Research And Application On Geometric Correction Methods Of Remote Sensing Image

Remote sensing image is the electromagnetic radiation record of surface features' high altitude satellite observation, two-dimensional projection of surface features' Three-Dimension info and comprehensive Description of surface features'spectrum, space and time info. Geometrical distortion and projection point drift would happen on account of satellite upper air orbit, high speed, changing angle of photography or other factors during the imaging. A better application preparation for image fusion and image classification require geometric correction.This dissertation do some deeply analysis and studies in remote sensor imaging model, geometric correction theory, factors of geometrical distortion, methods of choosing ground control point, analysis windows-based methods for control point collection, with one image automatch algorithm, and design and achieve a geometric correction system. Here are the main achievements of this study:Adopt the mentioned windows-based new collection methods for control point, Regarding the problem we can do nothing which coming out from presently commercial software's error when it has been making image automatch during control point collection, We develop this new methods which can combine the automatch with artificial work, so we can adjust the control point's position manually if the error go beyond allowance range after choosing point based on image automatch. The new methods match the merits of automatch and artificial work was confirmed by simulation experiment in accuracy enhancement.Adopt the automatch algorithm based on the cross-correlation of gray's normalization and cross-correlation edge normalization also, we put the unchanged centroid point as control point into resolving after success automatch, can reach a higher accuracy.Based on the platform of VC 6.0, adopt the MFC,OpenCV,HALCON during the development, we design and develop the remote sensing image geometry correction system, with main function of image access, control point choosing, Polynomial geometry correction, and confirm the mentioned new collection model and image automatch algorithm.