非平稳海杂波的消除趋势波动分析

Detrended fluctuation analysis of non-stationary sea clutter

根据消除趋势波动分析(DFA)的基本原理,研究了非平稳条件下两种不同波段实测海杂波数据的分形标度特性以及海洋环境参数对分析结果的影响.采用分段数据均值/方差的差异、随机游走模型和对数方差-尺度法等对海杂波的非平稳性、分形特性和长程相关性进行定性判定.利用DFA得到的波动函数-尺度的双对数坐标曲线,分析了不同尺度范围内海杂波和目标单元的标度指数特性.对交叉标度现象的研究结果表明:它由海杂波本身的相关性在不同尺度范围存在差异导致的,且在VV极化时差异性更加显著.当海况改变时,不同尺度范围内标度指数的变化趋势与极化方式密切相关.最后,以拟合误差为检验统计量,对分析结果在目标检测算法中的应用进行了验证性分析.

According to the basic principle of detrended fluctuation analysis（DFA）, this paper researches the fractal scaling property of measured sea clutter data with two different bands under non-stationary conditions,as well as the influence of o- cean environment parameters on the analysis results. Firstly, with the methods of variation mean/variance between different data segments, random walk model and logarithm variance verse scale etc., the properties of sea clutter such as non-station- ary,fractal and long range correlations are qualitatively justified. The scaling expo- nent property of sea clutter and targets bins is then analyzed within different scale ranges by making use of the double logarithm fluctuation function verse scale curves worked out by DFA. The research results of scaling crossover phenomenon shows that it is caused by the discrepancy in correlation properties of sea clutter at differ- ent scales,and the discrepancy is more prominent for VV polarized data. When sea state changes, the changing trend of scaling exponents in different scale ranges is closely related to polarization patterns. Finally, the fit error is chosen as detection statistics to validate the application of analysis results in the field of target detection algorithm.