机构地区: 北方工业大学信息工程学院
出 处: 《计算机测量与控制》 2014年第6期1735-1737,1753,共4页
摘 要: 针对机场目标检测定位时目标特征提取过于单一,识别结果不理想的问题,文章提出一种基于多源遥感图像的机场目标检测与定位方法;分别提取多光谱机场图像的跑道特征、纹理特征,以及相同区域SAR图像的后向散射特征,并将这些特征输入SVM分类器进行识别来判断候选区域是否包含机场,解决了目前特征单一、识别率较低等问题;实验结果表明该方法能更好地识别机场区域、提高识别率。 Aimed at single feature extraction and dissatisfied recognition results about airport target detection and location, this paper proposes an airport target detection and location method based on multi source remote sensing images. By extracting runway characteristics, texture on multispectral images of airport candidate regions, and backscattering characteristics on SAR image of the same regions, this meth- od inputs the features to SVM classifier to identify and determine whether the candidate regions contain the airport or not. This approach solves the problems of single feature extraction and lower recognition rate. The experimental results show this method can better identify candidate regions and improve the recognition rate.
关 键 词: 连通域 平行线 纹理特征 后向散射特征 支持向量机
领 域: [自动化与计算机技术] [自动化与计算机技术]