The Study Of Key Technology Oriented Data Processing In Production Of UAV

Unmanned Aerial Vehicle(UAV) remote sensing technology can remedy detects of satellite remote sensing and aerial remote sensing. It has been widely used in the fields of resource investigation, environmental monitoring, new countryside construction, maps updater. But compared with the traditional aerial or aerosapce photography, It is very different in platform, flight altitude, sensor. The UAV image data obtained by these effects has difference with the traditional, such as small coverage area, large rotation angle, inconsistent route, large amount, irregular overlap etc, this impacts of process internal data. In order to improve automation, efficiency and accuracy of data processing, thesis tries to study it from data source, image matching, aerial triangulation etc. The main issues are as follows:1) UAV image data quality mainly depends on the quality control in the process of image acquisition.UAV flight environment directly affect the choice of the flight platform. This paper according to the characteristics of the plateau region to verify the importance of environmental selection of flight platform; This paper takes The UAV image rotation angle, drift angle, pitch angle value as the quantization standard of image data quality to study the displayed status meeting the requirement of rotation angle, drift angle, pitch angle; Non-metric digital camera as the UAV using,this paper describes the concept as well as the correction mode, and explain the necessity of distorting error correction for non-metric camera.2) There are some shortages in traditional image matching algorithm, and single use cannot meet the current requirements of data processing. This paper according to the analysis of the Forstner and Harris' s characteristics. Obtained the advantages and disadvantages of the two kinds of operators, and according to the features of UAV data,which based on Forstner + SIFT operater combined with matching strategy to prove that the proposed method about this paper has high applicability, and eventually throughed experiment.3) Relative orientation and model connection are the key steps of the aerial triangulation. This paper tries to put forward the optimization model and thecombined matching strategy to improve automation of the relative orientation by analyzing the reasons for the relative orientation failure. And then propose the intermediate image model connection as a reference image, using inverse distance weighted method to assign connection error, decrease image distortion by analyzing the model connection film like distortion.Image distortion correction, automatic relative orientation and model connection of the UAV images in this research have been applied to the Tibet data processing by the new PixelGrid, which is used as a part of UAV data processing, showing a better data processing effect, and overcome the problems in the past.