基于三种不同绝对对齐方法的分类器分析

Analysis of Three Kinds of Classification Based on Different Absolute Alignment Methods

雷达高分辨距离像识别时的一个重要问题是其平移对齐。相对于相关对齐方法,绝对对齐方法具有计算简单的优点,虽然对齐精度不如相关对齐法,但识别率并没有降低,并且由于利用最简单的分类准则———最小欧氏距离准则来进行分类,运算量与运算复杂度有明显的降低。文中介绍了三种基于不同绝对对齐方法———零相位对齐法、零线性相位对齐法、重心法的分类器,并对它们进行了理论分析与性能比较。实验结果表明,这三种分类器都能在低运算量的情况下获得了较好的识别性能。

Range alignment is a crucial problem in high range resolution profile （HRRP） based on radar automatic recognition. Comparing with correlation based range alignment, absolute alignment has the advantage of low computation complexity. Although its alignment accuracy is decreased, recognition ratio are not decreased, what＇s more, computation complexity is observably lowered because minimum Euclidean distance rule which is the simplest classification is applied to classify, This paper will introduce three kinds of classification based different absolute alignment method - Zero Phase Representation （ ZPR）, Zero Linear Phase Representation（ZLPR）, Barycenter Representation. Then we take theoretic analysis and the comparison of their performance. The experimental results show the three methods can all achieve a good recognition performance with low computation complexity.