John R. Smith and Shih-Fu Chang,
Querying by color regions using the emphVisualSEEk content-based visual query system,
In M. T. Maybury ed., Proceedings of the IJCAI Workshop on Intelligent Multimedia Information Retrieval,
1997.
John R. Smith and Shih-Fu Chang,
Querying by color regions using the emphVisualSEEk content-based visual query system,
In M. T. Maybury ed., Proceedings of the IJCAI Workshop on Intelligent Multimedia Information Retrieval,
1997.VisualSEEk is a highly functional image database manipulation system that provides advanced tools for searching, browsing and retrieving images. VisualSEEk is distinct from other content-based image query systems in that the user may query for images using both the visual properties of regions and their spatial layout. Furthermore, the image analysis for region extraction is fully automated. VisualSEEk uses a novel system for region extraction and representation based upon color sets. Through a process of color set back-projection, the system automatically extracts salient color regions from images. This paper describes the implementation of the color query system and examines its role for content-based visual query in image and video databases. The next phase of the VisualSEEk implementation will provide additional tools for querying by texture, shape, embedded text and motion features.