Measurements of image shifts using an edge-extracted joint transform correlator

Abstract

Abstract. To measure accurately sub-pixel image shifts for the space cameras with high resolutions, an improved joint-transform correlation (JTC) algorithm is put forward, which combines fuzzy-edge detection and high-boost filtering. First we explain the principles of measurements of the image shifts based on the JTC. Considering that most of the important information is kept in and described by the edges of image, a modified fuzzy-edge detection algorithm by Pal and King is then used to extract the edge information of the input image. The joint power spectrum is subjected to high-boost filtering to enhance detail-containing image components and eliminate low-frequency noises. Finally, an experimental platform is built and the performance of image-shift measurements based on the improved JTC is studied. The results testify that the measurement errors are normally distributed, with a zero mean and mean-square deviation less than ~ 0.1 pixel, whenever the shift between two correlated images changes from 0 to 25 pixel. As a result, our JTC technique facilitates highly accurate measurements of sub-pixel image shifts.