To achieve high speed, high precision, and device miniaturization, an improved joint transform correlation (JTC) method is proposed. In this method, the correlation input image is constructed by superposing the intensity of two images to be detected. To effectively separate autocorrelation and cross correlation peaks, the method makes use of the constant image motions resulting from satellite movement. The size of each image used for the correlation operation is reduced to the size of the image to be detected, which greatly reduces the calculation of the measurement process, which is beneficial for realizing high-speed digital joint transform correlation. The simulated and experimental results show that the new method maintains the advantages of high accuracy, high robustness to image noise and low dependency on image texture of traditional JTC. Additionally, it also has the potential for high-speed image motion measurements.