Automated particle detection in Atomic Force Microscopy images of environmental films using two novel algorithms

Reliable particle identification in Atomic Force Microscopy (AFM) images is challenging, particularly for environmental films—surface layers that accumulate diverse nanoscale particles outdoors. The commonly used Igor Pro™ software requires extensive manual input, introducing user variability while producing merged clusters and zero-area artifacts. By implementing steps such as the white top-hat transform and Adaptive Gaussian Threshold, this work presents two fully automated algorithms for AFM particle detection. One emphasizes speed, reducing computational time by an order of magnitude, while the other prioritizes accuracy, improving the F1 score by over 40% compared to Igor Pro™. Both approaches eliminate common artifacts and allow for reproducible, high-throughput particle analysis. These results demonstrate that automated algorithms can significantly advance AFM-based studies of films, bridging the gap between complex, resource-intensive and simple, error-prone methods.