Motion-Compensated Frame Interpolation With Multiframe-Based Occlusion Handling

Abstract

In this paper, an occlusion handling algorithm is presented for motion-compensated frame interpolation (MCFI). The proposed algorithm first estimates the unidirectional motion vector fields (MVFs) between two reference frames in both forward and backward directions. Then, a multiframe-based occlusion detection method is presented to identify the uncovered, covered, and motion ambiguity regions within the intermediate frame plane. Based on the detection results, bidirectional motion vectors of the intermediate frame are assigned accurately from unidirectional MVFs. For bidirectional motion compensation, the conventional overlapped block motion compensation (OMBC) method is improved by adjusting the support weight of each pixel within the overlapped window to suppress the blurring and ghost artifacts. Experimental results show that the proposed method not only provides interpolation results with better subjective visual quality, but can also improve the average peak signal-to-noise ratio of the interpolated frames by up to 1.788 dB, when compared to benchmark algorithms.