![Sample our Physical Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110819/TOC-Subject-Sample-2021-Physical-Sciences.jpg"})
Abstract
The identification of deep green infection plays an important role in high-quality production of tobacco leaves. However, it is difficult to evaluate the infection level automatically and accurately at present, especially at the asymptomatic stage. In this study, a novel infection identification method for the severity of deep green tobacco infections using portable near-infrared spectroscopy (NIR) and the extreme learning machine (ELM) algorithm is proposed. Firstly, measurements of the deep green leaf infection at different levels were obtained using a portable digital light projection (DLP) NIR spectrometer directly from fresh tobacco leaves without defoliation and any sample preparation procedures in the field. Next, the qualitative and quantitative models were both constructed to characterize the extent of deep green tobacco infection by the employment of an ELM algorithm. The qualitative model was able to automatically identify if a tobacco leaf was infected, while quantitative model was demonstrated to accurately characterize the degree of condition at the asymptomatic stage. These methods are simple, rapid, precise and allow the development of appropriate decisions to precisely control the deep green leaf infection in the field.