Automatic video image moving target detection for wide area surveillance

Two image processing techniques developed for moving target indication from video infrared imagery in natural scenes are presented. It is shown that frequency domain spatio-temporal filtering of video sequences and spatio-temporal constraint error of image frame pairs are able to detect and track moving targets (e.g., personnel) in natural scenes in spite of low image contrast, changes in the target´s infrared image pattern, sensor noise, or background clutter. The effectiveness of the motion filtering algorithms when applied to sequences of data corrupted with additive noise is shown. The effectiveness of the CFAR (constant false alarm rate) adaptive threshold algorithm in controlling the false alarm rate for motion detection has been demonstrated. These steps have permitted substantial data reduction so that real-time processing is possible