Developing a New Model for Description of Micro Size Particles in SEM Images for Safeguards Purposes | ||
| Arab Journal of Nuclear Sciences and Applications | ||
| Article 7, Volume 53, Issue 4, October 2020, Pages 64-71 PDF (886.75 K) | ||
| Document Type: Original Article | ||
| DOI: 10.21608/ajnsa.2020.26198.1340 | ||
| Author | ||
| Sameh ElSayed Shaban* | ||
| Nuclear Safeguards and Physical Protection, Egyptian Nuclear and Radiological Regulatory Authority, Cairo , Egypt | ||
| Abstract | ||
| Scanning Electron microscope (SEM) is a fast screening tool for identification of uranium and thorium particles. In the field of micro-particle analysis for safeguards purposes, estimation of uranium size is necessary information especially for scanning of swipes or environmental samples at nuclear facilities. SEM can be used to visualize micro sized particles of any nuclear material. It gives secondary electron images (SEI) at different magnifications power. In this work, the model JEOL 6510 LV instrument that has capabilities to image samples in a range of magnification from X5 to X300,000 was used. The process of imaging was done at optimum conditions; SS=50, WD=10 and appropriate pressure. The system determines the size of any particle in SEI using internal image. Many trials were performed to accomplish a correlation between micro size and pixels to estimate the size of targeted particles. Here, a mathematical model was developed to describe the relation between micro size and pixels depending on an assumption that is called "numbers to zeros" assumption. The mathematical description was manipulated in three different ways; linearly, exponentially and by using artificial neural network (ANN). The model in its three forms was validated using large numbers of SEIs and comparisons with commercial softwares. The fitting correlation coefficient for linear and exponential manipulation was found to be 0.99997 and 1 respectively. While, the performance of the trained data in ANN was 6x10-10. | ||
| Keywords | ||
| Micro-particles; SEM images; Mathematical model; Artificial neural network | ||
|
Statistics Article View: 471 PDF Download: 353 |
||
