The research aims to offer a solution to raise the detecting SNR of coincidence counting in intensity correlation imaging method and to obtain a clear image of high-orbit satellite. A simplified model of coincidence counting in intensity correlation was built based on the Hanbury Brown & Twiss Effect and the photoelectric conversion semi-classical model. As well as the influence on the detecting SNR of coincidence counting caused by the time measurement error of detection equipment, the effect of coincidence window and the observed spectral shape was analyzed. Furthermore, an optimization on window function according to these observing conditions was offered. Finally, the coincidence window′s effects on detecting SNR with the Monte-Carlo method were simulated. The simulation result shows that, by practicing the process of optimized coincidence window in intensity correlation imaging, the detecting SNR increases 39.2 times more than that with traditional method. The detecting SNR and the image quality of dim target can be efficiently improved as well by the window optimization.