Investigation On The Coregistration Method Of Lidar Dsm Image And Aerial Image

Airborne LiDAR is a new remote sensing technology in last decade, which has been widely applied in fields of basic surveying and mapping,3D reconstruction, digital city construction and so on. However, compared to the traditional remote sensing, airborne LiDAR is lack of object surface texture and institutional limitations on imaging mode; therefore, it has a certain theoretical and application significance on the research of the registration and integration of airborne LiDAR and the traditional remote sensing data.On the other hand, due to the different source images having their own limitations, the study which is based on the heterogeneous image registration and fusion,3D reconstruction technology has become a hotspot in the field and the difficulty of remote sensing. At present, there are three kinds of registration ways between LiDAR data and aerial image:the way based on the LiDAR intensity image, the method with LiDAR point-cloud data and the means by LiDAR DSM image. Due to the drawbacks of LiDAR intensity image itself, the accuracy of registration method based on LiDAR intensity image is poorer; although the accuracy of registration based on LiDAR point cloud data is higher, the algorithm is relatively complex, which limits applications of the registration method. Image registration method based on LiDAR DSM depth image is worthy popularization in different source images registration field, therefore this thesis mainly aims at registration method of LiDAR DSM image and aerial image. And the details are as follows:(1) This thesis firstly introduces the registration theory of LiDAR data and aerial image, such as the registration principle, the base problem of registration, registration method, and feature selection and transformation model of registration. Moreover, the present methods of the registration between LiDAR data and aerial image are summarized, and the advantages of registration method of LiDAR DSM image and aerial image are showed.(2) The theoretical frameworks which are based on a series of procedures, such as LiDAR point cloud data interpolating depth image, edge detection, linear feature extraction of building, are tested and verified through experiments, which provides a more complete theoretical foundation for the subsequent software development.(3) This thesis introduces photo pin-pricking in the traditional photogrammetric theory into the line feature extraction of aviation image, which ensures that more accurate linear characteristics of homonymous lines are extraced through the way of human-computer interaction, even though aerial image texture feature is against the extraction of building line features. These works ensure registration adjustment calculation to go well by the collinearity condition equation of the generalized points.(4) This thesis devlelops the software of the registration of LiDAR DSM depth image and aerial image by the collinearity equation on the software platform of Visual Studio2010.The experiment results of the registration of LiDAR DSM depth imagery and aerial image covering the same area by self-programming software are showed that the results of registration by our procdures and methods can obtain relative high registration accuracy and our researches have a practical value. Meanwhile, it means more reference values to the research on data registration between aerospace photogrammetry, InSAR and aerial photogrammetry with the development of high-resolution remote sensing.