Spatial-temporal adaptive clutter classification suppression and dim small moving targets detection

A new method was proposed for the solution of an important class of multidimensional signal detection problems: the detection of dim, small and moving targets of unknown position and velocity in heavy clutter in a sequence of digital images. By studying temporal gray-level moment of input sequence, the pixels were classified into two categories: stationary clutter and variational clutter. And a nonparametric temporal filter and a LS adaptive filter were applied for suppressing clutter respectively, thus the raw images were transformed into quasi SPGWN model. Then according to a target model of multi-pixel per frame, a detection algorithm integrating signal energy in spatial and temporal domain jointly was employed. The algorithm can improve SNR evidently and can easily be implemented in real time. The theoretic analysis and many simulations of real data verify the validity of the method.