Deinterlacing using motion compensated local spectra

Abstract

A new algorithm for deinterlacing of interlaced scanned video sequences is presented. The proposed algorithm is based on local spectra analysis of even and odd image fields. The local spectral analysis is performed based on overlapped block decomposition and motion compensation. The aliasing relationship of matching blocks in the interlaced fields and the sampling lattice information obtained by subpixel registration are utilized to reconstruct deinterlaced frames. Simulation results demonstrate that the resulting frames contain less spatial aliasing and sharper edges than frames generated using commonly used approaches such as median filtering and linear interpolation.