基于预测的无人机视频快速匹配算法研究

针对无人机视频数据，提出一种快速逐帧匹配算法。该算法首先使用计算机在每一帧中均匀标记一定数量的点，接着利用视频相邻帧之间的特点使用绝对差值法完成同名点的寻找，之后引入内插理论建立预测模型，缩小了搜索窗口和匹配窗口大小，降低了算法时间复杂度。实验结果表明，该算法可以高效地完成相邻帧之间的匹配。

Fast Matching Methods for UAV Video with Prediction

This paper proposes a fast matching algorithm for UAV＇s video.Firstly，the algorithm use computer to sign some points which are uniform distribution on the image.Secondly，the algorithm take advantage of the adjacent frames characteristics to find the correspond points.At last，we introducing the interpolation theory to establish the prediction model to reduce the size of the searching window and matching window，make the time complexity lower than before.