Image reconstruction from multiple 1-D scans using filtered localized projection

The state-of-the-art procedure for image acquisition in downlooking remote sensing and astronomical applications is to use scanning linear arrays. These consist of discrete IR solid-state detectors. We consider the spatial resolution that may be attained with such an array. The individual detectors in the array are typically significantly larger than the blur spot of the imaging optics. Hence a naive recombination of data would produce images which are limited in resolution by the size of the detectors. However, the information content of all available data, including intentional overlap of successive scans and rescans of the same area from different directions, may include much higher spatial frequencies. In the following, we outline a method of filtered localized projection (FLP) which attempts to reconstruct these higher frequencies. Mathematically, FLP consists of a localized summation or projection followed by an inverse-filter operation in analogy to filtered backprojection of computed tomography. The FLP method is applied, using a linear array, to simulated data for staggered parallel scans and to multiple scan directions. Effects of noise and limitations of the approach are discussed.