Adaptive Filtering for Color Image Sharpening and Denoising
Takahiko Horiuchi, Kunio Watanabe, Shoji Tominaga · 2007
Both tasks of sharpening and denoising color images are eternity problems for improving image quality. This paper proposes a filtering algorithm which brings a dramatic improvement in sharpening effects and reducing flat-area noises for natural color images. First, we generate a luminance slope map, which classifies a color image into the edge areas and the flat areas with uniform color. Second, the Gaussian derivative filter is applied to sharpening each segmented edge area. In order to keep color balance, the edge-sharpening filter is applied only to the luminance component of the color image. In the flat areas, a filter based on the SUSAN algorithm is applied to reducing the background noises with preserving sharpened edges. The performance of the proposed method is verified for natural image with Gaussian noise and impulse noise.