A denoising method based on CNN through multi-layer separation

In the process of electromagnetic information leakage detection, the traditional denoising methods have the problem of insufficient ability to identify useful information under low SNR. This problem results in the useful information being easily removed as a noise signal in the denoising process. Aiming at this problem, this paper proposes a denoising method based on convolutional neural network (CNN), which is achieved by decomposing red and black signals layer by layer. This method can not only show a better denoising effect under a low SNR condition, but also locate, identify and enhance the useful information effectively.