Abstract: An important, problem in the field of signal processing, sequence of infrared image detection which arouses peoples attention widely is the detection and tracking of dim moving targets in very low Signal-to-Noise Ratio (SNR) or signal-to-clutter ratio environment. In fact, dynamic programming is a problem of multi-stage decision, and Dynamic Programming Algorithm (DPA) is an effective approach which was adopted during the process of Track-Before-Detect(TBD). In current work, the mechanism of dynamic programming algorithm is analyzed, and a number of technical innovations that improve the performance of DPA are presented. Simulation results show this approach has high ability of detection for dim moving targets, and achieves 3~5dB SNR. gains comparable to those of other detection algorithms.