The Design And Implementation Of Chinese Paper-cutting Generation Method Based On Its Artistic Characteristics

Chinese paper-cut is one of the world's intangible cultural heritage,it can exhibit a variety of shapes and reveal profound cultural morals with cheap paper materials.It has high ornamental and collectible values,is thus the treasure of Chinese traditional handicraft.However,most of our paper-cutting works scatter in the folk,and paper-based paper-cuts can not be long-term preserved,which result in few of them are handed down,thus the digital protection of paper-cuts is extremely urgent.For this reason,Chinese paper-cut has also been listed as one of the most important ten national cultural heritage rescue projects.More importantly,the design of paper-cuts relays on the rich experience,and the existing paper-cutting themes are limited to the traditional wedding and fete ceremony.It is difficult for amateurs to cut new themes for modern things and objects,which greatly limits the propagation and inheritance of paper-cutting art.In this thesis,we focus on the shape matching method for paper-cutting pattern and image-based automatic paper-cut generation using computer technologies,based on artistic characteristics of paper-cuts,such as pattern element repetition,the richness of cultural meanings and the overall geometric connectivity of Chinese paper-cuts.The major research works are as follows:1.We propose a continuous shape descriptor for the distribution of discrete contour points,which is integrated into the contour point distribution histogram and shape context methods.Firstly,a polar plate is constructed,and then continuous influence value of each bin in the polar plate is computed to have the descriptor vector.Finally,the difference between shape descriptors is calculated quantitatively.Due to the large amount of noises in the scanned paper-cutting images and zigzag artifacts in the extracted contours from natural images,we introduce a continuous local histogram for discrete contour points that improves existing shape descriptors to be less sensitive to noises.2.We propose an image-based fast paper-cut generation method that automatically transfers patterns from reference paper-cuts to the target image.This method categorizes paper-cutting patterns into physical patterns and decorative patterns.The transfer of physical patterns is accomplished by an automatic matching approach.A boundary band map of the target object is constructed by an improved intensity graph firstly,and then a vector field of the physical pattern is computed and finally boundary band matching method is used to find the target position of the physical pattern transfer.We separate decorative patterns according to their distances to outer contours.Ones away from outer contours are transferred by remapping and rescaling,and ones close to outer contours are transferred based on their spatial relationships with closed contour edges to avoid going beyond the contour.Finally,all transferred patterns are examined whether they exceed the range of outer contours or overlap with other patterns,and the problem of boundary adhesion between different shape elements is solved.Paper-cutting works produced by our method not only have the shape features of real objects but also reveal three artistic characteristics of paper-cuts,which is crucial for amateurs to design paper-cuts quickly.3.We design and develop a Chinese paper-cutting generation system,which implements the aforementioned algorithms and transferring strategies from a natural image to a paper-cut.The purpose of this thesis is to use computer technologies to help amateurs to design personalized paper-cuts.A large number of user studies have demonstrated the effectiveness of our method in reducing the difficulty of paper-cutting design and increasing the efficiency of paper-cutting design,and thus it is of great significance to explore the new model of digital protection of intangible cultural heritage based on computer graphics technologies.