在雷达地面目标识别中, 采用合适的特征选择算法有助于在复杂的特征空间中挑选出对不同类型目标具有较强区分能力的特征。结合信息论和合作博弈理论, 提出一种基于Owen值的特征选择算法, 旨在选择出与类别相关度高、特征与特征之间冗余性低、依赖性强的最优特征子集, 达到在雷达目标识别中提高识别率的目的。利用雷达实测数据验证其性能, 结果表明, 所选特征子集的平均识别准确率优于两种经典的Filter式特征选择算法, 且该算法具有良好的噪声/杂波稳健性。

The cross-correlation method for temporal characterization is investigated using simulations of the twocolor above threshold ionization (ATI) on He induced by a vacuum ultraviolet (VUV) free-electron laser (FEL) in the presence of an infrared (IR) field. Non-linear dependencies of the sideband structure produced in the two-color ATI process are expressed as a function of IR laser intensity by considering the spatial distributions and temporal jitter of both lasers. The temporal properties of the FEL pulse can be characterized accurately using the cross-correlation method at a low IR laser intensity of ~3 \times 10^{10} W/cm2 but with low cross-correlation signals. When the dynamic range of sidebands is increased to high IR intensity, the accuracy of the cross-correlation method becomes crucially dependent on the actual nonlinear index. An approach of determining this index is proposed here to improve the accuracy of temporal characterizations.