Detail Pattern recognition and 3D object reconstruction using active vision Bibliografi Author: Yi, Xilin ; Camps, Octavia I. (Advisor) Topik: ENGINEERING; ELECTRONICS AND ELECTRICAL|COMPUTER SCIENCE Bahasa: (EN )    ISBN: 0-591-41961-0     Penerbit: THE PENNSYLVANIA STATE UNIVERSITY     Tahun Terbit: 1997     Jenis: Theses - Dissertation Fulltext: 9732408.pdf (0.0B; 0 download) Abstract 3D shape reconstruction and modeling can be applied to many applications as hazardous environment and industrial automation. In this research, three correlated approaches for 3D shape reconstruction with active sensor and illumination are proposed, namely, shape from occluding contours, shape from occluding contours and highlights, and shape from specular highlights. Shape from occluding contours method differs from previous affine based computational approaches. It uses Hausdorff distance based pattern recognition system to track stationary features. With the hypothesis that surface markings behave similar to stationary points when camera moves around the object, Hausdorff distance tracking scheme distinguishes surface markings from occluding contours. This method extends the validation to weak perspective projection, does not require to search for correspondences between frames, can handle scaling between consecutive images, thus can estimate the full Euclidean surface structure, and does not require to know camera motion. The extensive experimental results show that it is robust to the occlusion of feature points and image noise as opposed to previous affine-based approaches. Shape from occluding contours and highlights method explores the relationships between highlight and occluding contour. With the help of active camera, it can predict highlight from occluding contour, and vice versa, and quantitatively recover the 3D object shape with the aid of both. All previous approaches either consider only occluding contours and eliminate highlights, or consider only highlights and ignore occluding contour information. The proposed approach presents a simple theorem to use highlights and occluding contours to recover 3D information. It doesn't have the problem of solving for unknown constants as in previous approaches. This approach needs to identify the occluding contours with the help of the first approach. Shape from highlights method takes advantage of active illumination, and presents a novel relation between the specular highlight movement and local depth. For the first time, we develop a simple algorithm for 3D shape reconstruction from specular highlight. This method needs to find the depth of initial points with the help of the second approach. Rigorous experimental protocols have been developed for Hausdorff distance based pattern recognition system and 3D shape reconstruction approaches. Opini Anda Klik untuk menuliskan opini Anda tentang koleksi ini! Lihat Sejarah Pengadaan  Konversi Metadata   Kembali

Copyright © 2006, 2007 Unika Atma Jaya, all rights reserved