碩士 / 國立臺灣大學 / 造船及海洋工程學研究所 / 84 / The objective of this thesis is to set up a vision model for underwater vehicles; this is, to establish a recognition model for underwater objects by means of optic images and range images which are obtained by the optic image equipment--a video camera and light sources and the simulated acoustic distance measurement equipment--a scanning SONAR in underwater vehicles. Under water, light, absorbed and scattered, reduces its power and contrast; therefore, the viewing range and the quality of underwater images also decline. In this condition, the image contrast can be improved by means of the geometric position of the light source and the camera. If a perfect 3-D range image can be gained, it is possible to calculate the geometric position of this 3-D object and to recognize this object. On the other hand, acoustic range images are obtained through pulse-echo time of flight, which are gained through the acoustic time of flight in media. Acoustic is likely to be affected by some factors such as the quality of the reflection surface; consequently, range data are the image resulting from strong reflection signals. Therefore, the emphasis of this thesis is to utilize both optic images and range images as the vision model for automatic underwater vehicles(AUV).
This thesis has finished common optic image processing, edge-detecting, range image simulation, range image coordinate transformation, given object recognition model. The final stage is to utilize Hough transform to obtain the edge characteristics, and then to describe the distinctive features so as to establish recognition rules. The last step is to take advantage of expert system as an assistant tool to recognize the given underwater objects.
| Identifer | oai:union.ndltd.org:TW/084NTU03624002 |
| Date | January 1996 |
| Creators | 師樂善 |
| Contributors | 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 | 67 |
Page generated in 0.0014 seconds