Automatic correction of distorted aerial images and extraction of multiangular information

The extraction of multiangular information can be used in the field of inversion and ground target classification, etc. A method has been developed to correct distorted images based on obtaining registered data. We designed a series of steps to extract enough registration control point pairs. The distribution of control point pairs is homogeneous for the mapping function. Based on gray-valued corner detection the interest points can be extracted automatically. Then the interest points are matched for generation of RCP. The technology of local surface spline fitting is employed for solving the problem of correction of distorted images. In our study we use Advanced Solid-State Array Spectroradiometer (ASAS) airborne simulated images, Polarization and Directionality of Earth Reflectances (POLDER) airborne simulated images and the wide-angle airborne images of IRSA, CAS. The results show that the distorted images can be corrected automatically with high accuracy