Clutter metrics in infrared target acquisition

Recent research on human target acquisition using electro-optic systems in cluttered environments has proposed that an appropriate measure of the clutter content of a scene is the average variance of the intensity of the scene when subdivided into blocks whose linear dimensions are approximately twice the size of the target. We show that, for natural scenes characterized by a particular fractal pattern and spatial frequency spectrum, such a criterion is equivalent to the analysis of the number of edge pixels appearing on the picture after passing through a DOOG filter. In addition, the statistical criterion can be related to the phase spectrum of the picture and the fractal characterization of the picture. Based on psychophysical data, this suggests that a search process based on 'popouts' and edge detection for the human visual system is appropriate. Incorporating this effect both in the analysis of the clutter and of the target image allows us to propose a new target detection model; good corroboration with experimental data has been obtained.