Surfel Set Simplification With Optimized Feature Preservation

Yuxue Fan, Yan Huang, Kangying Cai, Feihu Yan, Jingliang Peng

Abstract: In this paper, we propose a novel scheme for simplifying a surfel set with the resultant surfels computed and distributed to preserve prominent geometric and textural features. It works by iteratively collapsing local neighborhoods around surfels until a given data reduction ratio is reached. For optimized feature preservation, novel techniques are proposed in various steps of the scheme. The local neighborhood collapses are prioritized according to a cost metric that takes into account the local complexities of both the geometric and the textural information. Methods for surfel attribute computation are proposed for faithful representation of geometric and textural features in the simplified model. The proposed algorithm is further extended to support out-of-core simplification for large models by optimally determining the reduction ratios for different parts of the model in consideration of the surface features. Experimental results demonstrate the outstanding performance of the proposed scheme.