Development Of A Reconstruction Technlogy Of 3d Map Based On Pictures From Uav

UAV (unmanned aerial vehicle), especially the small unmanned helicopter with ability of autonomous flying ,has important value in the field of military, civil and scientific research.The basic structure of UAV contains 6 parts: the UAV body, navigation system, flight control system, wireless information communication system, ground station system and the mission-application system. The small helicopter"Skywalker"of Shanghai Jiaotong University has realized the front 5 systems above. In this paper, we make the small unmanned helicopter "Skywalker" as the main application object, innovate and realize the mission-application system of it.In the process of application and flight control, UAV need to perceive the obstacles ahead and the environmental information. Using the binocular cameras on-board to catch the 2D images, the UAV can reconstruct the 3D environmental information by taking use of the pattern recognition technologies and the three-dimensional reconstruction techniques. These 3D information can effectively supports the implementation of UAV tasks like: obstacle avoidance, path planning and flight control application. Therefore, the study of 3D reconstruction based on simultaneous 2D aerial images, has important theoretical and practical value.The main contents include the following 3 aspects: 1,Design and build the stereo vision system on the UAV based on the physical characteristics of"Skywalker"and geometrical principle of stereo vision. Use the UAV ground station's"aerial information processing module"to send the images to our computer.2,Do the image processing, detect the feature points of images. Then match all the detected feature points in the right and left images, in this step, we use the SIFT and SUSAN algorithm, realize the feature points matching process in MATLAB.3,Reconstruct the 3D scene of these images. Use the matched feature points, we calculate the 3D coordinates of them in MATLAB, then, reconstruct the scene in 3D.In this paper, our stereo vision system and the 3D reconstruction process achieved good experimental result. This technology is of great importance in UAV navigation, UAV investigation and UAV obstacle avoidance.