50 likes | 142 Vues
Develop tools to search and index multimedia data by content within image databases for various domains. Plan includes feature extraction, database creation, and image comparison using software tools like Matlab and Java. Learn about image analysis, retrieval, and clustering. Schedule involves poster and paper presentations, software implementation, and team meetings. Results include implementing algorithms in Java for query by image content.
E N D
Problem Query image by content in an image database
Motivation Image databases are becoming more and more common and larger in several distinct application domains, such as: multimedia search engines, digital libraries, art, medical, geographic, criminal investigation. There is a need to develop efficient tools for searching (indexing) and browsing the multimedia data by content.
Plan of attack • Ideas • extract features from the query image (color, texture, shape, position, regions) • create a database, indexing the pictures • compare the query image to images from database, releaving similarities • Tools • Matlab, Jasc Photoshop, Java, DBMS • Things to learn • Image analysis, Image retrieval, comparing features, clusterizing images (SOM), Java, SQL
Schedule • Poster - 1st October 2003 • Paper - 1st November 2003 • Software implementation starts on 1st October 2003 • Team meetings - twice a week, on Tuesdays and Saturdays • 3 subgroups: research, implementation and documentation
Results • Good knowledge in image analysis applied in image retrieval, query by content • Algorithms (Matlab, Java) for query-by-image-content field • Try to implement these algorithms in Java application