An infrared dim and small target detection algorithm is proposed based on adaptive Two-Dimensional LMS (TDLMS). Based on the traditional TDLMS algorithm, the improved TDLMS algorithm can evaluate the difference between the background and the template by using template compression and the hollow method, and adjust the iterative step size adaptively to carry out background filtering. And then a differential operation is performed to the background image for target detection. The experimental results show that the improved TDLMS method has a higher detection rate than the traditional TDLMS method.