SAR图像启发式搜索直线提取  

Straight Line Extraction by Heuristic Search for SAR Image

作　　者：曾阳帆 刘伟[1] 陈建宏[1,2] 赵拥军[1] 

机构地区：[1]解放军信息工程大学导航与空天目标工程学院 [2]73906部队

出　　处：《信号处理》2015年第3期282-289,共8页Journal of Signal Processing

基　　金：国家自然科学基金资助课题(41301481)

摘　　要：直线特征是识别线状目标的基础,现有启发式搜索算法仅适用于光学图像的直线特征提取,对SAR图像效果并不理想。针对该问题,提出一种SAR图像直线特征提取的启发式搜索算法。首先对SAR图像进行Frost滤波,然后利用指数加权均值比(ROEWA)算子进行边缘检测,再利用非极值抑制得到边缘二值图,最后采用二次搜索策略及"直线走原则",充分考虑局部信息,设计启发式搜索的代价函数及搜索规则,实现直线特征提取。实验结果表明,该方法具有较好的抗噪性和抗断裂能力,能够有效地提取出SAR图像中的直线特征。Straight line feature is the basis of linear target recognition.The existing heuristic search algorithms are suit-able for straight line feature extraction in optical image while performing poor in SAR image.For this problem,a straight line extraction method for SAR image based on heuristic search is presented.Firstly,SAR image is filtered by Frost filter. Then,edges are detected by ratio of exponential weighted average （ROEWA）operator and non-maxima suppression location algorithm is adopted to obtain the binary edge map.Finally,considering the local information sufficiently,the strategy of search twice and the principle of “go straight”is utilized and the cost function and rules of heuristic search are designed. Then the straight line extraction is completed.The experiment results show good anti-noise performance and fracture resist-ance of the proposed method which can effectively describe the straight line feature of SAR images.

关 键 词：合成孔径雷达图像 直线特征提取 启发式搜索 代价函数 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]