LAUSR

In this paper, we introduce a weighted sum-based color filter array interpolation method using Taylor series cubic approximation. We use a high-order approximation to predict accurate pixel values during the reconstruction of quincunx grid-sampled green channels, and perform a weighted average-based interpolation in a large local window. We also perform prediction utilizing a color difference model in a small local window to generate additional green values. By applying the weighted sum method to the predicted green values in two local windows, green channel can be enhanced. The remaining color components, namely, the rectangular grid-sampled red and blue channels, are interpolated utilizing the weighted average method of the color difference model. Additionally, we propose a post-processing method that removes the zipper artifacts generated during the interpolation process. Experimental results demonstrate that the proposed color filter array interpolation system outperforms existing algorithms in terms of both objective and subjective performance.