Moire removal algorithm based on deep learning

• 1、School of information and Communication Engineering,Beijing University of Posts and Telecommunications,Beijing,100089

Abstract：In daily life and industrial sites, moire interference often occurs when the camera takes digital screens, resulting in damage to image quality, which not only affects the user experience, but also affects the image analysis and processing in industrial sites. The traditional moire removal algorithm uses low-pass filter to filter the high-frequency information to achieve the function of Moire removal. This algorithm not only loses the details of the image, resulting in reduced quality, but also the moire removal is incomplete. In order to solve the above problems, an adaptive window filtering algorithm is proposed to optimize the traditional algorithm to improve the image quality, and a multi-scale fusion network model is designed, which greatly improves the network performance and image quality by using multi-scale context aware fusion.

Keywords： Artificial intelligence, neural network, moire removal, digital image processing