Abstract: We all know that the exterior elements of images,which are taken by CCD from different orbits,are difficult to obtain through ordinary bundle adjustment process because between the unknown exterior elements there is strong correlation which makes error equation badly ill.Some methods are provided to overcome the problem but they cannot bring satisfied results.Under the situation the paper gives other algorithms to successfully solve ill error equation and get the real exterior elements. Firstly depending on the base principle for handling the images,an adjustment process in object coordinate system is designed and a separating-combining method is used in some correlate elements to modify the base model in the bundle adjustment in order to reduce the strong correlation between the exterior elements.Therefor,more stable result is acquired in the calculation of the exterior elements of remote sensing images. Another emphasize of the paper is designing a searching control solution based on the principles of AI,which separates the whole process into several layers with respective data resolution,and the global optimization is searched in each layer.It needs a fewer control points,removes the correlation between X_s and φ,maintains the RC and CC in a lower lever.The solution is a direct,simple and strict method. The experiments,based on practice data,have proven that the algorithms,proposed in the paper,are correct and efficient.