论文部分内容阅读
合成孔径雷达(synthetic aperture radar,SAR)具有全天时、全天候观测,穿透能力强等特点,在灾害监测与评估、资源勘探等方面得到广泛应用,但其固有的相干斑噪声严重限制了单一利用SAR影像进行快速的信息获取。本文提出了一种基于GIS与贝叶斯网络的高分辨率SAR影像道路损毁信息提取方法。在GIS数据的辅助下,利用水平集分割与改进的D1检测融合的方法在影像上提取疑似道路损毁区域;再综合多证据及疑似损毁区观测值构建贝叶斯网络模型,对疑似损毁区进一步判断提取出实际道路损毁区域。实验结果表明,该方法能够快速、准确地对道路损毁信息进行提取。
Synthetic aperture radar (SAR) is widely used in disaster monitoring and assessment, resource exploration and so on because it has all-weather observation, all-weather observation and strong penetrating ability. However, its inherent speckle noise severely limits single Use of SAR Images for Fast Information Acquisition. This paper presents a high-resolution SAR image road damage information extraction method based on GIS and Bayesian networks. With the help of GIS data, the method of fusion of level set segmentation and improved D1 detection is used to extract the suspected road damaged area on the image. Then the Bayesian network model is constructed based on multiple evidences and the observed values of the suspected damaged area, Determine the actual road damaged areas extracted. The experimental results show that this method can extract the damaged road information quickly and accurately.