无人机近红外传感器数据匹配方法  

Research on matching algorithm of UAV infrared sensor data

作　　者：李迁[1,2] 甘拯 支晓栋[2] 刘玥[4] 王建超[2] 金鼎坚[2] 

机构地区：[1]中国地质大学(北京),北京100083 [2]中国国土资源航空物探遥感中心,北京100083 [3]长江空间信息技术工程有限公司(武汉),武汉410010 [4]北京航空航天大学自动化科学与电气工程学院,北京100191

出　　处：《国土资源遥感》2017年第1期86-91,共6页Remote Sensing for Land & Resources

基　　金：中国国土资源航空物探遥感中心对地观测技术工程实验室航遥青年创新基金项目"基于红外光学传感器的低空遥感应急监测方法研究"(编号:2013YFL09);中国地质调查局地质调查项目"地质灾害低空遥感应急监测方法技术研究"(编号:1212011120219)共同资助

摘　　要：随着无人机低空遥感技术的不断发展,其已成为一种重要的遥感技术手段。无人机传感器也从普通可见光向多/高光谱传感器发展,但鉴于无人机载荷小对传感器的限制,这些新型传感器数据质量较差,现有方法难以直接处理。因此,以无人机近红外传感器所获取数据为研究对象,基于尺度不变特征转换(scale-invariant feature transform,SIFT)匹配方法进行参数优化和粗差剔除,以解决无人机多/高光谱数据后期成图应用的关键技术,并对该方法进行实验验证。实验结果表明,通过该方法能够获取稳健的匹配结果,对提高无人机多/高光谱等新型传感器的应用效果具有重要价值。Low - altitude UAV remote sensing technology has become an important means of remote sensing technology. With the continuous development of the technology, its sensors have also changed from the visible ones to the multi/hyperspectral ones. However, due to the limitations of a small UAV payload on the sensors, the data quality of these new types of sensors is poor and hence it is difficult to deal with existing methods directly. Therefore, the authors studied the data obtained by UAV infrared sensors and then optimized parameters and removed gross errors based on the SIFT matching algorithm. This method has made robust matching results and can solve the key technology of the late mapping application of multi/hyperspectral data. The authors used a set of UAV infrared data to test and verify this method. The experimental results show that this method is capable of obtaining robust matching results and has a great value in improving applications of UAV multi/hyperspectral sensors.

关 键 词：SIFT匹配 无人机 近红外影像 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]