| Summary: | 博士 === 國立中央大學 === 機械工程學系 === 106 === The shape-from-silhouette (SFS) method has been widely used in three-dimensional (3D) shape reconstruction. It uses silhouettes of a series of two-dimensional (2D) images of an object captured from multiple viewpoints to generate a 3D model that describes the visual hull of the object. The inherent problem with this method is the virtual features that appear all over the model because of being formed by the intersection of multiple polyhedra from different views. In addition, concavities on the object may wrongly be modeled as convex shape because they are invisible on image silhouettes. On the other hand, real object features may exist which should be preserved on the model. The purpose of this study is to propose a method for the generation of 3D model from silhouettes of multiple images as well as a quality improvement method to overcome the afore-mentioned problems. The proposed 3D modelling method focuses on accurate evaluation of 3D points intersected by all polyhedra from different views and the removal of poor meshes on triangulation. The quality improvement method is essentially an iterative procedure, in which the first iteration is used to eliminate virtual features and artifacts, and the remaining iterations are applied to smooth the model, while preserving the consistency of all silhouettes. The proposed method is designed to be used for product presentations in e-commerce, so the 3D model must be covered with the color and texture of the object. Several examples are presented to illustrate the capability of the proposed method.
|
|---|
