Image Pre-processing Algorithms for Detection of Small/Point Airborne Targets

The problem of detecting small/point targets in infra-red imagery is an important research area for defence applications. The challenge is to achieve high sensitivity for detection of dim point like small targets with low false alarms and high detection probability. To detect the target in such scenario, pre-processing algorithms are used to predict the complex background and then to subtract predicted background from the original image. The difference image is passed to the detection algorithm to further distinguish between target and background and/or noise. The aim of the study is to fit the background as closely as possible in the original image without diminishing the target signal. A number of pre-processing algorithms (spatial, temporal and spatio-temporal) have been reported in the literature. In this paper a Survey of different pre-processing algorithm is presented. An improved hybrid morphological filter, which provides high gain in signal-to-noise plus clutter ratio (SCNR), has been proposed for detection of small/point targets.