Abstract: An improved algorithm using layered multistage hypothesis testing (LMHT) was proposed,aiming at detecting small and slow moving targets in the low SNR infrared image sequences.The targets’ moving parameters space is separated as to the minimum cumulative frames and their trajectories are detected based on the MHT.The precise moving parameters of the targets’ trajectories are achieved using iteratively searching We make simulation with the scenes disturbed by gauss white noise and clutter. The results show that the improved algorithm effectively captures the dim targets and improves the precision of the parameters estimating, especially reducing the computation and storage remarkably