基于形状与纹理特征的高分辨率SAR影像桥梁目标自动提取方法  

Automatic Extraction of Bridges with Shape and Texture Characteristics Using High Resolution SAR Images

作　　者：贾洪果 刘路遥 韦博文 张梦意 武燕 章昊 

机构地区：[1]西南交通大学地球科学与环境工程学院测绘遥感信息系,四川成都610031

出　　处：《测绘通报》2017年第12期82-85,共4页Bulletin of Surveying and Mapping

基　　金：国家重点研发计划资助(2017YFB0502700);国家自然科学基金(41701535);卫星测绘技术与应用国家测绘地理信息局重点实验室经费(KLSMTA-201602)

摘　　要：基于高分辨率SAR影像中桥梁目标特征的分析,提出了一种桥梁目标自动提取方法。首先使用多影像时序平均法对SAR影像进行预处理,去除SAR影像噪声;再使用阈值法对水体分割;最后采用形状特征与纹理特征相结合的方法提取桥梁目标。本方法在提取效果上明显优于传统桥梁目标提取方法,其提取准确率可达87.5%。证明本文方法在面向高分辨率雷达影像的桥梁目标自动提取应用上的可靠性和可行性。This study proposed an automatic extraction method of bridge information based on the shape and texture characteristics of bridges in high resolution SAR images. This method consists of three steps：（1） elimination of noise by time-series image averaging; （2）detection of water-areas by a threshold; （3）extraction of bridge information based shape and texture characteristics of bridges. The experimental results show that the precise of extraction by the method proposed in this study can reach 87.5% and is significantly better than traditional methods. The experimental evaluation indicated that this method has high reliability and feasibility for automatic extraction of bridge information from high resolution SAR images.

关 键 词：高分辨率 SAR影像 桥梁 形状特征 纹理特征 自动提取 

分 类 号：P237[天文地球—摄影测量与遥感]