Demoiréing for screen-shot images with multi-channel layer decomposition
Jingyu Yang, Xue Zhang, Changrui Cai, Kun Li · 2017
Moiré patterns on screen-shot images are mainly due to the aliasing of the grid of the display and the camera sensor, which heavily degenerated the image quality. This paper proposes an demoiréing method for screen-shot images via layer decomposition on polyphase components (LDPC). The layer decomposition model separates the image into a background layer and a moiré layer, which are both regularized by a patch-based Gaussian Mixture Model (GMM) prior. To enhance the distinguishability between the image patches and moiré patches, the input image is first subsampled into four polyphase components, each of which is decomposed with the GMM-based layer decomposition model. The proposed model is applied on luminance (Y) channel to weaken the intensity of moiré patterns, and on red, green, blue channels respectively to further remove moiré patterns. Experimental results demonstrate that the proposed method is able to efficiently remove moiré artifacts for screen-shot images and outperform several other methods.