Abstract: A technology based on adaptive image clutter suppression for detecting dim point moving target is investigated. The quad-tree algorithm is used for segmenting a non-stationary original image to some sub-blocks, which are quasi-stationary data. Then a LS adaptive filter is adopted in these blocks to estimate and suppress clutter, which leads to quasi GWN background obtained. Thus the hypothesis of movement continuity is proposed to model target positions in sequential frames with high-order markov chains and construct the trajectories state space. According to the model a multi-frame detection algorithm of nonlinear integration along tracks is adopted. Consequently it abstains from the disadvantage of excessive searching operations due to using traditional three-dimensional matching algorithm. And it also avoids the SNR fall that three-to-two-dimension projection detection can bring forth. Theoretic analysis and many simulations can prove its validity.

Key words: information processing technology, dim point moving target detection, quad-tree segmenting, Markov model, nonlinear integration