返回Google圖書搜尋

3D Reconstruction of Deformable Surface Based on Binocular Vision

出版	ACTI INNOVATION Incorporated, 2020-11-30
主題	Computers / General
ISBN	19543180069781954318007
URL	http://books.google.com.hk/books?id=0qkezgEACAAJ&hl=&source=gbs_api

註釋The 3D reconstruction of deformable surfaces is a classical problem in the field of computer vision, which has many important applications in virtual reality, surgical navigation, and training. For example, accurate 3D reconstruction for intracavitary soft-tissue scenes based on a stereo-endoscope can effectively promote surgeons and surgical robots' perception in a Minimally Invasive Surgery (MIS), and thus improve the precision and safety of surgical operations. However, the dynamic deformable soft-tissues bring great challenges for accurate reconstruction. This book provides an in-depth discussion about the binocular reconstruction techniques for dynamic deformable surfaces such as beating heart surfaces. The camera models and calibration methods used for stereo vision are introduced first. The classical stereo matching methods are then presented. The deformable-model-based reconstruction methods are discussed emphatically, from the common-used deformable models, such as the triangular affine transformation, the B-spline model, and the Thin Plate Spline (TPS), to some improved models proposed recently, such as the Quasi-Spherical Triangle (QST) model, the overlapping TPS model, and the Triangular Radial Cubic Spline (TRCS) model. This book includes Reconstruction techniques based on machine learning, statistical deformable modeling using principal component analysis, and Deep-learning methods for deformable surface reconstruction with selected experimental results.