Scholarlink

Abstract <p>The problem of compensating for distortions of images obtained by a scanning probe microscope, arising due to the finite dimensions of the probe, is solved. For this purpose, an approach based on a combination of two different data processing methods is proposed: an artificial neural network is constructed that segments the analyzed image to isolate an area of interest, and the surface relief is assessed in this area taking into account the size of the probe. A mathematical model of the image processing process is constructed, which allows determining the optimal operator that maximizes the assessment of the quality of compensation for distortions of scanning probe microscopy images on a fixed dataset.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Integrated Method for Segmentation and Relief Refinement of Atomic Images Based on Probe Shape Convolution and U-Net Architecture Neural Network

  • 期刊论文

  • 发表日期：

    2025年9月1日

查看全文
  • K. S. Zyryanov,
  • A. I. Oreshkin,
  • A. I. Chulichkov

摘要

Abstract

The problem of compensating for distortions of images obtained by a scanning probe microscope, arising due to the finite dimensions of the probe, is solved. For this purpose, an approach based on a combination of two different data processing methods is proposed: an artificial neural network is constructed that segments the analyzed image to isolate an area of interest, and the surface relief is assessed in this area taking into account the size of the probe. A mathematical model of the image processing process is constructed, which allows determining the optimal operator that maximizes the assessment of the quality of compensation for distortions of scanning probe microscopy images on a fixed dataset.