CBGIR: content-based geographic image retrieval

We describe and demonstrate CBGIR, a web-based system for performing content-based image retrieval in large sets of high-resolution overhead images. The system provides a familiar Google Maps interface to navigate the images and select regions of interest. A query-by-example paradigm is used to retrieve the most visually similar images to this region from a large target set of image tiles. Similarity can be computed with respect to a number of visual features including color, texture, and local invariant descriptors. We describe the salient components of the system and provide sample retrieval results. Categories and Subject Descriptors H.2.8 [Database Management]: Database Applications— spatial databases and GIS; I.5.4 [Pattern Recognition]: Applications; I.4.8 [Image Processing and Computer Vision]: Scene Analysis Keywords content-based image retrieval

• Type: Journal
• Year: 2010
• Where: GIS
• Authors: Shawn Newsam, Daniel Leung, Oscar Caballero, Justin Floreza, Jesus Pulido
• Keywords: Automated Reasoning, Content-based Image Retrieval, GIS 2010, Image, Local Invariant Descriptors