【摘要】 本文通过频率域信噪比分析评价奇异值分解、截断和重构法(SVDTR)消除地球物理数据中随机噪声的效果。首先给出频率域中分量和全频谱平均信噪比、含信比和信噪比增益的定义。然后根据这些统计量对SVDTR前后40×40的理论电磁数据中某一测线数据频谱变化的情况进行分析对比。结果表明,该方法可以提高几乎每一个频率分量的信噪比,最高可提高44倍之多,平均8倍多。全频谱平均含信比由0.4提高到0.7左右。最后,对全部二维数据x和y二个方向上的含信比谱分析表明,全频平均含信比增大1倍多。更多还原

【Abstract】 This paper has evaluated the effects of SVD noise rejection by analysing S/N ratio in frequency domain. Several statistical parameters for the S/N analysis have been defined. The changes in S/N ratio due to SVD filtering have been discussed in frequency domain using a column data from theoretical data matrix(40 × 40). The SVD filtering has increased S/N ratio almost at each frequency component ,averagely by a factor of 8 and most highly 44. The average S/(S+N) for total frequency band has increased from 0. 4 to 0. 7. Finally, the S/(S + N) for the 2-D data matrix has increased by once as compared with that befor SVD filtering, along both raw and column directions.更多还原