- 発表日 2021/01/01
- Computer-Aided Design and Applications, 18(5), pp.1080-1095,(2021)（英文ジャーナル)
- DOIコード 10.14733/cadaps.2021.1080-1095
Shape Descriptor-Based Similar Feature Extraction for Finite Element Meshing
The aim of this paper is to propose a feature extraction method that allows for the extraction of features that contain nonidentical geometries from a target shape, which are similar to that of a reference feature shape. In this method, the reference feature shape and target shape are represented by a set of shape descriptors defined in triangular meshes. The proposed similar feature extraction method was verified by some case studies in which the relationships between the reference shape and target shape were isotropic scaling, anisotropic scaling, and anisotropic scaling with distortion and complex deformation. In most cases, local regions having similarity to a reference feature could be successfully extracted from a target shape. The results show that the proposed shape descriptor-based feature extraction method effectively allows an engineer to identify where local regions, having similarity to a reference feature, are placed on a target shape. However, the estimated projective transformation in the method still includes non-negligible errors that are not good enough to fit the reference mesh model to the similar features extracted. This dilemma is seen as an appropriate avenue for future research.
関連記事
