Abstract
We propose a new similarity measure based on keyword positioning to search images collected from the Web, for the improvement of precision and recall performance. Our image representation model fully exploits the image semantics from the text associated with the images in the Web pages. A prototype search engine Image Search is implemented with these techniques.
Original language | English |
---|---|
Pages (from-to) | 883-886 |
Number of pages | 4 |
Journal | Lecture Notes in Computer Science |
Volume | 3007 |
DOIs | |
Publication status | Published - 2004 |
Keywords
- Digital techniques
- Image files
- Image processing
- Information storage and retrieval systems
- Semantics
- Web search engines