无人机影像局部增强方法及其在影像匹配中的应用  

Local enhancement method and its applications to UAV image matching

作　　者：唐敏[1] 李永树[1] 李歆[2] 刘波[2] 

机构地区：[1]西南交通大学地理信息工程中心 [2]78155部队

出　　处：《国土资源遥感》2013年第4期53-57,共5页Remote Sensing for Land & Resources

基　　金："十一五"国家科技支撑计划项目(编号:2006BAJ05A13)

摘　　要：无人机影像的质量与成像时的光线、成像角度及所拍摄地物的特征有着密切关系。很多影像都存在着视觉对比度差、影像畸变大及分辨率低等缺点,给后续影像匹配和正射纠正带来困难。为了提高无人机影像的匹配效果,首先利用原始影像的离散灰度信息计算每个像元灰度值与整个影像灰度平均值的离散程度,并把影像划分为不同区域;然后利用距离加权插值方法计算的变换函数对各区域影像进行不同程度的增强处理,对增强后影像的直方图进行修正;最后利用核线约束法对影像进行匹配和均匀度检验。研究结果表明,由于增强后影像的灰度梯度差变大,在畸变较大及灰度变化较小的林地分布区域,影像的匹配成功率和均匀度都有一定程度的提高。The quality of UAV image is closely related to light, imaging angle and the feature of the ground object. There are some shortcomings in many images, such as poor visual contrast, large distortion and insufficient resolution, which cause many difficulties in subsequent image matching and orthorectification. To obtain better results in UAV image matching, the authors first calculated the degree of dispersion between gray value of pixel and image gray average to divide the image into different areas according to the discrete gray information of original images. Then, the distance weighted interpolation method was used to calculate the transformation function and enhanced the various regions in different degrees. After that, the histogram of enhanced image was corrected. Finally, the epipolar constraint method was used in the image matching experiment. The results show that, because of the enhanced image gray value gradient difference, there exists some extent of increased matching success rate and uniformity in the woodland area which has smaller gray change, thus favoring the smooth progress of the work of subsequent image pro- cessing.

关 键 词：无人机影像 局部增强 匹配 可变窗口 梯度差 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]