Fast LIC image generation based on significance map

Abstract

Although texture-based methods provide a very promising way to visualize 3D vector fields, they are very time-consuming. In this paper, we introduce the notion of “significance map”, and describe how significance values are derived from the intrinsic properties of a vector field. Based on the significance map, we propose techniques to accelerate the generation of a line integral convolution (LIC) texture image, to highlight important structures in a vector field, and to generate an LIC texture image with different granularities. Also, we describe how to implement our method in a parallel environment. Experimental results illustrate the feasibility of our method.