The Research Of Epipolar Image Module Based On Digital Photogrammetry System

Throughout the development of the photogrammetry and computer vision technology,the format of the images processed increasingly diverse, and the degree of automationconstantly improves. Close photogrammetry and computer vision fields cannot avoid theproblem of image matching. Image matching is a bottleneck problem, which how to improvethe image matching accuracy and how to shorten the time required for image matching. In the1970s, since the epipolar line concept was put forward, epipolar correlation plays a key role inimage matching. Epiporlarline is the basic concept in the photogrammetry. In the stereocorresponding images, corresponding points must be located in the correspondingepiporlarline and corresponding points are one to one. So if we can determine thecorresponding epiporlarline in the stereo epiporlarline images, then may change2d matchinginto1d matching according to the nature of epiporlarline images, which greatly improves thecalculation efficiency and reliability.In order to facilitate the paper’s expansion as well as to the digital image processing,firstly designed a digital photogrammetric system framework, which includes creating project、reading images、reading and creating camera files and control point files. And then only forgenerating epipolar image module is discussed. There are some ways to generate epipolarimage, such as the method based on digital correction, the method based on condition ofcoplanar and the method based on Relativity Direct Solution. Because that the experiment ofimage data does not contain information such as the inner orientation element, and the formertwo methods are involved in calculation of collinear equation, so use the third method to do itby programming. And use Relativity Direct Solution for epiporlarline arrangement.Temporarily don’t consider the whole epipolar image generation, using only a single imagepoint for testing. The result reached a certain precision.The last part of the paper simply introduces the method based on projective geometry togenerate epipolar images. This method and the method in the photography measurement based on coplanar condition can be mutual transformated.But this part of the discussion is notas the focus of this article. The paper is an engineering applied article to discusses how togenerate epipolar image detailed, which can make foundation for the image matching.