The NRC Publications Archive is now operational; however, not all the features of the site are available at this time.
The following features remain unavailable:
- Viewing/Downloading of full text publications
- Author browse feature
- Affiliation/institute/portfolio search
- NRC Publication Archive statistics
NRC is currently working to restore these features and we will update this notice as these features become available. Thank you for your patience.
Application of the Hough transform for the automatic determination of soot aggregate morphology
; Grishin, Igor
; Thomson, Keven
; Migliorini, Francesca
Sloan, James J.
NRC Institute for Chemical Process and Environmental Technology; National Research Council Canada
Environment Monitoring Technologies; Technologies de surveillance de l'environnement
LII - Nanoparticle Diagnostics and Characterization; LII - Diagnostic et caractérisation des nanoparticules
Environment Monitoring Technologies Program; Programme des technologies de surveillance de l'environnement
We report a new method for automated identification and measurement of primary particles within soot aggregates as well as the sizes of the aggregates and discuss its application to high-resolution transmission electron microscope (TEM) images of the aggregates. The image processing algorithm is based on an optimized Hough transform, applied to the external border of the aggregate. This achieves a significant data reduction by decomposing the particle border into fragments, which are assumed to be spheres in the present application, consistent with the known morphology of soot aggregates. Unlike traditional techniques, which are ultimately reliant on manual (human) measurement of a small sample of primary particles from a subset of aggregates, this method gives a direct measurement of the sizes of the aggregates and the size distributions of the primary particles of which they are composed. The current version of the algorithm allows processing of high-resolution TEM images by a conventional laptop computer at a rate of 1–2 ms per aggregate. The results were validated by comparison with manual image processing, and excellent agreement was found.