東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Rendering driven image based modeling.

Li, Yin.

FindBook

Google Book

Amazon

博客來

Rendering driven image based modeling.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Rendering driven image based modeling./
作者:	Li, Yin.
面頁冊數:	88 p.
附註:	Source: Dissertation Abstracts International, Volume: 65-09, Section: B, page: 4671.
Contained By:	Dissertation Abstracts International65-09B.
標題:	Computer Science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3148580
ISBN:	0496072595

Rendering driven image based modeling.
Li, Yin.

Rendering driven image based modeling. - 88 p.

Source: Dissertation Abstracts International, Volume: 65-09, Section: B, page: 4671.

Thesis (Ph.D.)--Hong Kong University of Science and Technology (People's Republic of China), 2004.

Previous works on image-based rendering (IBR) suggests that there is a tradeoff between the number of images and the amount of geometry required for anti-aliased rendering. However, to reconstruct the necessary geometry information from images is not trivial. Even for latest computer vision techniques, the quality is still not good enough for image based rendering. One major problem is that they lack the ability to properly recognize the occlusion boundaries and group the object at different depths.

ISBN: 0496072595Subjects--Topical Terms:

626642
Computer Science.

Rendering driven image based modeling.
LDR:03258nmm 2200301 4500 001 1846888
005 20051103093559.5
008 130614s2004 eng d
020 $a 0496072595
035 $a (UnM)AAI3148580
035 $a AAI3148580
040 $a UnM $c UnM
100 1 $a Li, Yin. $3 1259288
245 1 0 $a Rendering driven image based modeling.
300 $a 88 p.
500 $a Source: Dissertation Abstracts International, Volume: 65-09, Section: B, page: 4671.
500 $a Adviser: Chi-Keung Tang.
502 $a Thesis (Ph.D.)--Hong Kong University of Science and Technology (People's Republic of China), 2004.
520 $a Previous works on image-based rendering (IBR) suggests that there is a tradeoff between the number of images and the amount of geometry required for anti-aliased rendering. However, to reconstruct the necessary geometry information from images is not trivial. Even for latest computer vision techniques, the quality is still not good enough for image based rendering. One major problem is that they lack the ability to properly recognize the occlusion boundaries and group the object at different depths.
520 $a In this thesis, we introduce an interactive image based modeling framework, rendering driven image based modeling, which leverages the vision algorithms with human's interaction. It reconstructs the geometry in an iterative way. The criterion for the reconstruction is to improve the rendering quality. Through a well designed user interface, the user can supervise the reconstruction process and specify certain information to computer, such as grouping of pixels, errors, or confirmation of correctness.
520 $a One of our essential observations is that the alpha matting along occluding boundary is very important for anti-aliasing rendering, while the geometry error inside opaque object is not as important. Therefore, we introduce the coherent layer representation to rectify the boundary location and extraction. To correctly separate the foreground colors and transparencies for the coherent layers, we also designed coherent matting, an enhanced Bayesian matting algorithm. This system, also known as pop-up light field, is a practical interactive system for sparse lightfield modeling and rendering.
520 $a In order to save user's interaction time, we also designed an interactive image segmentation tool, called Lazy Snapping. It separates the segmentation process into two steps, from coarse scale to fine detail. We deliberately designed two user interfaces for the two tasks. One is a coarse but quick marking process, and another is a polygon editing UI for boundary refinement. Moreover, Lazy Snapping provides instant visual feedback due to our enhanced region-based graph cut segmentation algorithm. Our usability studies indicate that Lazy Snapping system provides a better user experience and produces better segmentation results than the state-of-the-art interactive image cutout tool, Magnetic Lasso in Adobe Photoshop.
590 $a School code: 1223.
650 4 $a Computer Science. $3 626642
690 $a 0984
710 2 0 $a Hong Kong University of Science and Technology (People's Republic of China). $3 1249812
773 0 $t Dissertation Abstracts International $g 65-09B.
790 1 0 $a Tang, Chi-Keung, $e advisor
790 $a 1223
791 $a Ph.D.
792 $a 2004
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3148580