Geometric correction techniques for remotely sensed image

Abstract

In the analysis of remotely sensed imagery it is necessary for the distorted image data to be corrected. For correction of the distortion, techniques which are utilizing GCP``s, and frequency characteristics under the linear transform are discussed. In the next chapter, the modified PRL technique which can be used as a post-classifier of a specified region and as a pre-processing technique for the correction of the distortion utilizing shape measures is covered. And at the last part of this thesis, the algorithm utilizing shape measures for the correction of distortion and for reconstruction of the original image from the distorted image data is suggested. Two methods to find distortion angles are suggested. That how much an image is distorted can be known is a great advantage of the method using shape measures and if there are "perspective distortion" in the image, the method is more useful.