Application Of Digital Image Inpainting In Mural

Mural painting of Tang Dynasty is one of the precious historical and culturalheritage. It is an image data which is important to the research on social life in TangDynasty, especially the aristocratic life and spiritual quest. Mural painting of TangDynasty has thousands years history, now various forms of diseases include mildew,fade and crack has been appeared in the unearthed Mural painting of Tang Dynastybecause of the factors such as the collapse of the Tomb, the erosion of the flood andvandalism. By using digital image restoration technology simulate the process ofmanual repair murals to repair the damaged mural virtually can not only reduce thedifficulty of protecting ancient murals and improve the work efficiency, but alsoprotected murals from the Man-made secondary damage.This paper describes the preservation of the status quo for mural painting of TangDynasty, and summarizes the current classical digital image restoration techniques andthe application of digital image restoration technology in the ancient murals. Using it asa base, the main works of this research are as follows:(1) Segmentation and annotation of damaged areas of mural based on FuzzyClustering Manual. Due to the manual label of damaged areas cost numerous time andenergy, this paper has Segment mural combined the image segmentation algorithmbased on FCM with the color and spatial information. By regarding the damaged muralareas as the foreground, and the non-damaged areas as the background, thesegmentation problems of the damaged areas would be converted to the foregroundsegmentation. At last, labeling segmented regions. Application of this method couldcomplete the auto label, So that the efficiency of the mural annotation has beenimproved.(2) Adaptive mural image inpainting algorithm based on content. Considering thebother that exemplar-based image inpainting algorithm might take quite a long time, damaged areas can be divided into flat area and texture area by using local gradientchange in this paper. For the texture regions, improved exemplar-based imageinpainting algorithm would be applied, and for the flat regions, FMM would be used. Atthe same time, in order to avoid the deviation accumulation problems during the processof repairing and the extreme impact, this paper redefined the formula of prioritythrough applying addition instead of multiplication, and the priority computationformula is revised by adopting a weighting factor of dynamic increment to adjust thepriority of filling edges, The experimental results show that this method would ensurethe improve the image repair rate under the premise of guarantee of repair quality.