Detecting and tracking dense multi-target under intense interference remains a major challenge in scientific research.It also has important applications in national defense and economy as well.In this paper

we proposed "signature" using the target physical characteristics and motion features of fusion measurement in spectral

spatial and temporary spaces.Thus we could eliminate noise at the earliest time by applying the kinetics Markov belief and the feature Markov belief of the target signature as well as using data association result as feedback of adaptive video surveillance to achieve high-precision multi-target tracking method.The experimental results have shown its outstanding performance.