Abstract: An automatic target tracking system comprising video means, target tracking means, correlation means and aimpoint location means which predicts the subsequent position of a target after each viewing by the video means and then correlates this with the actual viewing of the target position so as to estimate that position within a given field of view.

Abstract: This invention relates to a method for analyzing a viewed scene containing a target. The scene is digitized forming a binary image in which the target is distinguished from the background. The image is then processed to form a contoured representation of the target, and a target skeleton. Finally a contoured skeleton is produced and the target aimpoint is determined.

Abstract: An item of interest may be located in a scene image by scanning a reference image containing the item (extracted from a previous image of the scene for example) over the scene image and, for each position of the reference image, correlating corresponding pairs of pixels from the reference and scene images. Herein, to avoid inaccuracies due to bright spots say affecting a small portion of the scene image, for each position of the reference image, all the pixel pairs are correlated and an average similarity value assigned to that reference image position.