Mining for hidden information in scanning electron micrograph subpixels

Fibics公司是带电粒子束（CPB）显微镜仪器的领导者（通常是 ** 扫描电子显微镜（SEM）和聚焦离子束显微镜（FIB）），并确定了一个 ** 未开发的信息来源，通常提供给用户的图像：他们开发了一种获取和传送包含在标准像素内的“亚时间”信息的方法，以及一个或多个线平均/积分通道。停留时间和行 ** 平均设置被认为是定义CPB扫描的“标准”参数，然而，允许图像 ** 处理功能以时间和空间分辨的方式访问该亚时间像素信息是一项 ** 新的创新，并且以称为“高、宽亚像素”的格式传达。在这个项目中，我们将 ** 应用分类方法，包括卷积神经网络，支持向量机和 ** 统计方法来表征SubPixel信息，并寻找与其他 ** 多传感器数据的相关性，并设计策略来加速基于此 ** 信息的实时分析的采集。

Fibics Incorporated is a leader in the instrumentation of charged particle beam (CPB) microscopes (typically**Scanning Electron Microscopes (SEMs) and Focused Ion Beam Microscopes (FIBs) ), and has identified an**untapped source of information within the images normally made available to users: they have developed a**means to acquire and convey the "subtemporal" information contained within a standard pixel that is acquired**with the standard "dwell time" and one or more line averaging/integration passes. Dwell time and line**averaging settings are considered "standard" parameters to define a CPB scan, however allowing image**processing functions access to this subtemporal pixel information in a time and spatially resolved manner is a**new innovation, and is conveyed in a format referred to as "Tall, Wide SubPixels". In this project, we will**apply classification methods including Convolutional Neural Networks, Support Vector Machines and**statistical methods to characterize the SubPixel information and also look for correlations with other**multiple-sensor data, and design strategies to accelerate acquisitions based on real-time analysis of this**information.