基于信息粒增强的无人机航拍图像与卫星影像匹配方法  

Image matching method between UAV aerial images and satellite imagery based on information granule enhancement

作　　者：任艳[1] 张天博 刘胜男 张檬 REN Yan;ZHANG Tian-bo;LIU Sheng-nan;ZHANG Meng(School of Artificial Intelligence,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳航空航天大学人工智能学院,辽宁沈阳110136

出　　处：《计算机工程与设计》2025年第4期1013-1021,共9页Computer Engineering and Design

基　　金：辽宁省教育厅基础研究基金项目(LJKZ0219);辽宁省科技厅自然科学基金面上基金项目(2021-MS-265)。

摘　　要：无人机航拍图像与卫星影像在不同季节下的地物信息差异给图像匹配带来的挑战。针对不同季节下异源图像的匹配困难问题,提出一种基于信息粒增强的图像匹配方法。构建信息粒并对无人机航拍图像和卫星瓦片图像进行增强;利用无人机航拍图像和卫星影像的地面分辨率差异对增强后的航拍图像进行尺度预处理;利用Sobel算子对图像进行边缘检测,并匹配无人机航拍图像与卫星瓦片。在冬季与夏季数据集上的实验结果表明,该算法可以有效解决四季变化导致的无人机航拍图像与卫星地图地物景象不一致时的准确定位问题。The differences in ground object information between UAV aerial images and satellite images across different seasons present a challenge for image matching.An image matching method based on enhanced information granules was proposed to address the difficulty of matching heterogeneous images across different seasons.Information granules were constructed,and both UAV aerial images and satellite tile images were enhanced.Based on the ground resolution of the UAV aerial images and satellite images,the enhanced UAV aerial images were preprocessed for scale.The edge detection was performed on the images using the Sobel operator,and UAV aerial images were matched with satellite tiles.Experimental results on winter and summer datasets demonstrate that this algorithm can effectively solve the precise positioning problem caused by inconsistencies in ground object scenes between UAV aerial images and satellite maps due to seasonal changes.

关 键 词：图像匹配 图像增强 信息粒 边缘检测 相似性度量 无人机航拍图像 卫星图像 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]