Recovery of nanowire morphology and distribution by a computer vision-assisted approach

Nanomaterial has demonstrated its advantages in a variety of application fields such as biosensors and next-generation computer chips. It is important yet challenging to precisely capture and recover the morphological characteristics of nanomaterial for nanomanufacturing quality control. We propose a computer vision-assisted approach in this paper to recover the structural details of nanowires and their spatial distribution. First, we develop a geometric model for nanowire length measurement and a model calibration method based on statistical fitting to deal with bias. Second, we devise a morphology-based detection algorithm to identify nanowires from Scanning Electron Microscope (SEM) images in the presence of defects, and design a simple yet novel similarity function for correspondence establishment. The identified nanowires are fed into the geometric model to calculate their lengths. The experimental results show that our proposed method can effectively identify nanowires and establish correspondence with high detection accuracy and low false alarms in estimating lengths of a great number of nanowires on a substrate. Our approach can be readily integrated to assist nanomanufacturing and its quality control.