Satellite jitter detection and compensation using multispectral imagery

Abstract

ABSTRACTSatellite jitter is an important factor that affects the geometric accuracy of high-resolution satellite imagery. However, it is often ignored when there are no high frequency and accurate attitude observations. This letter proposes a jitter detection and compensation method using multispectral imagery, which can improve the band-to-band registration accuracy and eliminate image distortion caused by satellite jitter only using image data. In the method, a jitter detection and compensation model is built to estimate the periodic distortion from band-to-band registration error of multispectral imagery. Both simulation data and Ziyuan-3 (ZY-3) satellite multispectral images were used to validate the presented method. In the experiment of ZY-3, the band-to-band registration accuracy of the compensated image is improved to 0.05 pixels in both across and along track directions and the distortions across and along the track are both supressed to less than 0.2 pixels as determined from the known positions of ground control points (GCPs), showing that the distortion caused by satellite jitter can be accurately estimated and corrected by the proposed method.