Partial Distortion Correction of Rectification Error of Remote Sensing Images

At present,geometric rectification of remote sensing image mainly adopts the method of digital rectification in which control points are very important to image rectification quality.In actual work,the accuracy of control points is always affected by many factors that cause partial dis-tortion in rectified remote sensing image.Partial distortion error existing in rectified remote sens-ing images is not paid much attention to,and many remote sensing processing software lacks the a-FK(W21?40ZQbility to deal this problem. This paper presents a method to check and display the partial distortion error by overlaying remote sensing image and digital raster graphics (DRG) or digital line graphics (DLG) with the support of image roaming and scaling,which visually shows whether partial distortion exists or not in rectified image.To solve the problem on partial distortion,triangle transformation algorithm,a scheme of disassembling polygon into multi-triangles,the principle of triangle geometric distortion are introduced.In the experimental research,this paper selects aerial image and 1:10000 topographical map to match and to be rectified.First,digital raster graphics (DRG) is generated from 1:10000 topographical map rectified by ERDAS remote sensing image processing software,then 16 points are selected separately from aerial image and DRG,image rectified by polynomial equation in ERDAS with the reported mean error of 3.3887 pixels,image distortion error is checked by overlaying DRG with rectified image.In flat areas,the distortion can not be observed,while in mountain areas the distortion is obvious.With the software developed under Windows using simplified algorithm presented here,the partial distortion is corrected in two steps with different quad size.The partial distortion rectification is done in the environment of image overlaying,so it guarantees the correctness of the partial distortion.The partial distortion rectification is very fast,because the rectification area is small,and the calculation amount is also small as well.Results show that algorithm presented in this paper is feasible.Further research will be focused on overlaying image with image having different time phases and distortion rectification inside arbitrary polygon.