碩士 / 國立高雄第一科技大學 / 資訊管理所 / 98 / Although video technology have been widely used and developed in recent years, however there are still few researches for fish surveillance. Taiwan is a beautiful island and aquariums stand everywhere. But there is still a shortage of fish knowledge or learning systems for the country people. Therefore, it is very important to develop related fish information systems.
Mean shift tracking is one of the popular object tracking methods. But this method is suffered when the object moves with high speed or overlaped with other objects. In this study, we introduce a modified Mean-Shift method with a region prediction strategy to improve the fish tracking performance. The region prediction strategy can provide extra information to predit the way of object moving. In addition, a region-based tracking method is provided to quickly find back the correct object when the original object tracking is missed.
The applications of the proposed method include the tank environment analysis and construction, fish object detection, fish tracking, fish recognition and knowledge-base construction. And we accomplish the real-time fish indexing using the integrated computer vision techniques. Experimental results show the high performance of the proposed system.
| Identifer | oai:union.ndltd.org:TW/098NKIT5396027 |
| Date | January 2010 |
| Creators | Jhih-Wei Kang, 康志偉 |
| Contributors | Jia-Hong Lee, 李嘉紘 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 69 |
Page generated in 0.0015 seconds