Digital zooming for color filter array-based image sensors

In this paper, zooming methods which operate directly on color filter array (CFA) data are proposed, analyzed, and evaluated. Under the proposed framework enlarged spatial resolution images are generated directly from the CFA-based image sensors. The reduced computational complexity of the proposed schemes makes them ideal for real-time surveillance systems, industrial strength computer vision solutions, and mobile sensor-based visual systems. Simulation studies reported here indicate that the new methods (i) produce excellent results, in terms of both objective and subjective evaluation metrics, and (ii) outperform conventional zooming schemes operating in the RGB domain.