An ever-increasing number of remote sensors have necessitated data integration among different sensors. This is due to more gained benefits from differences in spectral resolution, acquisition time, and so on. With reference to primitive method of image registration, an image-to-image space is commonly mentioned for the process of registration when multiple images are considered, while image-to-object spaces is namely used as orthorectification, which is relied on DEM, camera parameters, and images configuration. This method is claimed as complex and needed for more carefully on processing. When the heterogeneous area is considered, it causes more problems on matching. Moreover, it is uncomfortable when the process is lacking of automatic procedure.

This paper, therefore, introduces improved model for automatic image registration with respect to image-to-object spaces based on edges and corners consideration. Polynomial transformation was mentioned as a major tool for transforming all tie points from image space to object space. Then accuracy was reported for whole area and for checked points. According to transformation, the RMS residual of whole area was approximately two pixels regarding systematic error, while efficiency index was about 0.99 at checked points. By improvement of procedure, edges and corners play an important role. Existing operators generated both features, then the redundancy of them was considered. Finally those redundant features are employed for improving output’s accuracy.