摘要
针对单一传感器的局限性,通过多传感器融合技术,最大限度地获取对目标场景的边缘信息描述。由于光学图像和SAR图像成像机理不同,即便配准精确的图像提取出来的边缘也会出现不重合的情况。首先采用离散域Canny算法提取光学图像边缘特征,用ratio算法提取SAR图像边缘特征,通过不变矩算法对两幅边缘特征图中的边缘进行匹配融合,得到了比单一图像边缘特征图更完整更清晰的边缘特征图,获取了更多的目标场景的信息描述。
Aiming at single sensor＇s limitation,this paper uses the technology of multi-sensor fusion,which furthest obtains the information description of target scene.Due to different imaging mechanisms,even though the remote sensing images are accurately matched,the edges extracted will not be in the same location.This paper firstly adopts Canny discrete criteria to extract the optical image,and adopts the ratio algorithm to extract the SAR image.By adopting the methods of moment invariants to match each edge in the two edge feature images,obtaining the clearer and more complete edge feature images than the single one means more information descriptions.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第29期171-176,共6页
Computer Engineering and Applications
关键词
边缘提取
图像匹配
图像融合
edge detection
image matching
image fusion
作者简介
李璟旭（1982-），女，硕士研究生，主要研究领域为图像识别、图像特征级融合。
