惯导约束下的高光谱图像行相关几何校正算法

The line related geometric correction algorithm under the inertial navigation constraint for the hyperspectral image

针对搭载于无人机的推扫式高光谱成像仪受气流影响显著,获取的高光谱图像几何畸变明显,无人机搭载POS精度相对较低,传统方法难以消除图像几何畸变等问题。该文提出了结合图像行相关和低精度GPS惯导数据校正的思路,首先进行行相关匹配完成图像行之间的相对平移量,然后经过反算修正POS的翻滚角数据,最后基于更新后的POS数据进行几何校正。该方法充分结合了行相关匹配和POS校正的优势,最大限度地消除了几何畸变,经过对无人机图像验证,比其他单一方法处理效果更为明显。

Hyperspectral image that carrying unmanned aerial vehicle（UAV） significantly affected by the air flow, the line scan characteristics caused significant geometric distortion, so it is difficult to conduct the subsequent remote sensing application without geometric correction. But limited to the cost, size, and weight factors, the precision of current POS which carried by domestic hyperspectral drone platform is low, so it is difficult to eliminate the geometric distortion. Based on this, we put forward the thinking of combining with image line related and low precision GPS to derivative correction, first, conducting the line correlation matching and complete image translation amount of relative position and thenuse the GPS data to complete the constraint correction, then revising POS data of roll Angle, finally using the updated POS data for geometric correction. After validating the image of UAV, the proposed method fully combines the advantage of correlation matching and POS correction, eliminates the geometric distortion to the greatest extent, treatment effect is more apparent than any other single method, it has certain promotion value for this kind of geometric correction to low cost airlines line scan image of linear UAV under POS system.