The photoelectric stabilization platform is a servo system installed on the moving carrier, which is installed on the ground or air detection device to track the ground or air targets. The problems of gyroscope measurement noise, phase lag and carrier motion causing interference to the visual axis are carried on research work. This paper takes the two-axis and four-frame vehicle-mounted photoelectric stabilization platform for border and coastal patrol as the object to study the frame structure and control principle of photoelectric stabilization platform and determine the models of the speed stabilization loop and position follower loop. By analyzing the power spectral density of MEMS gyroscope noise signal, the source of noise and its impact on the system performance are determined, a Kalman sliding mode filtering method based on dynamic model is proposed, the filtering effect of the system is tested by simulation experiments, and the statistical characteristics of the data are analyzed. In order to restrain the disturbance of the visual axis system, the lead-lag correction control algorithm is introduced, combined with the automatic disturbance rejection control to compensate the observed quantity, and the closed-loop bandwidth of the speed stabilization loop is analyzed by experiment when the dynamic and static performance of the system is optimal. The experimental results show that the mean square error of the measured data of Kalman sliding mode filter based on dynamic model can be reduced to 0.104°/s, and the signal noise after filtering is significantly improved. 15Hz~25Hz is the ideal closed-loop bandwidth of the speed stabilization loop, and the optimized control algorithm can effectively improve the closed-loop bandwidth of the speed stabilization loop.