Image Matching In Photogrammetry Exploration And Study

As the shooting time, shooting angle, the natural environment changes, the use of avariety of sensors and sensor's own shortcomings, so that images taken not only by the noiseimpact, and there is a serious distortion and geometric distortion of gray.Under suchconditions, the matching algorithm how to achieve high accuracy, high rate of matching thecorrect speed, robustness and immunity to interference as well as the parallel implementationbecome the goal.In the image matching system, the matching algorithm is essential.At present, thedomestic and foreign researchers on the image matching has launched a lot of research work,made a lot of image matching methods, and achieved good results.Generally divided into twogray-based matching and feature-based matching method two categories, based on graycorrelation matching algorithm is a treat in order to match the image pixel gray-scale array ofa certain size the window by one or a few similarMeasure sequential search matching method,the performance of these algorithms depends primarily on similarity measure and the choiceof search strategy.Feature-based matching method, the class method first from the pendingregistration of the image feature extraction, using similarity measure and a number ofconstraints to determine the geometric transformation, the transformation acting on the finalimage to be matched.Match the characteristics of common edges, contour, line, points ofinterest, color, texture and so on.This article details the ideological point of feature extraction, using Moravec operator,Forstner operator experimental analysis of each test image is obtained experimental results,also introduced the relevant sub-channel matching principle, the algorithm under test by threeGauss and laplaciantower structure of two images.Finally, a manual matching and automaticmatching combination of ideas, aerial photograph with a group of relative as the test object, toachieve the same tomorrow as the match point.This final chapter will briefly with C #. NETdevelopment of software installation and function.