A self-adaptive algorithm for small targets detection in clutter scene inspired by insects compound eye

It is an intriguing challenge that insects can catch small moving targets amidst the clutter and moving scene accurately by the compound eye system which is with limited resolution. Especially when the velocity of the target is matched to the background, insect can also detect and catch it. Inspired by the neural mechanism of compound eye, this paper proposed a bio-inspired nonlinear self-adaptive detector algorithm to detect small move targets in the clutter scene. At first, multiple spatial-temporal filters were used to degrade the clutter background and enhanced the contrast between target and background. Then the scene was divided into two independent polarity contrast channels, so that the texture and noise were suppressed effectively. At last the different polarity channels were correlated like Reichardt Elementary Motion-Detector. This proposed algorithm didn´t need to estimate the background and the relative motion between target and background, just dependent on the multi-spatiotemporal filter to extract the small targets. Compared with the common signal process algorithm, the bio-inspired detection algorithm was fast, easy and robust.