岷江流域无人机遥感影像拼接及灾害信息提取  

FAST MOSAIC ALGORITHM OF THE UAV REMOTE SENSING IMAGE FOR MINGJIANG RIVER AREA

作　　者：胡文艺[1] 王绪本[1] 

机构地区：[1]成都理工大学信息工程学院,成都610059

出　　处：《物探化探计算技术》2012年第3期340-343,241,共4页Computing Techniques For Geophysical and Geochemical Exploration

基　　金：国家自然基金项目(60841006);中央高校基本科研业务费(ZYGX2010J085);四川省青年基金(09ZQ026-058);四川省科技支撑计划(2009GZ0157)

摘　　要：针对无人机遥感影像的特点,提出了一种基于特征点和重叠度的快速自动拼接算法。该算法实现了用于计算参考图像和待配准图像之间重叠度的相位相关法,并提出只在重叠区域中进行特征提取和特征匹配的方法。利用岷江流域无人机遥感图像进行了实验,结果表明,该算法使图像拼接有效地防止了图像非重叠区域中信息对算法的干扰,提高了拼接算法的精度和速度。并利用拼接的图像对茂县至都江堰的岷江主干河流进行了灾害信息提取和统计分析,为建立无人机高分辨率低空遥感数据的快速处理和应急灾害信息提取具有重要的指导意义。A fast algorithm of automatic mosaic is presented for the unmanned aerial vehicle(UAV) images.A phase correlation method used to calculate the overlap degree between reference images and matching images is realized.And a new method to carry out feature extraction and match only within the overlap region is proposed,which reduces the computational complexity dramatically,prevents non-overlap image region′s data from influencing the algorithm and increases the accuracy of mosaic algorithm.The algorithm calculates the overlap degree between two images through phase correlation method,extracts feature points by using the improved harris corner detection algorithm,eliminates approximately feature points failing to match according to the overlap degree,and confirms matched points pairs based on similarity function.The paper obtained a seamless image by fusion splicing.Experimental results demonstrate that the algorithm is effective and valuable.

关 键 词：无人机 特征点 匹配 图像拼接 

分 类 号：TP7[自动化与计算机技术—检测技术与自动化装置]