Simulation And Implementation For Chinese Ink And Wash Painting Which Based On Multi-target Segmentation Of Colorful Images

Chinese ink-wash painting is a Chinese traditional art which has a long history,the simulation technology of the ink painting has become the art in the field of computer graphics and digital image processing is one of the most challenging tasks.Combining computer technology and traditional article culture by using non-photorealistic rendering technology in Chinese characteristics of ink painting style,not only can be extended to the field of image processing,but also carry forward the Chinese traditional culture,that has high realistic meaning and applied prospect.In this paper,author researchs in ink painting art style,brush model,and the ink diffusion model,so author puts forward a simulation method of Chinese ink painting style image based on multi-object segmentation,the method relies on a little prior knowledge which is from users.Finally,according to my design algorithm,the two-dimensional image which has few channels can translate into grey-scale.Through the study and observation of ink painting,and visiting some ink painter,author found that the painters usually use fine pen with a little pure water to sketch the contours of objects,and the contours are not continuous,then the internal even have affordable shades to freehand coloring.Based on this point,author divides the simulation process into four steps.Firstly,according to the prior knowledge,this paper will associate the related objects,and divide different objects which are with no relation to each.And then,divide the input into multi segmentation by the algorithm of the strong correlation through neighbor search and morphology.Next,extract the edge of the object,and then establish fuzzy operator for the edge of Chinese brush and paper.Finally,establish the central operator for diffusion of ink.When extracting different targets,author use 8-local search method to grow regions.At the same time,taking into the instability of prior knowledge,so using the morphological close to process the specific areas which aren't belong to each region.author applied Laplacian to extract fine-grained edges.On the other hand of the edge's operator of brush and paper,author doesn't have to simulate the dynamic characteristics of the brush,and also doesn't calculate the distortion of brush,but create fuzzy convolution kernel for that's shape.When fill internal,author did not use pixel equalization to mainstream ink because of the ink painting style.The ink painting pays attention to the coordination of gray,so thesis uses the morphological algorithm for filling,and finally create the diffusion kernel to filling the internal of outer edge.The proposed algorithm requires only a small amount of user prior knowledge and achieves complex multi-channel color ink painting style conversion perfectly and smoothly.Compared with the previously used grab Cut,watershed and mean of shift algorithm,the algorithm improves the efficiency mostly.