论文部分内容阅读
根据ROC(receiver operating characteristics)技术能评估分类器在所有可能工作阈值下总体性能的特点,建立包含边缘像素点相关分析与ROC分类决策的ROC融合准则。依据该准则组合多种SAR边缘检测算子,并得到合成孔径雷达(SAR)影像的“理想”边缘检测结果。实验结果表明,本文方法能融合多种边缘检测算子的优点,有较强的开放性与目标适应性,并且不需要手工设置阈值,自动化程度高,有很强的工程实用性。
According to the receiver operating characteristics (ROC) technique, the overall performance of the classifier can be evaluated under all possible working thresholds, and the ROC fusion criterion including edge pixel point correlation analysis and ROC classification decision is established. Based on this criterion, a variety of SAR edge detection operators are combined and the “ideal” edge detection results of Synthetic Aperture Radar (SAR) images are obtained. The experimental results show that the proposed method can combine the advantages of multiple edge detection operators, have strong openness and target adaptability, and do not need to manually set the threshold, and have a high degree of automation and strong engineering practicability.