Lagrange Based Color Image Demosaicing Using Iterative Residual Interpolation

Interpolation technique in color image demosaicing is a big challenge to obtain RGB channels as perceived by humans. The existing demosaicing methods adopt i terative residual interpolation (IRI) technique, which performs iterations on the residuals. These methods are unable to reconstruct G channel efficiently with less iterations. Hence a new demosaicing method is proposed, which performs less iterative interpolations on the residuals using Residual Interpolation (RI) and Minimum Laplacian Residual Interpolation (MLRI) methods. Initially, Lagrange interpolation technique is used to estimate RGB channels independently. The G channel is reconstructed based on the estimated values of RGB channels with minimal iterations. R and B channels are reconstructed from RI method using reconstructed G as a guide. Experimental results demonstrate that, this algorithm performance is better in most cases, compared with the existing demosaicing methods both on subjective and objective evaluations. Keywords— Color Filter Array, Guided Filter, Image Demosaicing, Residual Interpolation.