposes a novel frame rate up-conversion algorithm based on improved bilateral motion estimation. Edge-weighted motion estimation is firstly applied to get bilateral motion vectors with high precision and then multi-stage motion vectors refinement is adopted to get more credible motion vectors. Subsequently, we apply the weighted index-based bilateral motion compensated interpolation to produce an initial frame. Finally, a denoising filter is applied on the initial frame to correct unreliable pixels and smooth the image according to the high correlation among pixels in temporal and spatial domain.Experimental results show that the proposed algorithm effectively solves the problem of overlap and hole, improves the accuracy of motion vector and suppresses the interpolation noises and motion compensation errors. Compared to the conventional algorithm, the proposed algorithm provides better subjective and objective quality.