Images acquired by electron microscopy usually contain noise information,which can obscure useful signals and impact subsequent analysis.This work addresses the denoising of electron microscopy images by deep learning techniques.A self-supervised learning approach was employed and a novel denoising neural network model was constructed based on the U-Net architecture to obtain clean electron microscopy images.The proposed model incorporated maximum blur pooling and attention mechanisms to enhance denoising capabilities.Experimental validation demonstrated the effectiveness of the proposed approach.This study is particularly suitable for electron microscopy images,making it more favorable compared to supervised learning method.Furthermore,it outperforms traditional machine learning techniques in terms of denoising performance.
关键词
电子显微图像/自监督学习/神经网络/图像降噪
Key words
electron microscopy image/self-supervised learning/neural network/image denoising
维普
万方数据