期刊
SCIENTIFIC REPORTS
卷 12, 期 1, 页码 -出版社
NATURE PORTFOLIO
DOI: 10.1038/s41598-022-15486-y
关键词
-
资金
- Swiss National Foundation [200021_197107]
- Swiss National Science Foundation (SNF) [200021_197107] Funding Source: Swiss National Science Foundation (SNF)
A low-dose imaging technique is proposed, where recognition is used instead of recording high-resolution images to verify structural hypotheses. By detecting only a few scattered particles, a structural hypothesis can be confirmed with a probability exceeding 95%. This technique can be applied to optical character recognition and minimizing radiation damage in sensitive materials.
A low-dose imaging technique which uses recognition rather than recording of a full high-resolution image is proposed. A structural hypothesis is verified by probing the object with only a few particles (photons, electrons). Each scattered particle is detected in the far field and its position on the detector is analysed by applying Bayesian statistics. Already a few detected particles are sufficient to confirm a structural hypothesis at a probability exceeding 95%. As an example, the method is demonstrated as an application in optical character recognition, where a hand-written number is recognized from a set of different written numbers. In other provided examples, the structural hypothesis of a single macromolecule is recognized from a diffraction pattern acquired at an extremely low radiation dose, less than one X-ray photon or electron per angstrom(2), thus leaving the macromolecule practically without any radiation damage. The proposed principle of low-dose recognition can be utilized in various applications, ranging from optical character recognition and optical security elements to recognizing a certain protein or its conformation.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据