Translator Disclaimer
1 April 2009 Effective soft-decision demosaicking using directional filtering and embedded artifact refinement
Author Affiliations +
Demosaicking is an interpolation process that transforms a color filter array (CFA) image into a full-color image in a single-sensor imaging pipeline. In all demosaicking techniques, the interpolation of the green components plays a central role in dictating the visual quality of reconstructed images because green light is of maximum sensitivity in the human visual system. Guided by this point, we propose a new soft-decision demosaicking algorithm using directional filtering and embedded artifact refinement. The novelty of this approach is twofold. First, we lift the constraint of the Bayer CFA that results in the absence of diagonal neighboring green color values for directionally recovering diagonal edges. The developed directional interpolation method is fairly robust in dealing with the four edge features, namely, vertical, horizontal, 45-deg diagonal, and 135-deg diagonal. In addition, the proposed embedded refinement scheme provides an efficient way for soft-decision-based algorithms to achieve improved results with fewer computations. We have compared this new approach to six state-of-the-art methods, and it can outstandingly preserve more edge details and handle fine textures well, without requiring a high computational cost.
©(2009) Society of Photo-Optical Instrumentation Engineers (SPIE)
Wen-Tsung Huang, Wen-Jan Chen, and Shen-Chuan Tai "Effective soft-decision demosaicking using directional filtering and embedded artifact refinement," Optical Engineering 48(4), 047003 (1 April 2009).
Published: 1 April 2009


Nonlinear quincunx interpolation filtering
Proceedings of SPIE (September 01 1990)

Back to Top