A stabilization imaging system is proposed, and the feasibility of design and realization is verified by studying the robustness of gray scale projection algorithm for under-exposured images. The effect of under-exposure on the performance of gray scale projection algorithm is studied by synthetic and real video experiments. The results show that, even with 1/25 exposure time and certain random noise, a favorable estimation performance of motion vector is also achieved. The projection algorithm is proved to be a fast and robust digital image stabilization technique even for under-exposured images. The projection algorithm can be used in aerial photography and spatial remote sensors to obtain stable and high resolution images, thanks to its small computational burden and robustness for under-exposured images.