To compensate the image motion and improve the image quality, it is necessary to accurately detect the displacement vector between images. Based on the principle of image motion vector measurement with joint transform correlator, an improved method is proposed. To detect the edges of low altitude images, the improved self-feedback fuzzy edge detection method is presented to realize the input image preprocessing. By eliminating the 0th diffraction peak, the coordinates of relative correlation peaks are extracted with centroid method, and the high-precision detection of relative correlation peaks is achieved. The measurement accuracy and performance of proposed techniques are analyzed with simulation test and real low imaging experiments. Simulation tests show that the detection accuracy can be controlled within 0.03 pixel, when motion blur in the range of 0~10 pixel, Gaussian noise is less than 0.002 and the displacement between image in 0~20 pixel. Experimental results further show that the measurement accuracy of the proposed method reaches 0.2 pixel, and it can detect the image displacement vector of low altitude images with high accuracy and robustness.