Autonomous interior orientation of aerial photographs with weak constraints

This paper proposes an autonomous approach to determine center location of the fiducial marks included in aerial photographs without precise geometric information and human interactions. The conceptual model based on two assumptions makes it possible to localize the exact center of the mark by considering the symmetric characteristics of the mark. This approach is composed of four steps: (a) determining the symmetric center of the fiducial area, (b) finding the center of the mark with unit pixel accuracy, (c) applying the self-diagnosis function, and finally (d) localizing the exact center up to sub-pixel accuracy. In the experiment using 284 samples of the marks, it was visually evaluated that the approach had resulted only one case of erroneous identification. This shows robustness and reliability of the proposed approach.