Aiming at problems that near space target is hard to detect and sampling period and measurement precision of multi-radars are different

a track-before-detect（ TBD） algorithm for multi-radar based on probabilistic grid and multi-dimensional Hough transform is proposed. Firstly

radar measurement data are transformed into probability values by using probabilistic grid. Then

primary candidate tracks are set up in multi-dimensional space through multi-dimensional Hough transform. At last

the number and tracks of the targets are confirmed by spacefusion method based on unscented transform. Simulation results show that this algorithm can realize effective for detection of dim targets under dense clutter environment.