基于正则化方法的SAR图像分割及目标边缘检测算法

Edge detection in SAR segmentation based on regularization method

为了准确地分割图像并获取清晰、连续的边缘特征,在系统分析正则化技术的基础上,提出了一种基于正则化技术的SAR图像分割及目标边缘检测算法。该算法首先利用一种改进的正则化方法对SAR图像进行预处理,然后分析图像的统计特性,利用阈值化技术获取SAR图像的目标区域和阴影区域,最后通过加窗处理技术对分割后的目标区域进行边缘特征提取。并用MSTAR数据进行大量的仿真实验,结果表明,与经典的边缘检测方法相比,该方法在获取良好的分割结果的同时能更精确、更完整地检测出目标的边缘特征。

Since right segmentation, clear and continuous edge feature retrieval are important for the analysis and interpretation of SAR images, a new segmentation and edge detection algorithm for SAR images was presented by analyzing regularization method. In this algorithm, the images were preprocessed with a modified regularization technique first, and then after the statistic characteristic analysis, the interested target and shadow region of these images were segmented, at last edge detection was realized by windowing. Comparing to the traditional methods, experimental results with MSTAR dataset show that the algorithm can well segment images and maintain detail feature of the target region.