Filter-based Bayer Pattern CFA Demosaicking

In this paper, we introduce an efficient demosaicking method based on an advanced nonlocal mean (NLM) filter using adaptive weight with consideration of both neighborhood similarity and patch distance. The to-be-interpolated missing color information is considered to be populated by the weight average of directional estimation in six directions: north, south, west, east, pro-diagonal, and anti-diagonal. The NLM sets adaptive weight by obtaining neighborhood similarity between a missing color pixel and the given neighbor color pixels, and similar local neighborhoods are selected to calculate the adaptive weight. The proposed advanced NLM filter is designed to assign weight as a combination of NLM and spatial location distance. The nearer a pixel to a missing pixel, the larger is its chance of having a local structure similar to that of the to-be-interpolated pixel. Considering this evidence and from the viewpoint of reducing computational complexity, we employ location distance to control weights. The new advanced NLM filter is better parametrized, and the weights therein are impacted both by patch similarity and location distance. We present experimental results in terms of CPSNR, S-CIELAB E*, and FSIM to clarify the performance of the proposed algorithm objectively and subjectively. The experimental results indicate that the proposed method outperforms existing approaches both objectively and subjectively.