Artifact Removal and Image Restoration in AFM:A Structured Mask-Guided Directional Inpainting Approach

TL;DR

AFM images suffer from environmental noise and scanner artifacts that degrade interpretability. The authors present a lightweight, geometry-aware pipeline that automates artifact detection, artifact localization, and restoration via mask-guided directional inpainting, with a Tkinter GUI for real-time control. Key contributions include a ResNet-18-based four-class classifier, a fast segmentation module with aspect-ratio-based stripe expansion, a mask-guided Smart Flatten, and directional inpainting with Telea and localized smoothing, achieving high accuracy and robust restoration on AFM data. The approach preserves nanoscale surface features and enables high-throughput, reproducible AFM data analysis, with potential to integrate into digital twins and multi-channel AFM workflows.

Abstract

Atomic Force Microscopy (AFM) enables high-resolution surface imaging at the nanoscale, yet the output is often degraded by artifacts introduced by environmental noise, scanning imperfections, and tip-sample interactions. To address this challenge, a lightweight and fully automated framework for artifact detection and restoration in AFM image analysis is presented. The pipeline begins with a classification model that determines whether an AFM image contains artifacts. If necessary, a lightweight semantic segmentation network, custom-designed and trained on AFM data, is applied to generate precise artifact masks. These masks are adaptively expanded based on their structural orientation and then inpainted using a directional neighbor-based interpolation strategy to preserve 3D surface continuity. A localized Gaussian smoothing operation is then applied for seamless restoration. The system is integrated into a user-friendly GUI that supports real-time parameter adjustments and batch processing. Experimental results demonstrate the effective artifact removal while preserving nanoscale structural details, providing a robust, geometry-aware solution for high-fidelity AFM data interpretation.

Figures (10)

• Figure 1: Overview of the proposed AFM image restoration framework. Raw SPM data are converted to 2D height maps for CNN-based artifact classification. Defective images undergo Smart Flatten, which supports automatic mask-aware or user-guided interactive exclusion, followed by Mask-Guided Restoration via segmentation-based mask expansion, directional inpainting, and localized smoothing.
• Figure 2: CNN-based classification module for AFM artifact detection. The model is based on ResNet-18 pretrained on ImageNet, with only the final residual block and fully connected (FC) layer fine-tuned for AFM data. Four classes are distinguished: Good, Imaging Artifacts, Not Tracking, and Tip Contamination. Evaluation is based on accuracy, precision, recall, and F1-score. The classification stage acts as a decision gate: Good images are directly exported, while corrupted ones are passed to the subsequent restoration pipeline.
• Figure 3: Lightweight semantic segmentation model for AFM artifact mask generation.
• Figure 4: Schematic of the Smart Flatten process.
• Figure 5: Schematic of the AFM Repair GUI workflow. The interface integrates file I/O for raw SPM data, DLL access, and model loading; configurable flattening options (row/column/bidirectional fitting, polynomial order, mask-aware mode); centralized parameter adjustment with default ranges and tuning hints; and a visualization panel displaying original and modified AFM data (2D/3D views, raw artifact mask, and expanded mask).