The Technology Of Image Processing And The Monitoring Inspection Method Of Flight Quality For Unmanned Helicopter

UAV(Unmanned Aerial Vehicle) Remote Sensing system can neatly operate at low cost without perfect weather conditions, so it has wild application future in disaster monitoring, emergency surveying, local environmental monitoring, regional topographic mapping, and so on. Unfortunately, UAV platform often has weaknesses of flight attitude instability, limited load. And because of non-professional small digital cameras, acquired images distort with significant differences. Therefore, positioning accuracy and processing efficiency of images got by UAV system are limited by a certain extent. To solve these problems, this paper studies theories and methods of feature extraction and matching of UAV images. Also automatic inspection methods of UAV flight quality, distortion and geometric correction methods, DEM(Digital Elevation Model) extraction, orthographic images production are referred. As the results, a feature matching method based on box of triangle boundary is formed, automatic inspection and acceptance technical system of UAV flight quality is built, and UAV images are processed automatically. At the end, these methods or technologies are validated by AF25 B UAV images.The idea of triangle boundary box is adopted to extract or match features. And a matching algorithm of Harris feature point based on triangle boundary box is designed to eliminate mismatching points using RANSAC(RANdom SAmple Consensus) algorithm. Automatic methods of feature point extracting and image matching are suitably applied to UAV system. Then, these methods reduce necessary feature point numbers, shorten automatic processing times, and improve matching accuracy of point pairs. So this new technological way has been provided for fast and accurate matching of remote sensing imageries.This paper has formed an excellent automatic inspection and acceptance methodology of UAV flight quality, and its reliability is confirmed through actual cases. In this methodology, inspection and acceptance system of UAV images is divided into three cases: POS data, GPS data and only image data, according to load characteristics and auxiliary flight data, and based on in-depth analysis of UAV flight inspection and acceptance contents and methods. Different steps of inspection and acceptance are designed for these three cases. By introducing relative orientation theory, topological relationships of images are calculated out using relative orientation results. Later, formulas to calculating flight parameters in different cases are derived. Results show that inspection and acceptance methodology proposed in this paper has evaluated UAV flight quality quickly and accurately to dealt with practical needs.Moreover, technologies such as distortion correction, geometric correction, DEM extraction and orthographic images production are experimentally analyzed to build a general process of UAV image processing technology, and these technologies are validated by actual UAV images.