| Summary: | 碩士 === 國立勤益科技大學 === 電子工程系 === 105 === Mesh segmentation is an essential geometric processing tool for a variety of applications such as motion capture, shape recognition, gesture recognition, 3D model retrieval, and product reverse design or manufacturing, etc. To assist automatic segmentation, the theory of part salience and minimal rule introduced by Hoffman et al.[1][2] has been applied extensively to the field of machine vision and mesh segmentation. In this paper, we proposed an approach to part salience approximation and its application to mesh segmentation. Unlike previous attempts, we assume that the part salience to be a linear combination of the degree of protrusion, the boundary strength, and the relative size. To verify our approach, a part salience-based iterative polygonal mesh segmentation is devised on the basis of such assumption and the segmentation results are scored according to a recent benchmark. According to the test scores, the segmentation algorithm based on our approach obviously out performs a number of concurrent well-know approaches.
|
|---|
