【摘要】 本文提出用奇异值分解方法消除地球物理信号频段上的干扰噪声。通过对地球物理数据的奇异值分解,可以确定信号和随机噪声所对应的奇异值范围。截断随机噪声所对应的奇异值,利用与信号对应的特征象重构数据,可以消除随机噪声,而对应随机噪声的特征象中包含各种频率成份,因而该方法可消除信号频段上的随机噪声。根据对电磁法数据进行计算和频谱分析的结果表明,信号频段上随机噪声的影响可降低60％多。更多还原

【Abstract】 This paper presents a new approach to eliminate the random noise which has the same-frequency spectrum range as the signal by using singular value decomposition. This approach is based on the decomposition of geophysical data matrix into eigenimages. The distribution of the singular values in geophysical data indicates which singular values are related to the signal and which to the random noise. By truncating the singular values associated with the noise, the data may be recons tructed from only a few of the first eigenimages associated with the signal, and the random noise including that in the signal frequency range can be rejected because the rejected eigenimages consist of various spectrum content according to spectrum analysis of the eigenimages. The test of the method on synthetic electromagnetic data shows that the approach can reduce the random noise in the signal frequency range more than 60％.更多还原