The characteristics of dim point target during infrared image acquisition is analyzed, aiming at detecting them under a very low signal-to-noise ratio (SNR) and strong background interferences. Based on spatial distributions of targets, background and noise, a novel spatial matching filter is proposed to raise SNR from original low level. This spatial matching filter is optimized from 1D version considering the diffraction effects of targets. Then a novel morphology is presented to reduce background and enhance target saliency. A threshold selection algorithm is adopted to extract dim point targets adaptively. In all three stages, diffraction effects of targets and intensity difference of targets with background are taken into account to improve detection performance. Experimental results prove that under a low SNR (≤2), the successful detection rate of proposed method is over 95% with 10-5 false detection rate.