Luis Cardona, Suguru Saito (Tokyo Institute of Technology)
In this paper, we propose an object-space version of the Gradient Vector Flow (GVF) and present a quantitative evaluation of a line drawing stylization system based on user annotations. In our system, the stylistic annotations of the user are extrapolated to arbitrary viewpoints by matching the user annotated strokes to view-dependent feature lines extracted from 3D models. The new object-space attraction field proposed in this paper is robust to global and local occlusion, and is evaluated by numerically comparing its line matching accuracy with previous methods. We also present an extended evaluation of the usability and performance of our stylization system.