A GAN-based Denoising Method for Chinese Stele and Rubbing Calligraphic Image

Abstract

Chinese calligraphic images have important artistic and historical values. However, subjected to hundreds of years of natural weathering, corrosion and man-made destruction, Chinese calligraphic images inevitably contain some special noise, such as dotted noise, flake corrosion noise and scratch noise. How to denoise this special noise is a challenge for digital preservation of Chinese calligraphic. In this paper, we propose an end-to-end calligraphic image denoising algorithm based on a well-designed generative adversarial network. The generator contains a recurrent network and a denoising autoencoder. By introducing an attention mechanism, we use a recurrent network with multiple progressive network units to generate a noise attention map. Through the noise attention map, the denoising autoencoder can restore the noisy calligraphic image into a clean image with reduced noise or even no noise. The extensive experiments results show that the results of our method are better than those of other comparison methods in terms of visual effects, PSNR and SSIM.

Tasks

Reproductions