纹理特征辅助的S AR影像冰川识别

青藏高原的冰川监测对气候变化研究有着重要的意义，通过遥感图像可以大范围长时间的监测冰川的变化，识别冰川边界是研究的重点。为了研究SAR影像纹理特征在冰川识别中的作用，以喀喇昆仑山地区的克勒青河上游为研究区，利用2018年Sentinel-1A数据进行干涉处理得到相干系数，然后基于相干系数提取了均值、方差、同质性、反差、相异性、熵、相关性共7种纹理特征，并对不同纹理特征组合之间的提取效果进行了比较。结果表明VV极化方式下均值、方差、同质性、相异性的特征组合冰川识别效果最好。据此提取了克勒青河上游区域的冰川边界，最高精度达到91.36%，该方法明显优于基于相干系数图的阈值分割法和基于光学影像的波段比值法，冰川识别精度提高了约2%。

Texture-assisted glacier recognition based on SAR image

Glacier monitoring on the Tibetan Plateau has important implications for climate change research. Remote sensing can be used to monitor glaciers in a wide range over long periods of time, and the identification of glacier boundary becomes the focus of research. In this paper, for studying the role of synthetic aperture radar (SAR) image texture features in glacier recognition, the upper stream of the Keleqing River basin in the Karakorum Mountains is used as the study area. Sentinel-1A SAR image in 2018 is used for interferometry processing to obtain the coherence coefficient, and then texture features, such as the mean, variance, homogeneity, contrast, dissimilarity, entropy and correlation are extracted based on the coherence coefficient. The combined extraction effects between different texture features are compared. The results show that the combination of mean, variance, homogeneity and dissimilarity of SAR image with VV polarization mode has the best effect of glacier recognition. Based on this, the glacier boundary in the upper streams of the Keleqing River is extracted, and the highest precision reaches 91.36%, which is superior to the threshold segmentation method based on coherence coefficient diagram and the band ratio method based on optical image, improving the accuracy of glacier recognition by about 2%.