Lack Of Control Points Of The Uav Remote Sensing Image Geometric Correction Technology

Unmanned Aerial Vehicle (UAV) Low Altitude Remote Sensing System has been applied widely in various resources prospecting, disaster monitoring, etc, because of its advantages in flexibility, economy, high efficiency and real time monitoring. During the period of the earthquake relief work for 5.12 Wenchuan Earthquake, a great amount of high resolution(0.2-0.4m) remote sensing data from serious disaster region are acquired immediately by using UAV with several kinds of sensors. Because ground control points (GCPs) cannot be obtained for the bad geographic environment, frequently occurred landslides, collapses and debris flows and ruined roads after earthquake, the geometrical rectification cannot be performed through traditional polynomial fitting. Moreover, the final image mosaic effect and the quality of disaster information extractions are degraded seriously by UAV images features of small image range and large quantity and the heavy work burden and low efficiency in images preprocessing. These two respects impede severely the application of UAV remote sensing data. In order to let UAV remote sensing data reflect the geographic environment more realistic and be applied in disaster information extraction and analysis more efficiently, a UAV remote sensing geometrical rectification model without control points is need to established urgently.For problems existed in UAV low altitude remote sensing image, this thesis has researched mainly on the following work:The main factors causing the geometrical distortion of UAV remote sensing images are analyzed. According to the work environment of UAV low altitude remote sensing system, the influence on the accuracy of the whole image preprocessing from different error sources is discussed, so specific research is certified like the geometrical rectification on UAV images by using flight attitude parametersA geometrical rectification model of UAV remote sensing images based on different flight attitude is established. The influences on the relationship between projection points of UAV remote sensing images and realistic points, generated by kinds of flight attitude parameter variations like pitching angel, roll angel, yaw angle and flying height, have been discussed individually on the foundation of the imaging models of UAV remote sensing data. Comparing with the classical geometrical rectification of polynomial fitting, this function reduces the requirement degree of ground control points and enhances the speed and accuracy in preprocessing of considerable UAV remote sensing data.According to the established geometrical rectification model, an algorithm is designed and UAV remote sensing geometrical rectification experimental software is realized, which can be used in fast processing geometrical rectification for UAV remote sensing images on deprivation of ground control points and cooperated with various information about flight attitude provided by plane control system and can also provide powerful technique support for fast images mosaic and quantified information extraction afterward.The quality assessment of the geometrical rectification results of this model is performed through experimental comparison and analysis. The ground control points information are collected by Ashtech GPS survey system and the geometrical rectification of classical polynomial fitting on the remote sensing images of Pingle central region is simulated by ENVI software, which act the basis of the assessment. Comprehensive assessment has illustrated geometrical rectification algorithm can help to obtain high resolution remote sensing images with good rectification effect immediately and improve the quality, speed and accuracy of UAV remote sensing data processing and quantified disaster information extraction.