Fast algorithm of arbitrary factor subpixel downsampling based on frequency analysis

Ketan Tang,Oscar C. Au,Lu Fang,Jiahao Pang,Yuanfang Guo,Jiali Li
DOI: https://doi.org/10.1109/ICME.2014.6890186
2014-01-01
Abstract:Subpixel-based downsampling has shown its advantages over pixel-based downsampling in terms of preserving more spatial details along edges and generating sharper images, at the cost of certain amount of color-fringing artifacts in the downsampled image. To balance the sharpness and color-fringing artifacts, some algorithms are proposed to design optimal anti-aliasing (AA) filters, which are either image independent, or computationally too expensive. And all of the existing AA filters are designed for fixed downsampling factor, which makes them impractical for real applications. In this paper we propose two fast algorithms to design AA filter for arbitrary factor subpixel downsampling based on frequency analysis of the input image. The proposed algorithms generate image dependent AA filter which is as good as the state-of-the-art algorithm, but much faster.
What problem does this paper attempt to address?