Aiming at the detection and tracking of multiple point targets distributed in cloudy background

a new method

namely the multi-stages trace association is presented.In the preprocessing

a module based on local auto-adaptive contrast threshold is given to remove the slowly varying clutter and at the same time

the noise with high intensity and the edge of clutter are filtered with a bi-directional high order correlation filter.When constructing deep-searching trees the adoption of searching strategy from both directions can be pruned earlier.In the post processing the energy along the candidate trajectories are accumulated and it can cut the false traces efficiently.Moreover

for the spatial-temporal consistency and continuity of the targets it can lock the true targets as soon as possible.The experimental results show that it can detect and track dim overlapping point targets with arbitrary trajectories accurately.It is advantageous over the traditional correlation algorithms in searching space and time complexity.