BLOCK ADJUSTMENT OF MULTISPECTRAL IMAGES WITHOUT GCP AIDED BY STEREO TLC IMAGES FOR ZY-3 SATELLITE

Abstract. Multispectral images are the main data source of optical satellite remote sensing application. Consistent geometric accuracy is the basis of image registration and fusion. But the multispectral images usually collected by the nadir imaging camera are of weak geometric intersection, which will lead to the rank defect of the adjustment equation when performing block adjustment (BA) directly, making the solution unstable. Thus, a planar BA aided by additional digital surface model (DSM) which can overcome this weak geometry is often used to improve the geometric accuracy and consistency of regional planar images. However, the inevitable elevation error and the indispensable ground control points (GCPs) make the method limited in practical application. In this paper, a new method aiming to the BA of planer multispectral images of ZY-3 satellite without use of GCPs is presented. This method introduces the constraint of stereo images to assist the BA of planar images. By configuring appropriate weights of different observations, the integrated optimization of positioning models of planar and stereo images can be achieved together. The effectiveness of the method was verified by the planar multispectral images and stereo three linear camera (TLC) images collected by ZY-3 satellite. The satisfactory results indicated the rationality and effectiveness of the presented method.