Comparative analysis and evaluation of geometric transformations on distortions for image rectification

Image rectification is an important process in remote sensing application. Uncalibrated aerial images acquired from a non-metric digital camera which carried by an aircraft normally contain lens and perspective distortions. Various geometric transformations have been used for image rectification. However, the efficiency of the rectification would be affected if the geometric transformation is blindly used. This paper studies the performance of different types of geometric transformation for aerial images rectification especially in reducing the barrel effect of lens distortion and perspective distortion of inconsistencies in flight attitude. A simulation test is conducted using grid images to examine four common used geometric transformations. The results are measured using the total root mean square error (RMSE). It demonstrates that geometric transformation with both global and local accuracy, adjust transformation in this case, has the lowest total RMSE. It also confirmed that lower order global transformation has limitation in rectifying images with complex distortions.