联合灰度阈值分割及轮廓形态识别的河道提取

在合成孔径雷达(SAR)所获取的遥感图像中,河道区域由于其中水介质的电磁波反射率较低,在图像中呈现灰度低且灰度起伏小的特征。此外,河道区域由于水流的冲击作用,也具有显著的条带状轮廓形态特征。鉴于河道的这2种特征,联合灰度阈值分割及轮廓形态识别方法,采用多级分割策略,实现对河道区域的准确提取。实验结果证明,与经典的Otsu灰度阈值分割及其多种改进方法相比,多级分割方法能够更好的提取河道区域轮廓,且漏警率和虚警率均较低。

River channel extraction by combining grey threshold segmentation and contour form recognition

In the remote sensing image acquired by the synthetic aperture radar, the grey level of the pixels inclu-ding in the river channel is relative low and with low moderation since the electromagnetic wave reflectivity of the water is low.Moreover, because of the effect caused by water flow the river channel also has a saliency contour. Combining these two features of the river channel, a multistage segmentation method to correctly extract the region of the river channel is proposed in this paper.The experimental results prove that, comparing with the classic Otsu grey threshold method and its updated editions, the proposed method which combines the grey feature and the con-tour feature of the river channel has better performance to correctly extract the region of the river channel, both the missed alarm rate and the false alarm rate are lower.