Hyperspectral Imaging for e-waste Material Identification

Due to swift changes in consumer demand and technological advancements, electronic products are becoming obsolete at a high rate. This generates a huge amount of electronic waste with a huge potential for recycling. However, on the other hand recycling of electronic waste is becoming challenging due to its diverse and constantly changing material composition. Thus, non-destructive material characterization has become significantly important in order to recover material efficiently. Therefore, this study illustrates the use of non-destructive visible near-infrared Hyperspectral Imaging (VNIR-HSI) technique to identify material accurately in combination with machine learning algorithms such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN) and Neural Network, without using any additional features from any third apparatus and normalization techniques. Combined with an auto-labelling step, not only the material identification is achieved with an overall accuracy of 93.7%, but the processing time is highly reduced and easily automatable.