Accuracy Analysis Of Automatic Registration Of Airborne LiDAR Data And Digital Aerial Image

So far, the analytic theory of photogrammetry has been basically completed, is one of theimportant means of surveying and mapping production unit to produce maps. We can getimages’ orientation elements and restore images’ position and attitude through a series ofaerial photography measurements processing, then the production of digital elevation model(DEM), digital line map (DLG), digital orthophoto map (DOM) and3D city model. Theairborne LiDAR (Light Detection and Ranging) system can quickly acquire the large surfacearea of discrete point coordinate data and other information. With LiDAR point cloud densityincreasing, the precision becomes better; the application of LiDAR is likely to cause anotherrevolution measurement technology. Target’s three-dimensional spatial information can bedirectly obtained by laser scanning, classification information is also relatively easy to obtain,and however, it is difficult to show the texture information of the target surface. If only byusing digital image to restore target’s geometric information, it requires complex dataprocessing, including image feature extraction, image matching, image resection,triangulation and texture mapping etc. In the course of3D information restoration processing,laser scanning data and aerial image data each has advantages and disadvantages, cancomplement each other. On the one hand, aerial image data can make up for the defects ofLiDAR data on the performance of image texture. On the other hand, the efficiency of image3D information restoration can be improved by using LiDAR data.Based on traditional Aerotriangulation, this paper gives research according to POS(position and orientation system)-supported Aerotriangulation. Firstly, tie points of aerialimages with POS are extracted automatically supported by LiDAR data. Secondly, in order toobtain good photo control points, multi-images matching point cloud are obtained based onobject-space constraints in part area according to the plan of setting GCPs during aerialphotogrammetry. Thirdly, matching image point cloud and airborne LiDAR point cloudthrough ICP algorithm, good registered points are chose as control points for bundle blockadjustment, and the orientation elements of images are carried out, and the registration ofaerial images and LiDAR data is realized. Finally, the3D coordinate of LiDAR check pointsare obtained through professional LiDAR measurement software, meanwhile the3Dcoordinate of aerial image tie points are obtained accurately in stereoscopic measurementmode, then RMS Error (Root Mean Square Error) is applied to analyze accuracy of automaticregistration, and the accuracy evaluation of registration of aerial images and LiDAR data isrealized.Some key points and innovation suggests involved in the process of registering andaccuracy evaluating are as follows: (1) Aerial image matching in the condition of object-space constraints via POS andLiDAR data;(2) some aerial images are taken as control ones for multi-images dense matching underconstraint of LiDAR and initial orientation elements, and gets the three dimensionalcoordinates corresponding to the feature points with intersection. So, the problem ofregistration between aerial images and LiDAR data is turned to register image matchingpoints to LiDAR points;(3) The ICP algorithm is applied to the registration of aerial image matching points andLiDAR points;(4) Registration accuracy evaluation;(5) This paper puts forward to use good registration points as the aerial image controlpoints instead of manual selection of ground control points, and improves the measurementaccuracy of control points, also the image orientation elements calculation precision.This paper solves the automatic registration problem of aerial images and airborneLiDAR data, and the quality of automatic registration is assured through registration accuracyevaluation. The automatic registration of aerial images and airborne LiDAR data is one of thekey technologies of integration of digital aerial photogrammetry and LiDAR, it is propitiousfor a more extensive application of aerial photogrammetry and airborne LiDAR technology.The content of this paper is integrated into the software of airborne LiDAR data processing(such as: LiDARStation.Pro) and aerial image data processing software (such as: PixelGrid)in some surveying and mapping production units, and it not only improves the efficiency ofdata processing, but also achieved certain economic benefits.With the development of economy and technology, there is a higher requirement forsurveying and mapping in society. It becomes a new demand for related departments ofsurveying and mapping to farthest take advantage of aerial photogrammetry and airborneLiDAR technology. The research of this paper is the premise of integration of digital aerialphotogrammetry data and LiDAR data, which provides theoretical basis and technicalreference for obtaining digital products of higher quality, especially the true orthophoto and3D city modeling.