Abstract

It is common nowadays to collect image data by scanning the image field with a staggered array of solid state detectors. This may also be from several directions. In view of the finite size of each detector, the reconstructed image is limited in resolution or pixel size. How small may this be made? An efficient way of combining the scan data is localized projection. By this method, the required pixel size is first fixed. Then, the intensity within each pixel area is incremented by the readings from all scanning detectors that spatially intercept it, regardless of the detector size, its direction of scan, etc. This locally projected image will have a characteristic point spread function which may spill out over a few pixels. To reduce this spillage (perhaps to one pixel) a simple filtering operation to accomplish partial deconvolution may be performed. This filtered-locally projected approach has so far been applied to simulated IRAS astronomical data where all the scans are parallel. The effects of noise and dead detectors have also been investigated with encouraging results.

© 1986 Optical Society of America