提出了一种引入先验约束的合成孔径雷达(SAR)图像的目标分割技术,以解决强杂波背景干扰下的目标分割困难问题。不同于基于统计理论的目标检测,文中利用目标图像切片在图像域的稀疏性,通过稀疏分解的方法构建目标特征窗函数实现目标的检测,并引入目标的形状先验对目标区域进行修正;然后,利用目标阴影的空间约束对基于统计检测的阴影区域进行修正,实现目标的分割;最后,基于实测数据验证了算法的有效性。

A target segmentation method for synthetic aperture radar (SAR) images is proposed by introducing prior constraints, which is used to overcome the difficulties of segmentation under strong clutter circumstance. Combined with statistical theory based target detection, the proposal makes use of the sparsity of target image chips. The target feature window is constructed based on sparse decompsition, which is used to detect targets. Afterwards, the prior knowledge of target shape is brogught in to modify the target region. Then, the space constraint of target shadow is utilized to modify the detected shadow region, by which the segmentation is completed. Finally, experiments are conducted on real measured data to validate the effectiveness of the proposed algorithm.

TN957.51