不同噪声背景下基于广义斜投影算子的滤波方法

摘要

常规的斜投影(conventional oblique projection,COP)滤波通常假定背景噪声是高斯白噪声,而实际环境中的背景噪声多为色噪声.通过拓展COP的概念,得到一类广义斜投影(generalized oblique projection,GOP)算子,提出一种同时适用于白噪声和色噪声背景的最小化干扰约束广义斜投影(minimized interference constrained generalized oblique projection,MIGOP)自适应滤波方法,实现了在不同噪声背景下应用统一的斜投影算子进行滤波,并讨论了MIGOP滤波方法与其他各种相关的斜投影滤波方法之间的区别与联系.仿真实验验证了理论分析结果,通过对比不同噪声背景下各种GOP滤波方法的性能证实了所提方法的有效性.%Conventional oblique projection (COP) filter concentrates on white background noise, whereas the colored noise often shows in real circumstances. Through extending the concept of COP, a generalized oblique projection (GOP) operator is defined and a minimized interference constrained generalized oblique projection (MIGOP) filter technique which can be used in both white and colored noise scenarios is proposed. The filtering operator under different contaminating noises is unified and the relationship between the proposed MIGOP filter and some other kind of oblique projection filtering approaches is also researched. The theoretical analysis is verified by numerical simulations and the effectiveness of the proposed GOP filters is proved in both white and colored noise backgrounds.