Title: Feature-guided convolution for salient rendering of 3D meshes

Authors: Kyungha Min

Addresses: Department of Digital Media, Sangmyung University, Hongji-Dong, Jongro-Gu, Seoul, Korea

Abstract: We present a novel algorithm that depicts the salient shape of a 3D mesh by mimicking pencil drawing style. In estimating the saliency of a mesh, we consider both view-independent and view-dependent saliency. The view-independent saliency is estimated from the strength of silhouette and the view-dependent one is computed by approximating the curvature of a mesh. We extend the pencil drawing algorithm for a mesh and develop a convolution-based scheme that renders the saliency of a mesh in a pencil drawing style. Furthermore, we develop a scheme that controls the coherence of the depiction through an animation of a mesh.

Keywords: non-photorealistic rendering; NPR; saliency; silhouettes; mesh curvature; 3D meshes; pencil drawing; feature-guided convolution; salient rendering; mesh animation.

DOI: 10.1504/IJESMS.2015.066122

International Journal of Engineering Systems Modelling and Simulation, 2015 Vol.7 No.1, pp.1 - 5

Received: 10 Jun 2013
Accepted: 08 Oct 2013
Published online: 24 Jan 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article