A novel non-uniformity correction algorithm for parallel scan infrared imager

The parallel scan infrared imager consists of a linear focal plane array (FPA) detector aligned orthogonally to scan direction and imaging requires a one-dimensional scan to cover the field of view in horizontal dimension entirely. Each column of the image is corresponding to one frame outputted from the detector and the non-uniformity of detector is usually manifested as horizontal stripes. In other words, one image could contain many different detector frames, and this is very helpful in non-uniformity correction (NUC). Although a scene contains a lot of temperature information, we could assume that the scene is made up of numerous small fractions with same temperature; thus the fraction could work as a thermal reference source (TRS) and the difference in responses of detector pixels which view the same fraction can be obtained. When the amount of detector frames is large enough, we could find out all the differences in responses of the detector pixels, and offset drifts can be corrected by utilizing these differences. Based on these characteristics, we propose a novel NUC algorithm to correct offset non-uniformity in drift with time.