Artificial intelligence method developed for classifying raw sugarcane in the presence of the solid impurity

Abstract

An investigation dedicated to evaluating a big issue in biorefineries, solid impurity in raw sugarcane, is presented. This relevant industrial sector requests a high-frequency, low-cost, and noninvasive method. Then, the developed method uses the averaged color values of ten color-scale descriptors: R (red), G (green), B (blue), their relative colors (r, g, and b), H (hue), S (saturation), V (value) and L (luminosity) from digital images acquired from 146 solid mixtures among sugarcane stalks and solid impurity — vegetal parts (green and dry leaves) and soil. The solid mixture of samples was prepared considering desirable and undesirable scenarios for the solid impurity amounts. The outstanding result was revealed by an artificial neural network (ANN), achieving 100% of accurate classifications for two ranges of raw sugarcane in the samples: from 90 to 100 wt% and from 41 to 87 wt%. Low-computational cost and a simple setup for image acquisition method could screen solid impurity in sugarcane shipments as a promising application.