Multi-scale mesh saliency with local adaptive patches for viewpoint selection

Our visual attention is attracted by specific areas into 3D objects (represented by meshes). This visual attention depends on the degree of saliency exposed by these areas. In this paper, we propose a novel multi-scale approach for detecting salient regions. To do so, we define a local surface descriptor based on patches of adaptive size and filled in with a local height field. The single-scale saliency of a vertex is defined as its degree measure in the mesh with edges weights computed from adaptive patch similarities weighted by the local curvature. Finally, the multi-scale saliency is defined as the average of single-scale saliencies weighted by their respective entropies. The contribution of the multi-scale aspect is analyzed and showed through the different results. The strength and the stability of our approach with respect to noise and simplification are also studied. Our approach is compared to the state-of-the-art and presents competitive results. Graphical abstractDisplay Omitted HighlightsWe model the surface of the mesh by a non oriented graph. We associate to each vertex a normal vector and a 2D tangent plane.To describe the mesh surface, we use local patches with adaptive size. These patches are filled with a local height field of the spherical neighborhood.The single-scale saliency is defined as the vertices's degree measures with edges weights accounting for vertices similarities.The multi-scale saliency is defined as the average of single-scale saliencies weighted by their respective entropies.We discuss, compare the results with the state-of-the-art, and demonstrate the stability of our approach. We also present an application for selecting the most important viewpoints of 3D meshes.