In order to measure the sub-pixel image shifts of a remote sensing camera, induced by vibration, scrolling, the orbit regulation of the satellite, and so on, a photoelectric hybrid joint transform correlator is used. Two adjacent frames of the sequential remote sensing images are correlated by the joint transform correlator, and the joint power spectrum is processed by the Laplace transform to obtain the sharp shrinking cross-correlation peaks, the coordinate of which is found by the centroid method, and the image shifts can be measured accurately. In this paper, the measurement principle of the image shifts based on the joint transform correlator is given first; then the shifted image sequence is obtained by interpolation and down sampling, and the measurement results are simulated. Last, considering the relationship between the components of the joint transform correlator, the experiment platform is built and the image shifts are measured. The results show that the image shifts can be measured accurately, and the measurement error is less than 0.075 pixel.
© 2017 Optical Society of AmericaPDF Article