论文部分内容阅读
本文介绍了基于统计特征(StaF)的分类器,它是一种稳健的、用于高距离分辨率(HRR)雷达的飞机识别(ID)方法。在对特征的数量或位置不作先验假定的情况下,我们实时地选择HRR图像的峰值特征。特征提取依靠所观测到图像的信息内容,把特征的数量、位置和幅度作为随机变量。本项研究的主要目的是,通过把对未知目标的识别错误减少到最小,同时维持对已知目标的高识别率的方法来增强分类器的稳健性。结果表明,这种StaF分类器能够显著地减少与未知目标有关的识别错误,同时维持高概率的正确分类。
This article presents a StaF-based classifier that is a robust aircraft identification (ID) method for high-range resolution (HRR) radars. In the absence of a priori assumptions about the number or location of features, we select the peak features of HRR images in real time. Feature extraction depends on the information content of the observed image, using the number, position and magnitude of the features as random variables. The main purpose of this study is to enhance the robustness of classifiers by minimizing the recognition errors of unknown targets while maintaining a high recognition rate of known targets. The results show that this StaF classifier can significantly reduce recognition errors associated with unknown targets while maintaining a high probability of correct classification.