期刊名称：Archiwum Fotogrametrii, Kartografii i Teledetekcji
印刷版ISSN：2083-2214
出版年度：2007
卷号：17
语种：English
出版社：Main Board of Association of Polish Surveyors
摘要：One of the works conducted in the Department of Photogrammetry and Remote Sensing Informatics,University of Science and Technology,Kraków,Poland included the development of an algorithm for aerial images matching. The algorithm utilises the function of an available OpenCV opensource library,which is used to process images. The application algorithm,which finds homological points in images that form a stereogram,utilises the OpenCV library. The operation of the algorithm consists in determining,in a particular part of the right image,the best template downloaded from the left image,by means of successive passing through the levels of the right and left image pyramids. The left image template is a buffer of a specific size,which depends on the pyramid level,around the point to be searched for in the right image. Part of the right image,in which the search is performed,is determined as a buffer whose central point is equivalent to the anticipated position of the homological point in the right image. The determination of the probable position of the point searched for is made based on the displacement value between images,provided that that this is the first iteration after the pyramid levels,or based on the point's position from the previous pyramid level,when the value of the matching coefficient is higher than the threshold set. The search starts from the smallest image of the pyramid,it is from the highest level. Then,it passes to lower levels,one after another,and ends in the original image. On each level of the pyramid,particular points are corrected,and results are ascribed to proper vectors. The value of a matching coefficient determines the size of the area in the right image to be searched through. The values of displacement along the x and y axes are updated on a current basis,with respect to points that demonstrate high matching coefficient (the minimisation of errors). The determination of a given template matching within the search area consists in calculating the value of a two-dimensional co-variance function (correlation or distance) for two-dimensional signals that the images are,and presenting them in a form of a matching map for each template position in relation to the search area. The determination of a best match consists in finding the maximum value of pixel on the map,and returning it as the point searched for. In order to improve the efficiency of the algorithm,tests were conducted along with the analysis of image matching parameters' selection. An optimum size of the template was determined in the course of research,and also the behaviour of the image matching coefficient was studied.
其他摘要：W ramach prac prowadzonych w Zakładzie Fotogrametrii i Informatyki Teledetekcyjnej AGH,powstał algorytm dopasowania zdjęć lotniczych. Wykorzystuje on funkcję dostępnej biblioteki opensourcowej OpenCV służącej do przetwarzania obrazów. Algorytm aplikacji z
