首页 | 本学科首页   官方微博 | 高级检索  
     

SVD技术在二维线性预测谱估计中的应用
引用本文:殷林,邹理和. SVD技术在二维线性预测谱估计中的应用[J]. 西安交通大学学报, 1986, 0(6)
作者姓名:殷林  邹理和
作者单位:西安交通大学信息与控制工程系(殷林),西安交通大学信息与控制工程系(邹理和)
基金项目:科学院基金(技-85847)项目
摘    要:本文提出了一种在正弦加白噪声情况下的二维线性预测谱估计的改进方案,它实际上是一维Cadzow SVD方法在二维谱估计中的推广。这种谱估计的性能对于信号中所含正弦个数的准确估计是非常敏感的,本文证明了由二维自相关矩阵信号特征向量所排成的矩阵的秩等于正弦个数,由此可以得到一种二维正弦谱估计的新的判阶方法。这种新的判阶方法与改进的谱估计方法构成了一种新的谱估计方法,本方法所获得的谱估计可以有效地抑制峰裂、压缩伪峰,从而使谱估计的分辨率获得明显地改善,此方法可用于因果、半因果、非因果等各类不同的线性预测系统,并具有运算量不大,信噪比要求低等突出特点,实验验证了上述结论。

关 键 词:信息处理  频谱分析  数字处理

AN APPLICATION OF SVD TO 2-D LINEAR PREDICTION SPECTRAL ESTIMATION
Yin Lin,Zou Lihe. AN APPLICATION OF SVD TO 2-D LINEAR PREDICTION SPECTRAL ESTIMATION[J]. Journal of Xi'an Jiaotong University, 1986, 0(6)
Authors:Yin Lin  Zou Lihe
Affiliation:Yin Lin;Zou Lihe Department of information and Control Engineering
Abstract:This paper presents a modified 2-D linear prediction approach for spectralestimation of a finite number of 2-D sinusoids embedded in additive whitenoise. This approach could be viewed as a 2-D extension of Cadzow's SVDtechnique. Based on some properties of the rank of a matrix formed by eigen-vector associated with the largest eigenvalue, a new order-determining methodis also proposed. The spectral estimation based on the new approach results ina reduetion of peak splitting, compression of spurious peaks, as well as animprovement in resolution. This approach can be applied to all causal,semicausal, and noncausal linear prediction systems with the distinct advantageof having low computational requirement and low signal to noise ratio con-straint. Simulatioa results ave given.
Keywords:information processing  spectral analysis  data processing
本文献已被 CNKI 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号