针对现有基于?/?/?分解的全极化高分辨率距离像(HRRP)特征提取方法没有考虑度量尺度对特征性能的影响问题,引入动态互信息思想设计了度量尺度评价准则,并提出了基于平均度量尺度、不定度量尺度和金字塔型度量尺度3种特征提取方法.采用两类飞机目标全极化HRRP数据对提取的特征子集进行了有效性分析,并通过识别多类飞机目标验证了3种方法提取的特征子集具有良好的类别可分性和稳定性. In view of the fact that the existing HRRP feature extraction method based on? /? /? Decomposition does not consider the impact of metric scale on the feature performance, this paper introduces the dynamic mutual information theory and designs the evaluation criteria of metric scale, Three kinds of feature extraction methods based on average measure scale, indefinite scale measure and pyramid measure measure are adopted.The validity of the extracted feature subset is analyzed by using two types of target fully-polarized HRRP data, The feature subset extracted by the three methods has good category separability and stability.