Moiré Pattern Removal From Textured Images Via Image Decomposition

The term Moiré comes from the French,which usually refers to the interference artifact caused by the overlap of two sine waves with the same amplitude and closely frequencies.Due to the limited sampling rate of digital devices,moiré phenomenon easily appears on the digital imaging process when repetitive structures(such as cloth or LED screens)captured by digital cameras,heavily disturbing the qualitative and quantitative analysis of images.Moiré pattern is an aliasing interference caused by signal under-sampling,and its structure is closely related to the captured screen.Therefore,the research on textured image demoiréing is very challenging.So far,investigations on textured images demoiréing are still rather limited(hardware-based or professional software)and not yet achieve mature algorithms.In order to solve the problem,this paper propose a novel textured image demoiréing method by signal decomposition and guided filtering,which removes moiré patterns from textured images efficiently.The achievements and main creative points are as follows:1.Based on the study on statistical and related characteristics of image texture and moiré,this paper proposes to add the low-rank prior constraint to the texture component and add the sparse prior constraint to the distribution of the moiré component in frequency domain.Utilizing the symmetry of the frequency distribution between the texture and moiré components,this paper proposes to add the location constraint for the frequency distribution of moiré components,which further improves the discrimination between texture and moiré components.2.Based on the above priors,this paper builds low-rank and sparse matrix decomposition model to distinct texture and moiré efficiently,and designs an alternating direction method under the augmented Lagrangian multiplier framework for this non-differentiable model with variable elements.This model can remove moiré patterns as well as retain almost all texture details.3.Based on the sampling rule of the Color Filter Array(CFA)and the structural similarity among different color channels,this paper proposes remove moiré patterns from color textured images by using low-rank and sparse matrix decomposition model to restore the green channel image,and using the guided filtering model to restore the red and channel images based on the restored textured image.This method not only reduces the time complexity,but also improves the overall restored quality of textured images.Results on synthetic and real image data,the method proposed in this paper can restore texture image data with moiré interference efficiently,which promotes the research on textured image demoiréing.Comparing with state-of-the-art methods,both quantitative results and visual results,the method have the best performance in the field of textured image demoiréing.