【摘要】 时间域航空电磁数据经预处理后,仍存在残余噪声,影响电磁探测对地下异常的识别能力。笔者提出一种基于最小噪声分离的去噪方法,将一组含噪电磁数据通过旋转矩阵线性变换为按照信噪比大小排列的最小噪声分离成分,利用信噪比较大的最小噪声分离成分重构电磁数据,以达到分离噪声的目的。仿真数据去噪结果表明:最小噪声分离不仅能够有效压制晚期道剖面噪声,还能准确分辨异常信息;晚期道信噪比较测线滤波提高了11.28 d B,实测数据的噪声水平也由±50 n T/s降低到±10 n T/s。更多还原

【Abstract】 There is still residual noise in time-domain airborne electromagnetic data after preprocessing,which will affect the recognition of target. We proposed an approach to remove the residual noise based on minimum noise fraction. A set of noise-contaminated data will be linearly transformed by using the rotation matrix to the minimum noise fraction components,which are arranged in signal to noise ratio( SNR) from big to small.We use the minimum noise fraction components with the bigger SNR to reconstruct the electromagnetic data for separating the signal and noise. The experiment with the simulation data test shows that the minimum noise fraction can not only effectively suppress the noise of the profile of later channels,but also accurately identify the information of the target. The SNR has improved by 11. 28 d B compared with the survey-line filtering. The noise level for the field data is reduced from ± 50 n T / s to ± 10 n T / s after noise removal.更多还原