An Evolutionary Strategy for Georeferencing and Blunder Detection of High Resolution Satellite Imagery

The georeferencing of High Resolution Satellite Imagery (HRSI) is an important task for various remote sensing and photogrammetric applications. During the process of georeferencing، Ground Control Points (GCPs) in the image have an essential role to obtain the georeferencing parameters via the adjustment method. Therefore، the appearance undesirable blunders can cause the fundamental problem for image georeferencing and influence in adjustment results. There are traditional algorithms for detecting the presence of blunders in a set of points، such as data snooping and robust estimation. In case of blunders or systematic errors appearance either in the mathematical model or in the observations to be adjusted، both of these methods have considerable problems for isolating them and avoiding their influence. In this study، a novel method based on evolutionary algorithms is described for georeferencing and blunder detection simultaneously and is analyzed versus the traditional approaches. Based on the proposed method، through least square estimation، by maximizing the residual on each individual check point، the estimated values of parameters are computed and blunders in control points are found. The novel method is implemented on IKONOS and WorldView2 images with respect to the different number and propagation of blunders. The results show that the genetic algorithm based proposed method is a more efficient method for detecting blunders than data snooping and robust estimation even when the cluster of blunders exist in GCPs.