Study Of The Key Technology Of Low Altitude Photogrammetry Of Unmanned Aerial Vehicle And The Practice Of Large Scale Topography Formation

During the period of modernization, to complete different measuring scales and types of the highefficiency and high-precision plotting method is needed because the mapping task is heavy. Photographicsurveying surely will play a very import role in it. Surveying and mapping workers always dream to doengineering survey by using non-measurement digital camera and a set of automatic and low altitudephotographic surveying system which could satisfy the need of high-accuracy survey. Quick to obtain theresult, low in operating cost and better to handle are the characteristics of low altitude photographicsurveying. Low altitude photographic surveying system is an aerial photography system. In the system,carrying digital cameras, the unmanned aerial vehicle flies to catch regular overlapping images with theflight altitude lower than1000meters.According to the actual requirement, by analyzing the unmanned aerial vehicle made by the writerhimself, this paper deals with the study of the precise attitude control, the multi-GPS navigation, the cameraonline calibration, the quick matching of images, etc. Proven by the practical test, the system could meetthe requirement of large scale topography formation.The innovation points of this paper are as follows:1. The basic of the system is a stable platform with double gyroscopes. With the precision of0.5, oneof the two gyroscopes is placed on the platform to determine the attitude value of the platform; the otherone is arranged on the camera to determine the attitude value of the camera. The gyroscope on the platformis used to control the camera by determining the attitude value; the gyroscope on the camera is used toobtain high precision POS data. The purpose of using two gyroscopes is to do difference correction toeliminate the impact of the magnetic floating. If one gyroscope breaks down, the other one could amend theresult in accordance with related method of calculation.Considering the error the gyroscopes might have during the measuring process, the filter of Kalmanwhich is based on the time series algorithm is adopted to eliminate the random error while usinggyroscopes. Also the subsection interpolation compensation method is used to eliminate the scale factorerror of the gyroscopes. Because of the improvement of the precision and reliability of the gyroscopemeasurement, the precision of the attitude control theory of the aerospace camera reaches1degree.Influenced by the effect of the mechanical behavior, the actual measurement precision achieves to within2degree.2. The multi-GPS navigation system. To increase the precision of the location of the navigation GPSto get high-precision location information, only4navigation GPS are located on the flying platformbecause the unmanned aerial vehicle is with low load. During the flight, the flying platform is swiftlymoved, but the4GPS on the platform are relatively static with known relative position and distance.Taking the coordinate value the GPS get as the original value, taking the relative distance as the observationlength, there is an approximate equilateral triangle, which makes a triangulation network with other threeisosceles triangles. By using free net adjustment methods to adjust, the high-precision coordinate value is obtained to navigate and locate. Intersite interpolated with the heading value from the gyroscopes, thesimple intertial navigation system is formed.Adopting the multi-GPS navigation system, the coordinate of the flying platform could be real-timedetermined and the precision of the actual measurement instantaneous positioning could reach to about1meter.3. Online camera calibration technique. At present in order to do camera calibration there are specialcamera inspection places, which are usually set outdoors or on the top of high building. The numbercollecting is done by the camera while it is outdoors and collecting the data. But the aerial video cameraworks on the flying platform, which is different from working on the floor. The camera calibrationtechnique in this study is placed the camera directly on the flying platform, flying to the camera inspectionplace, shoots, collects data and calibrates. The data got in this condition is in accordance with the data thecamera collects during the work.Azimuth element calibration in the camera measures the space coordinates of the large number ofcontrol points which are set on the ground to form the testing ground. By calculating the data including thegeometric calibration laboratory data, photo coordinates, control points space coordinates, shootingposition coordinates while shooting, and the attitude angle of the camera, finally the inside azimuth elementis obtained. By the way of space resection and block aerial triangulation, the shooting position coordinatesand attitudes are calculated when shooting, which could improve the exterior orientation element precisionand the geometric correction precision.The main method to get the resolution ratio of images is to place resolution targets in the wild whiledoing aerial photographs; then analyze the resolution target images of the testing ground to obtain theresolution ratio of the images. And by calculating MTF of images to get the resolution ratio of images, thatis to say, by using the MTF from the images to sample the targets, calculate the edge spread function andpoint spread function and calculate MTF according to the edge method and pulse method.4. The quick image matching technology. Since the unmanned aerial vehicle’s characteristicsincluding large amount of data, relatively smaller size of the images, large number of the images, this paperuses SIFT to write the quick image matching program based on POS data and SIFT according to thehigh-precision camera central coordinates and attitude data from the multi-GPS location and doublegyroscopes attitude control system. By using the program, the image quality is checked while flying andthe orthograph map could be quickly got.Firstly the image sequence is set up and the coverage of the adjacent image can be calculated bydetermining the initial visual difference with the POS data. Then the characteristics of SIFT operators in thecoverage of the adjacent image is picked up and matched. The results from the former process are judged totest its dependability according to the error equation of continuous image pairs of relative orientation.Finally the mismatched points are got rid of.