An LMMSE-based merging approach for subpixel-based downsampling

Subpixel-based downsampling improves resolution at the cost of color fringing artifacts. In this paper, we propose a novel multiple descriptions approach for subpixel-based downsampling. Each description is treated as a noisy observation of the original image and an optimal linear estimator based on the LMMSE criteria is applied in the DCT domain to combine multiple descriptions. Simulation result shows that the proposed method gives sharper resultant images while being free of color fringing artifacts with compared to the resultant images of the conventional downsampling methods.