Image, Video, and Computer Vision
With rapid developments in satellite and sensor technologies, there has been a dramatic increase in the availability of multi-modal imagery albeit through remote sensing, multimedia or biomedical type applications. For example, the WorldView-2 sensor can capture images at less than 0.5 m resolution with a collection capacity of 300,000 sq mi/day. Similar challenges are also present in multimedia and biomedical areas. To this end, full motion video (FMV) content is being acquired on an ongoing basis via airborne sensors and UAVs for extracting intelligence to perform day-to-day reconnaissance, combat support, forensic analysis, security, and search/rescue duties. Hence, several FMV Terabytes are being uploaded daily and manually analyzed, contributing to a multi-billion dollar budget. Consequently, techniques for assisted analysis are urgently needed to support analysts in generating effective results in an efficient and timely manner.
The mission of the Image, Video and Computer Vision laboratory (IVCVL) is to conduct research and explore algorithms to establish a firm foundation for mining, exploitation, interpretation, enhancement, classification, storage and compression of multimodal imagery by performing meaningful segmentations//analysis that efficiently combine spectral, gradient, motion and textural information in order to facilitate effective classification of objects/regions that are similar but spatially separated and/or undergoing varying degrees of occlusion. Achieving these objectives will allow analysts/image experts to organize, sort, query information which will facilitate better decision making/understanding in the various image analysis tasks.