阈值
千分尺
像素
材料科学
人工智能
光学
物理
计算机科学
图像(数学)
作者
Jialong Hao,Wei Yang,Wenjuan Huang,Yuchen Xu,Yangting Lin,Hitesh G. Changela
摘要
NanoSIMS has been the most widely used analytical technique for measuring the elemental and isotopic ratios of micrometer to sub‐micrometer particles, e.g., pre‐solar grains, soil particles, and fog‐helium aerosol particles. Automated sample stage movement combined with particle recognition algorithm is commonly used to improve analytical efficiency. However, due to the effect of sample topography or variation in ion yield rates, the global thresholding method used in previous studies has a low recognition rate. In order to improve the recognition rate, here we have developed a high‐efficiency sub‐micron particle recognition method. This method searches for the threshold that minimizes the intra‐class variance of the local domain (the radius of the local is set to 10 pixels). The central pixel of this circular ROI is then tested against the threshold found for this region. Presolar grains from the Qingzhen (EH3) meteorite and soil organic particles from Oxisol were analyzed. The analytical efficiency was improved by 120‐200% compared with the global thresholding algorithms. This method can be widely applied in automated studies of large numbers of micrometer to sub‐micrometer particles.
科研通智能强力驱动
Strongly Powered by AbleSci AI