碩士 / 國立臺灣大學 / 工程科學與海洋工程學系 / 92 / This thesis presents an underwater dynamic vision system for biomimetic AUVs . We incorporate a mobile platform with three degrees of freedom to implement the dynamic underwater vision system.
For localization and tracking of a moving object, the bundle adjustment method, camera calibration and stereo imaging are taken into consideration. We construct an image pyramid and choose the winner -update strategy for template matching. Under the condition that optical axis of cameras are not parallel, rational function model is utilized. To estimate the external parameters of cameras, an aggressive algorithm is established. Object ranging is carried out using a DP mathematical model for camera calibration, combining with stereo imaging technique. To predict the position of a moving object, a gray theory is adopted.
To testify aggressive algorithm for the attitude of cameras, the iteration error of external parameters are converged to zero. By comparing the real and estimated coordinates of the reference points on the calibration board, the estimated error is 5%.
To prove the computer programs, two experiments are conducted: (1) Localization and tracking of a moving object under diffuse illumination. (2) 3D mapping of underwater surfaces by using a scanning point beam. In the working range, results suggest that the proposed dynamic underwater vision system is feasible.
| Identifer | oai:union.ndltd.org:TW/092NTU00345017 |
| Date | January 2004 |
| Creators | Huai-Wen Hsu, 許懷文 |
| Contributors | Sheng-Wen Cheng, 鄭勝文 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 65 |
Page generated in 0.2673 seconds