基于Laplace分布的非下采样小波包脉冲星信号消噪

发布时间：2018-06-15 02:51

本文选题：非下采样小波包(NWP) + 脉冲星信号　；参考：《光电子·激光》2017年06期

【摘要】：为了提高脉冲星信号的去噪效果,提出了一种基于非下采样小波包(NWP)分解的局部Laplace模型消噪方法。首先对真实脉冲星信号进行NWP分解,统计真实脉冲星信号NWP系数的分布特性,建立真实脉冲星信号小波包系数的Laplace分布模型;然后在Laplace先验概率分布的基础上,根据最大后验概率(MAP)估计准则,利用含噪脉冲星信号的小波包系数对真实脉冲星信号的小波包系数进行有效估算;最后对估算出的小波包系数进行NWP重构,得到消噪后的脉冲星信号。采用不同的脉冲星信号进行实验分析的结果表明,与经典的基于高斯分布的非下采样小波(NSW)消噪和NWP消噪相比,本文方法可以更有效地去除噪声,同时更好地保留信号中的微脉冲等细节信息,在信噪比(SNR)、均方根误差(RMSE)、相关系数(CC)和峰值相对误差(REPV)等都有较好的改善。
[Abstract]:In order to improve the de-noising effect of pulsar signal, a local Laplace model de-noising method based on non-downsampling wavelet packet decomposition is proposed. Firstly, the real pulsar signal is decomposed by NWP, and the distribution characteristics of the real pulsar signal NWP coefficient are analyzed, and the Laplace distribution model of the real pulsar signal wavelet packet coefficient is established, and then the Laplace priori probability distribution is based on the Laplace priori probability distribution. According to the maximum posterior probability (MAPP) estimation criterion, the wavelet packet coefficient of the noisy pulsar signal is used to effectively estimate the wavelet packet coefficient of the real pulsar signal, and the estimated wavelet packet coefficient is reconstructed by NWP. The de-noised pulsar signal is obtained. The experimental results of different pulsar signals show that the proposed method is more effective than the classical non-down-sampling wavelet Gao Si based de-noising and de-noising methods. At the same time, the detailed information such as micropulse in the signal is better preserved, and the SNR, RMSE, CCV and relative peak error are improved in SNR, RMSE, RMSE, RMSE, RMSE, RMS, RMSE, RMS, RMSE, RMS, RMSE, RMS and RMSE, respectively.
【作者单位】：武汉科技大学信息与计算科学系;武汉理工大学计算机科学与技术学院;
【基金】：国家自然科学基金(61473213,61671338)资助项目
【分类号】：TN911.7

【相似文献】