Grading tobacco leaves based on image processing and generalized regression neural network

Abstract

Tobacco quality is determined by its grade and the tobacco leaf grading is mainly based on manual classification, depending on people's senses. The area, perimeter, length, width, colors and so on are the key factors effecting tobacco grades. Almost of them can be shown from the leaf image. So the digital image technology is used to extract the leaf features and a generalized regression neural network is employed to determine its grade. The method of mean influence value is used to move the features which have small. Some tobacco leaves provided are graded by the proposed method. The results show that our method is practicable and effective.