基于区域生长法和变差函数的PolSAR影像山脊线提取

山脊线通常是地震引发的山地灾害的起始部位。在监测此类山地灾害时,需要明确山脊线所处位置,从而为制定出相关预防措施提供支持。根据实际经验,由于合成孔径雷达(Synthetic Aperture Radar,SAR)的成像特殊性,在检测含建筑物的全极化SAR(Polarimetric Synthetic Aperture Radar,PolSAR)影像山脊线时,往往受到高散射强度的建筑物干扰,从而出现误判。针对PolSAR影像中山脊线受到建筑物干扰出现误识的问题,提出一种结合区域生长法和变差函数的识别方法。在识别过程中,首先,对采集的PolSAR影像通过区域生长法进行分割处理,再通过模糊C均值(Fuzzy C-Means,FCM)分类方法对变差纹理等特征进行聚类分析,并进行相似度判断,获取建筑物区域,在此基础上,对比建筑物、山脊线的混分成分,进而提取山脊线。通过对该方法进行实验对比分析,发现其相较于阈值分割方法提取精度有明显提高,可以为PolSAR影像中山脊线识别提供一种新思路。

Extraction of ridge line from PolSAR image usingthe region-growing method and variogram

The ridge line is usually the starting point of mountain disasters caused by earthquakes. An accurate identification of the location of the ridge line while monitoring such mountain disasters is crucial to facilitate the development of relevant preventive measures. However, detection of the ridge line in PolSAR images containing buildings with high scattering intensity can be challenging due to some imaging issues of synthetic aperture radar (SAR), resulting in misjudgments. A recognition method combining the region-growing method and the variogram is proposed in this paper to address this problem. During the recognition process, the collected image is first segmented by the region-growing method. Then, the texture variations and other features are clustered and analyzed by the Fuzzy C-means method, and the similarity is assessed to obtain the building area. The ridge line can then be extracted by comparing this with the mixed image. A comparative analysis of the experimental results for the proposed method show significantly improved extraction accuracy as compared with the threshold segmentation method. Thus, this study can provide a new idea for the detection of ridgelines in PolSAR images.