狭窄重叠的无人机影像拼接方法研究

Research on Mosaic Method of UAV Images with Narrow Overlap

无人机影像拼接是利用无人机影像对大范围目标区域进行监控的一种手段,拼接过程需要对单个图像相对于参考平面进行精确和稳健的几何校正,目前使用广泛的几何校正方法通常要求相邻图像之间具有很宽的重叠度。为此,针对无人机飞行速度非常快,所获取的影像无法满足宽重叠的情况,本文研究了一种快速、鲁棒性强的窄重叠无人机影像拼接方法。本文主要对现有的变换模型进行了深入的分析,基于位姿参数提出了优化模型,保证了几何校正的精度和鲁棒性。同时,提出了基于重采样网格的图像重采样方法,用以实现几何校正的高效完成。实验结果表明:本文方法能对狭窄重叠的无人机影像实现精准拼接,具有快速监测应用潜力,同时也能为处理不同重叠的无人机影像提供最优的解决方案。

UAV image mosaic is a means of monitoring a wide range of target areas using UAV images.The mosaic process requires accurate and robust geometric correction of a single image relative to the reference plane,but the widely used geometric correction methods usually require a wide overlap between adjacent images.Therefore,aiming at the situation that the UAV flying speed is very fast and the obtained images cannot meet the wide overlap,a fast and robust narrow overlap UAV image mosaic method is studied in this paper.The existing transformation models are deeply analyzed,and an optimization model is proposed based on pose parameters to ensure the accuracy and robustness of geometric correction.At the same time,an image resampling method based on resampling grid is proposed to realize the efficient completion of geometric correction.The experimental results show that this method can realize accu-rate mosaic of UAV images with narrow overlap,has the potential of rapid monitoring application,and can also provide the best solu-tion for processing different overlapping UAV images.