点击次数：

论文类型：期刊论文

发表时间：2013-05-01

发表刊物：Journal of Information and Computational Science

收录刊物：EI、Scopus

卷号：10

期号：7

页面范围：1907-1916

ISSN号：15487741

摘要：In this paper, we propose a model-based reconstruction method for recovering a 2D parametric styling model with hierarchical structure from an image. Our method extracts characteristic features for industrial design with following essential components: construction of the template database as prior knowledge, contour detection from images and layered segmentation, key points generation via shape correspondence with reference model, sparse reconstruction based on shape space and hierarchical fitting. The whole recovery process is semi-automatic. We adopt the challenging vehicle styling as test examples, and establish associated styling model database. Numerical experiments show that our method is effective to recover the styling features with high precision and hierarchical structure. We also give an image reshaping application based on our hierarchical features. ? 2013 by Binary Information Press.