Image Processing Of Unmanned Helicopter Remote Sensing

Photogrammetry has entered the stage of digital photogrammetry from the stage of film photogrammetry, and digital photogrammetry has become an important method of acquiring spatial and attribute data. The unmanned aerial remote sensing system is to make the low altitude measurement and photography through the unmanned aircraft platform. We use the unmanned helicopter as the remote sensing platform. It is developed by the Engineering Research Center for Nanophotonics & Advanced Instrument, jointly with SL Aviation (SLA) Technology Company. This helicopter has the advantages of safety, flexibility and cost-effectiveness.The unmanned helicopter remote sensing system is different with the traditional aerial photogrammetry system, with which the captured images have some problems, such as small format images(use traditional CCD camera), excessive quantities of images(at most, more than 20 images per route), over tilt angle of images(there is perplexed external environment, so the helicopter may not fly steadily), irregular overlap(the image may be captured form different heights, so the formats of the images differ very much)etc. These problems bring a whole series of trouble into image matching, image orientation and some other related image rectification operations. It is also the problem with the Digital Photogrammetric System that both national and international researchers are facing. Only overcoming these issues can realise the value of the unmanned helicopter remote sensing system.To solve the existing image data processing problem, the dissertation has studied the following subjects:(1) We reviewed the international leading edge techniques about unmanned helicopter remote sensing systems, summarized their importance in the future, analyzed all of the characteristics of helicopters , and found the reasons with these problems. We also decided the research methodolgy.(2) Through the analyses of all kinds of camera, we decided the camera that is suitable to acquire spatial and attribute data. Then, from the camera, we found the geometry distortion mode of the photogrammetry image and the way to rectify it.(3) We analyzed the image distortion caused by flying conditions and the helicopter itself, and summarized the traditional image processing methods such as the automatic geometric rectification model, collinearity equation model and polynomial model. (4) We assessed the accuracy of these algorithms in Matlab, and in the Visual Studio2005. We analyzed the running time of these algorithms. Based on these comparative analyses, we got the right way to process photogrammetry images. We also analyzed the parameter needs in the algorithms.(5) To process photogrammetry images in batch, we analyzed the difficulties in image matching, and gave a simple way to match images which are suitable to the unmanned helicopter remote sensing system.