Zhanpei Zhang Beicheng Liang Tingting Ren Chengmiao Fan Rui Li Mu Li ^*

• Harbin Institute of Technology, Shenzhen, Shenzhen, China

Despite the fact that the text image based Optical Character Recognition (OCR) methods have been applied to wide applications, they do suffer from performance degradation when the image is contaminated with moir é patterns for the sake of interference between the display screen and the camera. To tackle this problem, in this paper, we propose a novel network for text image demoir éing. Specifically, to encourage our study on text images, a dataset including an amount of pairs of images with/without moir é patterns is collected, which is specific for text image demoir éing. In addition, due to the statistical difference among various channels on moir é patterns, a multi-channel strategy is proposed which roughly extracts the information associated with moir é patterns and subsequently contributes to moir é removal. Besides, our purpose on the text image is to increase the OCR accuracy, while other background pixels are insignificant. Instead of restoring all pixels like that in natural images, a character attention module is conducted, allowing the network to pay more attention on the optical character associated pixels and also achieving a consistent image style. Being beneficial from this strategy, characters can be more easily detected and more accurately recognized. Dramatic experimental results on our conducted dataset demonstrate the significance of our study and the superiority of our proposed method compared with state-of-the-art image restoration approaches. Specifically, the metrics of Recall and F1-measure on recognition are increased from 56.32%/70.18% to 85.34%/89.36%.