Weighted bilateral filtering for near-edge artifacts avoidance and impulse noise suppression
Ba Thai, Guang Yi Deng, Robert Ross, Phat Huynh · 2016
An important problem in image smoothing is to reduce noise while preserving sharp edges. In this paper we extend the bilateral filter for better impulse noise suppression and near edge artifacts avoidance. The proposed method weights each pixel prior to bilateral filtering by an extra coefficient that is dependent on a noisiness measurement. Consequently, the influence of noisy pixels can be significantly reduced in the filtering process. Experimental results have demonstrated that the performance of the proposed filter is competitive with bilateral filter in removing impulse noise while preserving sharp edges. Additionally, the proposed method can well avoid near edge artifacts in comparison with other state-of-the-art edge-aware filters.