Global motion–compensated preprocessing algorithm for block-based frame-rate conversion

摘要

We propose a global motion (GM) compensated prepro-ncessing algorithm for block-based frame rate conversion (FRC). Thenproposed method estimates camera motions such as zooms or ro-ntations between two input frames, and accordingly, produces non-ntranslational GM-free pictures by performing GM compensation withnrespect to the temporal location where the FRC method recon-nstructs an intermediate frame. To reduce the computational load, thenproposed method first subsamples input images, and then block-nwise motion estimation (ME) is performed. With the ME results, thenproposed method detects scene changes and nontranslationalnGMs. This allows us to determine whether to proceed with the GMnestimation and compensation processes. The geometric motionnmodel is adopted to describe camera motions with four parameters,nand these values are iteratively found on a motion vector (MV) field.nExperimental results show that the proposed algorithm achieves sig-nnificant performance improvements of subsequent FRC methods.