Title :
Acquiring and rendering high-resolution spherical mosaics
Author :
Kropp, Adam ; Master, Neel ; Teller, Seth
Author_Institution :
Comput. Graphics Group, MIT, Cambridge, MA, USA
Abstract :
We describe an acquisition and viewing method for high-resolution, high-dynamic range spherical image mosaics. These mosaics consist of hundreds of high-resolution, high-dynamic-range color images, each annotated with 3-DOF orientation about a common optical center. An automated acquisition platform captures a pre-specified hemispherical tiling of images. Each image is acquired with specified approximate camera rotation and bracketed exposures for high-dynamic range. Relative rotations among images, as well as internal camera parameters, are automatically refined using a correlation-based global optimization. The improved rotation and internal parameter estimates are used in an interactive viewer for smooth inspection of the resulting large spherical mosaic. The renderer uses a modified form of adaptive, incremental, parallel raycasting to sample the original images at pixel centers, preserving detail from the raw images while maintaining interactive display rates. We show an example dataset that resolves millimeter-scale detail at a viewing distance of ten meters
Keywords :
image reconstruction; rendering (computer graphics); acquisition; automatic reconstruction; camera rotation; image mosaics; rendering; textured 3D geometric models; urban landscapes; Cameras; Color; Computer graphics; Dynamic range; Geometrical optics; Inspection; Optical receivers; Optical sensors; Parameter estimation; Rendering (computer graphics);
Conference_Titel :
Omnidirectional Vision, 2000. Proceedings. IEEE Workshop on
Conference_Location :
Hilton Head Island, SC
Print_ISBN :
0-7695-0704-2
DOI :
10.1109/OMNVIS.2000.853803