A Real-Time Target Tracking Method for SEM Images Based on Template Matching

摘要

Micro-nano materials and their manufacture have become increasingly commonly used in the field of chip manufacturing as nanotechnology has advanced. In this area, the use of carbon nanotubes (CNTs) to build high-performance three-dimensional micro-nano devices has gotten a lot of interest. The micro-nano manipulation robot system built in the scanning electron microscope (SEM) is currently used to assemble carbon nanotube devices. Secondary electron detectors in SEMs are used to record secondary electron scanning pictures as a perceptual tool for automated assembly visual guidance. This paper describes how SEM images are used to provide visual guidance for the assembly of CNTs. SEM image noise sources and components were investigated. A carbon nanotube image tracking system based on template matching with scale invariance and rotation invariance was suggested, based on noise components and denoising. Its accuracy error is not more than 6 pixels and the frame rate reaches 26.39 FPS which meets the real-time requirements.