In order to realize the long period gaze exposure imaging for low-light application, an exposure period algorithm was designed to match the attitude change of gaze tracking image. Based on the mathematical model of the satellite to ground real-time gaze tracking, the velocity matching exposure imaging relations was set up. Image motion was induced by three-axis attitude angle and control accuracy of attitude angular velocity during the low-light imaging exposure period produces, and it can be calculated by using the method of the Monte Carlo. And exposure period was analyzed to meet the satellite attitude control accuracy in the low-light imaging. Finally, an on-orbit test of adaptive exposure period of low-light imaging algorithm was carried out on the smart verification satellite. Results show that: the corresponding imaging exposure time is18 ms and 55.2 ms respectively when the control accuracy of satellite attitude is 0.08°,0.0088°/s and 0.04°,0.003°/s . The actual deviation pixel of target point in the process exposure at imaging is less than 1 pixel. This bias has little influence on the imaging quality and image signal-to-noise ratio.