Adaptive format conversion of videos using wavelets

In this paper, we propose and implement adaptive schemes for video format conversions, involving spatial and temporal resolution changes. Individual adaptive schemes for spatial interpolation, deinterlacing and frame rate conversion (FRC) are proposed. The proposed method introduces adaptive schemes for all three video format conversion tasks, such that edges in the spatial and motion in the temporal domain is preserved. Using wavelet techniques, the best angle of interpolation is found so that interpolation is accomplished along the edges. Also, a second fold adaptation for interpolation is provided by frequency selective filters, whereby the interpolation filter is determined by the frequency characteristics of the local data. A scheme for accurate motion estimation for deinterlacing and motion vector estimation for high quality FRC using wavelets are developed. Extensive simulation performed on various images and videos shows the efficiency of the proposed method in comparison to other linear and non-linear methods in terms of objective and subjective measures.