Real-Time Beam Detection and Tracking From Pinhole Imaging System Based on Machine Learning
| Autor: | Nosych, Andriy, Iriso, Ubaldo |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: | |
| DOI: | 10.18429/jacow-ibic2021-tupp28 |
| Popis: | At ALBA Synchrotron each of the two in-air pinhole imaging systems is able to see several beam spots at once due to specific pinhole grid with 3x3 holes placed in the path of the X-ray fan. Each beam image has its own properties, such as source pinhole aperture size, its Point Spread Function (PSF) and copper filter thickness, all of which impact the electron beam size calculation. Until now, these parameters were applied manually to the pinhole device servers for numerical image analysis, so this semi-manual beam size calculator is subject to frequent adjustments and human monitoring. This study looks at feasibility of training and pointing an Artificial Neural Network (ANN) at image stream coming from pinhole cameras in real time, track all detected beam spots and analyze them, with the end goal to automate the whole pinhole beam image processing. Proceedings of the 10th International Beam Instrumentation Conference, IBIC2021, Pohang, Rep. of Korea |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|