Study On Feature Extraction And Matching Of Oblique Stereo Images

Since the outbreak of the information revolution,the information storage,transmission,presentation has undergone a qualitative change,which directly leads to the rapid progress of geographic information in data acquisition,processing and visual display.Aerial photogrammetry has become an important technical means for the acquisition and updating of the original spatial geographic information.Oblique photogrammetry is a major breakthrough in aerial photogrammetry.By carrying out multiple sensors on the same flight platform,it is possible to obtain a tilted image containing vertical texture information of a ground object while acquiring a vertical image.In the high-precision three-dimensional model generated by the technology,you can achieve the height,length,area and other information directly measured.Aerial oblique photogrammetry technology has a great advantage over traditional photogrammetry techniques,but the difficulty of data processing is improved by an order of magnitude: the oblique images has large affine deformation,serious perspective distortion and other problems due to the large angle of inclination;the large changes of resolution and the large-angle rotation,mutual shelter and other issues in the image is difficult to deal with;Aerial oblique photometric instruments also carry a large number of data redundancy.The existence of these objective conditions put forward a high demand to the image processing technology,and aerial image processing algorithms need to be redesigned.To solve the above problems,this paper studies the feature extraction and matching algorithm in digital image processing,and tries to solve the problem that the oblique image is difficult to obtain the same name due to deformation.The research work and innovation of the paper are mainly reflected in the following aspects:1.According to the research object,this paper introduced the basic knowledge of feature extraction and matching in oblique photogrammetry and digital image processing.This paper also focuses on the characteristics of oblique images and the limitations of the current popular feature extraction and matching algorithms.2.After studied scale-invariant feature transform algorithm and the maximum stable extremum region algorithm,this paper presents an SMSER algorithm combining MSER and SIFT-describing the affine invariant regions extracted by MSER using SIFT descriptor after fitting these regions,then matching images according to the obtained SIFT descriptor.Aiming at this method,the corresponding experiment is set up in this paper,which proves that the SMSER algorithm has certain effect.