论文部分内容阅读
提出了一种基于支持向量机的InSAR干涉图区域分割相位解缠法,以支持向量机为分类工具,利用干涉图中的残差点、相位导数方差等相位属性对干涉图中像元分类,把干涉图分割成高质量非掩模区域和低质量掩模区域。先用改进的Itoh方法对非掩模区域解缠,然后应用区域生长策略对掩模区域解缠。最后用真实的和模拟的干涉相位图试验表明算法比现有的传统算法如枝切法等InSAR干涉图相位解缠方法更有效。
A phase unwrapping method based on Support Vector Machine (SVM) for InSAR interferogram segmentation is proposed in this paper. Support Vector Machine (SVM) is used as a classification tool to classify the pixels in the interferogram using the phase attributes of residual points and phase derivatives in the interferogram. Interferogram is divided into high quality non-mask area and low quality mask area. First, the un-masked region is unwound with a modified Itoh method, and then the mask region is unwound by using the region growing strategy. Finally, the real and simulated interference phase diagram experiments show that the proposed algorithm is more effective than the existing phase-unwrapping method of InSAR interferograms such as the branch-cut method.