| Since the mid 1990s, Non-Photorealistic Rendering (NPR) has gradually become one of the hottest spots in computer vision and graphics.Different from the conventional photorealistic rendering, NPR focuses on the delivery of artistic features and effects of images instead of the truthfulness of rendering results. Digital art painting occupies an important role as well as a hot spot in the research of NPR. In the research of painting, the core process is to generate a personalized artistic painting from an image, through limited interactions.In order to make the effect closer to the artistic effect finally, based on painters' practices from of old, human beings' visual perception of arts and professional painting effects are analyzed, thus a mathematical model of the painting process is abstracted from perception to final rendering, most of operations are automatic, personal elements are added by few operations. In the whole model, the Mean Shift algorithm is used to do the"Over-Segmentation"; then, the result is used to build the Markov Random Field; and, using the Graph Cut algorithm, the objects in the image are segmented layer by layer. Furthermore, the main structure of the object is parsed by the Primary Sketch algorithm. According to the main structure, the orientation field is made by the oriented diffusion. And through color transferring, the image takes on the color of art paintings which provides pigments for rendering. Using the orientation field and the artistic pigments based on the original image, the digital art painting is rendered by the many layers'"pursuit"algorithm finally. The whole process simulates the human painting process basically.Compared with some existing results, the synthesized effects are better in both global artistic expressive force and local photoreality. Despite the obvious distance to real painting effects (especially masterpieces), it is believed that with the improvement of the model and related algorithms, it would become nearer and nearer. |
PDF Full Text Request