Geosynchronous satellite with big inclination is not completely stationary relative to the ground. Its sub-satellite point moves periodically in the north-south direction with ground track of shape like ``8''. The larger the inclination angle is, the larger the range of motion is. Such effect makes the effective exposure time of Geosynchronous satellite limited when observe it by optical telescope with stare mode, and the SNR (Signal Noise Ratio) of Geosynchronous satellite image could not be increased with long exposure time. In the monitoring of Geosynchronous orbit without changing hardware, a method called sub-pixel translating superposition of successive frames of images is proposed. By translating and aligning images in the sub-pixel scale according to the moving speed of the object and the time interval of the adjacent frame images, the positions of the Geosynchronous satellite in the image sequences are coincident, and then by superposition of these translational images, the SNR of moving Geosynchronous objects and the detection capability of the whole system can be improved. Results of superposition of real observed images show that the method can remarkably improve the SNR of moving. When five images are superimposed, compared with the SNR of the original image, the SNR of the superimposed image by integer pixels is increased by around 1.7 times, and the SNR of the superimposed image by sub-pixel is increased by about 2 times.