DIPSAM: Diffusion Inpainting for Restoring Water-Loss Degradation in Scanning Acoustic Microscopy
Published in Computers & Electrical Engineering (IF = 5.5), 2026
This paper proposes DIPSAM, a diffusion-based inpainting framework that restores water-loss degradation in scanning acoustic microscopy (SAM) images, enabling automatic and reliable inspection of semiconductor and industrial materials.
| Paper | Code |
Abstract
Scanning acoustic microscopy (SAM) relies on a water couplant to transmit ultrasonic waves, and any loss of that couplant during scanning introduces severe localized degradation that corrupts the acquired acoustic image and compromises downstream inspection. This work presents DIPSAM, a diffusion-based inpainting framework that reconstructs the regions affected by water-loss degradation while preserving the surrounding acoustic structures. The restored images recover the structural detail required for reliable defect assessment, improving the robustness of automated SAM-based inspection pipelines.
Key Contributions
- DIPSAM: a diffusion inpainting framework tailored to water-loss degradation in SAM imaging
- Restoration of degraded acoustic regions without disturbing intact structural content
- Improved reliability of automated defect inspection on real SAM acquisitions
Recommended citation: T. N. Nguyen, V. H. M. Doan, L. H. Tran, T. H. Vo, D. Q. Nguyen, T. T. Vo, V. H. Pham, J. Choi, S. Mondal, B. Lee and J. Oh, "DIPSAM: Diffusion inpainting for restoring water-loss degradation in scanning acoustic microscopy," *Computers & Electrical Engineering*, vol. 140, p. 111497, 2026, doi: 10.1016/j.compeleceng.2026.111497.