Shape Context Preserving 2D Illustrative Visualization Using Deformation

Illustrative visualization exploits effective visual abstractions to express the visualization in a clearly understandable way. It has been widely used in medical education, surgical training, and fluid simulation. We present a novel 2D shape deformation algorithm and exploit it for illustrative visualization.In the kernel of our approach is a new 2D gradient domain based mesh manipulation algorithm, which is solved with Poisson equation. Compared with previous methods, our solution has the following improvements:First, we introduce an anisotropic physical properties for individual illustrated objects, i.e., assigning different rigidities to separated 2D regions along x and y axe. This scheme facilitates physically plausible simulation for not only the full rigid objects like bones, but also anisotropic objects like muscle.Our second contribution is a curve analogy based shape style transfer technique from reference illustrations. Our approach allows for learning distinctive shape styles in an intuitive fashion, greatly comprehending the re-creation capability of illustrative visualization. Again, it is accomplished under the differential curve and mesh editing framework.Finally, we introduce the concept of shape context that has been adopted in biomorphism. We formulate the shape context of an underlying object as a mesh constructed from a set of selected feature points. By keeping the differential properties of the mesh unchanged during deformation, shape preserving illustrative visualization is conveniently achieved.We have built an interactive system and tested comprehensive examples. Experimental results show that visually and physically plausible illustrations of anatomy deformation can be easily fulfilled in real-time. Preliminary evaluations and feedbacks from medical illustrators are quite encouraging.