With frequent occurrence of earthquake disaster, earthquake monitoring becomes increasingly concerned. Global
observing by optical remote sensing is an emerging technology widely applied in monitoring temporal changes of
topography in earthquake. It provides advantages of large width of observation, fast data acquisition and high time
effectiveness. This technique takes advantages of accurate image registration of pre-seismic and post-seismic to spot
surface rupture zones. Therefore, the spatial alignment accuracy of multi temporal images becomes a problem that hinder
the earthquake monitoring. Considering the adverse impact of different imaging angle, camera lens distortion and other
factors on image registration, a new approach of high accurate registration based on constraining positioning consistency
in rational function model (RFM) is proposed. Ziyuan3 images of Yutian country in Xinjiang are used to perform the
earthquake monitoring experiment. After applying the proposed method, registration accuracy of pre-seismic and postseismic
images is better than 0.6 pixel; surface rupture zones caused by earthquake are acquired promptly.
Stereo adjustment is the key to stereo mapping of domestic optical satellite image. In recent years, the most common satellite data used for stereo mapping was the along-track stereo pairs from the tri-linear CCD mapping camera in domestic. With the improvement of satellite mobility and the measurement accuracy of orbit and attitude, using the single line array CCD camera, by side sway, to obtain across-track stereo pairs becomes possible. In this paper, the YG- 24 satellite panchromatic very high resolution (HR) images were used to obtain the across-track stereo pairs. Based on undistorted RFM, the stereo pairs were processed with affine transformation model under the condition of different number of ground control points (GCPs). The result shows that, the stereo positioning accuracy of bundle adjustment can reach 35m without GCPs; while the root mean square errors (RMSEs) of plane is 0.7m and the RMSEs of altitude is 0.9m with GCPs, which can meet the secondary accuracy requirements of 1:25000 mapping.
Owing to relatively simplistic domestic hardware technology and a lack of on-orbit geometric calibration, particularly interior calibration, the positioning accuracies of Optical-1 HR satellites can vary greatly depending on the presence of ground control points (GCPs). Without GCPs, accuracies are typically lower than 100 pixels, whereas when GCPs are plentiful, accuracy is higher than one pixel, demonstrating the potential for a large discrepancy between international optical satellites with the same image resolution. This study investigated a new method of geometric calibration for Optical-1 HR satellites. Experiments were conducted to demonstrate the positive effects on positioning accuracy achieved by the calibration method. After calibration by our method, a positioning accuracy of higher than one pixel was obtained with only a small number of GCPs, which is equivalent to the accuracy of advanced international optical satellites with the same image resolution.