信息几何是源于对统计流形几何性质的研究而发展起来的一套理论体系,为解决雷达信号处理等问题提供了一种新的思路和方法。本文基于信息几何理论,从一个全新的角度认识了雷达目标检测问题,并通过定义统计流形上两分布间的距离,进而提出了雷达目标检测的信息几何方法,该方法通过计算统计流形上由观测数据估计得到的分布与目标有无两种假设分布间距离之差的大小就可以实现目标的检测,将统计检测问题转化为了统计流形上的几何问题,简单实用,最后仿真比较了其与传统似然比检测的检测性能。
Information geometry is a set of system info originated from the study on the geometry properties of statistical manifold, which offers one kind of new thinking and method to the radar signal processing problems. A new kind of view to radar target detection problem is given based on information geometry theory in this article, simultaneously, information ge- ometry method to radar target detection is also proposed by defining the distance between the null and alternative distribu- tion on statistical manifold. The information geometry method to radar target detection is achieved by comparing the distance between oberserved distribution and the hypothetical distribution when target exists or not. Moreover, it is easy to be accom- plished by translating radar target detection problem into geometry problem on statistical manifold. Finally, detection per- formance of information geometry method and likelihood ratio detector is compared by simulations.