Combined denoising and demosaicing of CFA images

Color Filter Arrays (CFA) used by single sensor cameras captures single color information at each pixel location. The process of estimating the missing color samples to reconstruct a full color image is called color filter array interpolation or demosaicing. Demosaicing the CFA images without denoising leads to demosaicing artifacts that will reduce the perceptual quality of the image. This paper presents a denoising before demosaicing strategy for denoising of CFA images captured by single sensor cameras. The images captured by the camera using the Bayer pattern are first denoised by Principal Component Analysis and followed by multiscale gradient based demosaicing to preserve the edges and details. The demosaicing strategy is followed by a false color suppression method to remove the residual demosaicing artifacts.