Protrusion Guided 3D Mesh Segmentation

Mesh segmentation, as a crucial step in model analysis and understanding, is vital to the success of the applications, e.g., the 3D reverse engineering and modeling by example. A number of recent works suggested segmenting 3D meshes on the basis of the cognitive psychology concept of "part salience", which comprises three dominant factors: the protrusion degree, boundary strength, and relative size of the parts of the inspected object. In previous works, the estimation of salient feature such as protrusion degree is computation intensive. In this paper, we proposed a simpler measure of the protrusion degree based on the definition of 2D shape protrusion. With our new approach, the significant features can be successfully identified and extracted.