SAR图像阴影区域分割技术及其评估准则

A SAR image shadow region segmentation method and evaluation metrics

提出了一种阈值化技术与形态学方法相结合的SAR图像阴影区分割方法。该方法首先通过寻找图像平均强度最小的子区域对阴影区域进行初始分割,之后进行图像翻转、阈值化处理,最后利用形态学处理消除虚假像素,进行简单的连通性滤波,实现阴影区的自动分割。给出了3种图像分割评价准则,并基于计算机仿真和图像分割评价准则验证了该阴影区分割方法的有效性。

SAR imaging has shown the ability to identify targets at long ranges in adverse conditions. A recent thrust is to use the shadow information present in SAR imagery to identify the target. Clearly, it is essential to segment the shadow region from SAR image firstly. A SAR image shadow region segmentation method was presented based on morphology. The method segmented the shadow region originally. Then, morphological operations were used to perform some simple connectivity filtering to smooth the image and remove spurious pixels. And the shadow region was segmented automatically. To assess the shadow segmentation results, some target segmentation evaluation metrics were presented. The simulation results and target segmentation evaluation metrics show that the approach is effective.