基于不同准则函数的雷达目标特征空间变换

为解决雷达目标高分辨距离像的识别问题 ,将 4种基于不同准则推导出的线性特征空间变换进行了归纳。分别用这几种变换方法对 6类飞机缩比目标高分辨距离像数据进行特征空间变换 ,然后用同一分类器判决 ,并对其识别结果加以分析与比较 ,得出在该应用条件下的有益结论。

Radar Target Feature Transforms Based on Different Criterion Functions

In order to solve the radar target identification based on high resolution range profiles, four linear feature transform methods, such as Karhunen Loeve transform, discriminant analysis feature extraction including its modified version and Mahalanobis based transform, are compared and used for identifying six kinds of aircraft scaling models The identification results are given and analyzed. The experimental result shows that the modified discriminant analysis feature extraction method is practical, and that by feature extraction and transform the identification probability is higher than that by using high resolution range profile directly Perhaps this is because there is some redundant information in the range profiles Feature extraction and reduction will eliminate the effects from the redundant information ;