Application Of Image Inpainting Algorithm In Traditional Chinese Paintings

Chinese painting is one of our traditional treasures of art, which has made a great contribution to the heritage and development of the Chinese civilization. However, many paintings more or less have some kinds of damage because of their age, or as a result of not properly preserve, transport, damp and other factors. In this text, we analyze issues emerging in the traditional Chinese paintings, and attempt a classification, while also give some principles of image inpainting. We use different methods to solve different issues in the text.To repair crease in paintings, we use an improved exemplar-based inpainting algorithm. After the target region being designated through user's interaction, the algorithm will automatically complete the rest work. To make the algorithm more fit the problem, we use the result of image segmentation as a guide of the process of synthesis. Meanwhile, we use a method called GraphCut to get a seamless paste. After all is finished, the use of artifact detection is to find fault region, and further improve repairing effect.In the text, we use image interpolation with guided field to repair defaced background of Chinese paintings. We try to get a seamless result by solving the Poison equation. To save some useful information in defaced background, some seed points need set. Different to the traditional method, it doesn't need the source region as large as the target region. If the source region is not large enough, we use some method to expand the region, such as texture synthesis, or through a simple reflection expansion.Experimental results show that the methods we used to deal the two issues of Chinese paintings work effectively and produce satisfying results in visual.In the final chapter, we summarize the whole paper, draw conclusions and propose several possible future follow-up works.