The Aerial Triangulation Of Oblique Images Based On Camera Constraints

In recent years,with the development of aerial photogrammetry technology and sensor technology,the images obtained by oblique photography have been the main data source of 3D digital city and digital map update,which is an important means of spatial data acquisition.The oblique photogrammetry overcomes the limitations of traditional photogrammetry which can only obtain the texture of the roof and local information,and can acquire high-resolution images of the facedes of the building and the texture of the roof,which is widely used in the wisdom of the city,urban planning cultural relics protection,the collection and management of cadastral information and other fields.The aerial triangulation of oblique images which is as a key process in the production process has been the focus and difficulty of the study.The aerial triangulation of the traditional photogrammetry is no longer applicable in the oblique photogrammetry,because of the disadvantages of the geometric deformation between images,the larger changes of ground resolution and the difficulty of image matching caused by the different shooting angle of the cameras.Based on the existing research,this paper researchs the oblique images matching and bundle adjustment.The main contents of this paper are as follows:(1)In order to solve the problem of matching between oblique images with different angles,such as large rotation,deformation and large changes of ground resolution,this paper studies the ASIFT algorithm and proposes an oblique image matching algorithm based on camera constraint.(2)Aiming at the problem of large amount of oblique images data,the topological relations between images are established,which can be predicted whether the matching is performed by analyzing the overlap degree between images.The GPU acceleration can be used in the geometric transformation,the feature extraction and other aspects.(3)Aiming at the situation that the accuracy of the UAV pos(especially the attitude angle)is not high,a two-step aerial triangulation strategy is proposed.Firstly,the exterior orientations elements of vertical images can be obtained by the traditional aerial triangulation.And then the rough exterior orientations elements of oblique images can calculated according to the relative position parameter constrains between the cameras.Finally,the multi-camera bundle adjustment is be carryed out.In order to verify the feasibility and accuracy of the proposed algorithm and the process,the rotor UAV oblique images data of the low-precision pos system and theAMC580 oblique images data of the high-precision pos system are experimented.The results show that the algorithm and process proposed in this paper are feasible and can obtain the high accuracy of exterior orientations elements and the encryption points coordinates,which can be used as an effective method of the aerial triangulations of oblique images.