A motion-compensated frame interpolation method using symmetric motion estimation and variance-distortion curve

Abstract

This paper proposes a new motion-compensated frame interpolation (MCFI) method. The proposed method utilizes a symmetric motion estimation method, which is a new pixel-wise motion estimation method for MCFI. The proposed MCFI method estimates initial motion vectors (MVs) between successive frames by using the sum of absolute difference and image segmentation technique. Then, it regularizes the MVs by using a variance-distortion curve, which is a new metric for MV regularization. After refining MVs by using symmetric motion estimation method, the proposed method generates an intermediate frame between the successive frames. For occlusion areas, an adaptive frame interpolation is performed, which uses either the previous or current frames. The experimental results show that the proposed method outperforms other MCFI methods in terms of generating objectively and subjectively better interpolated frames.