基于SURF特征的无人机航拍图像拼接方法

Image Mosaic Method of UAV Aerial Photography Based on SURF Feature

设计了一种基于SURF特征的无人机航拍图像拼接方法。该方法首先对无人机航拍得到的图片进行预处理,利用几何校正和辐射校正方法对原始图像进行校正,为后续图像的拼接工作做准备;其次,根据无人机飞行高度、速度以及搭载相机参数信息,使用加速鲁棒性的SURF算法对预处理后的图像进行特征检测;再次,利用无人机飞行状态参数等信息对无人机航拍图像配准算法进行研究和改进,使后续图像拼接结果更加准确可靠;最后,采用一种改进的加权平均融合算法对图像进行拼接。此方法不但可以降低图像融合难度,而且可以消除在拼接过程出现的接缝,生成高清晰、平滑过渡的全景图像。

A method of splicing aerial image of drone based on SURF feature is designed. The method first preprocesses the images obtained by the aerial photography of the drone, and corrects the original image by using the geometric correction and the radiation correction method to prepare for the stitching work of the subsequent images;secondly, according to the flying height, speed and camera of the drone The parameter information is used to detect the pre-processed image using the SURF algorithm with accelerated robustness. Thirdly, the UAV aerial image registration algorithm is researched and improved by using the UAV flight state parameters and other information to make the subsequent image mosaic. The result is more accurate and reliable. Finally, an improved weighted average fusion algorithm is used to stitch the images. This method can not only reduce the difficulty of image fusion, but also eliminate the seams appearing during the splicing process, and generate a high-definition, smooth transition panoramic image.