Research On Image Matching Algorithm For Valley Topography Photogrammetry

With the continuous development of science and technology,oblique photogrammetry technology,as a new high-tech in the field of international surveying,mapping and remote sensing,has made rapid developments in remote sensing image data acquisition,processing and visualization technology.Tilt measurement technology acquires image maps by mounting multiple sensors on the same flying platform to acquire the image map at different acquisition angles,such as vertical and tilt,to obtain more complete and accurate image information,which has become a new acquisition of topographic maps.,An important means and method of updating the geographic information database and constructing a three-dimensional city model.In some areas with a large number of valley terrain areas,due to the continuous development of social technology,people's use of land and increasing attention to natural disasters require more accurate terrain data.Compared with traditional photogrammetry,In areas where there is a lot of valley terrain,the advantages of tilt photogrammetry are more obvious,and it is possible to more accurately obtain the accurate information of the terrain,but due to the huge amount of data obtained,the difficulty of data processing has also increased;at the same time,tilt photogrammetry itself During the shooting process,due to the large tilt angle,the captured images and ground features have a large affine transformation,severe perspective distortion,and occlusion of objects;and because the terrain features of the valley are not obvious during the later data processing,so As a result,the image does not match properly.In view of the above problems,this paper mainly studies the feature extraction and matching algorithms of digital image images.For the problems of valley terrain image features that are not obvious and difficult to match correctly,by studying the existing feature extraction and feature matching algorithms,this paper proposes a The image matching scheme for valley terrain is mainly as follows:I.Analyze the importance of feature points in the matching process,and summarize the methods and processes of image matching preprocessing.From the three aspects of image matching primitives,image matching constraints and matching methods,The basic principles and knowledge are introduced.Analyze the key technologies in image matching to further clarify the research purpose of this paper.2.Focus on the study of scale-invariant SIFT operators,and analyze the four feature point detection operators Moravec,SUSAN,Forstner,and Harris,which are excellent in feature extraction,and the algorithm for maximum stable extreme value(Maximally Stable Extremal).Regions,MSERs),and through further experiments,the performance of each operator is evaluated,providing a reference basis for the algorithm in this paper.3.Combining the characteristics of valley terrain,based on SIFT operator,using different extraction operators and SIFT algorithm,and experimenting with the combined algorithm to determine a feature extraction for low-altitude photogrammetry of valley terrain An algorithm with matching.The experimental results show that the algorithm combined with Harris operator and SIFT is more reliable and stable than other algorithms combined,and has certain practical value.