A Novel Weighted Edge-directed and Weighted-color-difference Interpolation for Demosaicking Bayer-pattern Color Images

摘要

Demosaicking is a process of obtaining a full-color image by interpolating missing color components of an image from CFA. This paper introduces a new interpolation algorithm for demosaicking Bayer-pattern color images. The first, step is to estimate interpolation direction taking advantage of all adjacent pixels, and use calculated optimal weighting factors to carry on G channel interpolation. Then according to the weighted color-difference rule, R/B components are interpolated and G components are updated with more known pixels. Lastly, using variance-constrained condition, the iterative interpolation repeats until approaching the optimal results. Based on the experiments of 24 Kodak images, the performance of the proposed one outperforms various traditional algorithms both in numerical and visual aspects. It is shown that the proposed algorithm can lead to an improvement in PSNR of about 20.99%, and in △E~*_(ab) of about 8.53%. Furthermore, this new algorithm has been applied on the images captured from panoramic camera and received excellent results.