The Data Processing Of UAV Remote Sensing Data And Its Precision Evaluation

From the application of the satellite remote sensing image and large aircraft aerial image, to the use of small unmanned aerial vehicle image, the access technology of remote sensing has been gradually brilliant now. Since present small unmanned aerial vehicle used for aerial survey are all under the remote control, the choice of their flight attitude and route is particularly vital, to cope with the development trend of remote sensing, the technology of automatic unmanned aerial vehicle(UAV) aerial survey is underway. From now to the future in the field of remote, unmanned aerial vehicle will carry the burden of aerial survey. Compared with satellite and large aircraft aerial survey, unmanned aerial vehicle(UAV), with high intelligent degree, can get high resolution images and need low requirement for sites and environment. Therefore, obtaining low-cost aerial images efficiently is imperative. If small unmanned aircraft, with characteristics of competitive prices and increasing popularity, are utilized completely to conduct a comprehensive measure the territory in our country and we will get better result.Aiming at the above problems, the topic begin research work by using unmanned aerial vehicle aerial technology, which with the characteristics of large scale, wide angle, higher currency, in combination with computer visual modeling method, centering on an ancient city wall and the ancient terrain 3D reconstruction problem. In order to establish a set of convenient and quick, low-cost modeling method, this study mainly studied the following aspects.This study expounds the current research situations of the unmanned aerial vehicle remote sensing measurement technology and 3D modeling technology of its data, discusses the unmanned aerial vehicle remote sensing system and the ways of its data access. As well as analyzes some key technologies in processing unmanned aerial photography such as lens distortion correction, beam area network aerial triangulation, resampling technique and 3D reconstruction theory algorithm. Through the ancient rampart unmanned aerial vehicle measurement project, using Photoscan software to process the images, this study carry out a precise analysis of experimental results and get the expected results. Through the project of measuring the topography around ancient grottoes, this study designs three different schemes of image control point layout and conduct an error analysis to the residuals of image control point location and accuracy of check point through experiments, which indicates that using model machine for geographic information data acquisition can meet the accuracy requirement topographic map of 1:2000, compares with three point layout schemes, the root mean square error of a point of check point decreases successively, which shows that with increase of the number of control points, the results, precision improves on a certain degree. The control point location distribution is also an important factor affects the accuracy of the results. Combining with the experimental procedure, this study briefly summarizes the factors causing the error. Finally this study utilizes three methods: image enhancement, uniform color and tailoring treatment from ERDAS, Photoshop and ArcGis and gets the elevation and 3D model of wall.According to engineering application experiments, this topic have different degrees of defects which need to be solved, first, it lacks of oblique aerial photographs for aerial which influence the texture image and shape of the 3D model. Secondly, compared to conventional aerial, accuracy of the model machine is lower because of increasing work load of obtaining image control points, which effected by the competitive price and available POS data.