Extraction of Interest Points on 3D Meshes Based on Bilateral Filtering

Abstract

Interest points are significant to 3D matching, recognition and 3D retrieval. We use bilateral filtering to extract 3D interest points. Differing from prior work, a novel approach is proposed to select the local neighborhood which is named trust area. And we also use multi-scale DoG to compute the saliency of the vertices. The interest points are extracted by saliency.