An edge-directed demosaicing algorithm based on integrated gradient
King-Hong Chung, Yuk‐Hee Chan · 2010
In this paper, an efficient edge-directed demosaicing algorithm for Bayer images is presented. A new edge-sensing measure called integrated gradient (IG) is exploited to extract gradient information in both color intensity and color difference domains simultaneously for the interpolations. This measure supports full resolution and allows one to interpolate the missing samples along an appropriate direction. By sharing it in different stages, it guarantees consistency of the interpolation direction in different channels and saves the effort required to repeatedly extracting gradient information from intermediate interpolation results. An IG-based green plane enhancement is also proposed to further improve the method's efficiency. Simulation results confirmed that the proposed demosaicing algorithm outperforms most up-to-date demosaicing algorithms in terms of both quality and complexity.