Search

A direct georeferencing method based on coplanar condition and reverse correction for unmanned aerial vehicle (UAV) images is proposed in this research. The main contributions of the method are embodied in three aspects. Firstly, a dual-step georeferencing algorithm based on coplanar condition is presented, which uses the analytic method to obtain the space position information of the target by two steps of relative orientation and absolute orientation according to the overlap and coplanar characteristics of UAV continuous images. Secondly, based on the coplanar condition georeferencing algorithm, this method introduces the idea of reverse rectification, and the georeferencing accuracy is transmitted from the current frame to the back frame according to the overlap relation between frames, which can effectively improve the attitude accuracy of the back image. Thirdly, a mediumaltitude UAV is employed to collect large amount of images and other telemetry data. Then the method contrast experiments are designed and performed. The results are analyzed and discussed, and present quite encouraging.

With the huge development of Unmanned Aerial Vehicle (UAV) in both military and civilian aspects, the image that shoot by UAV is very essential for the subsequent application. However, because of the bad weather, the images are always blurred, so select the appropriate model to restore the image quality is a key technology. In this paper, we proposed two forms of non-uniform multi-scattering models, and in different environment constraint condition, we simplify the models into 4 different models. And then we use the experiments to verify the different characteristics in the UAV medium-altitude imaging process. Experimental results demonstrate the superiority of the multi-scattering models over others with regard to contrast, color, and the number of SIFT.