Plastic and organic waste identification using multispectral imaging

World cities face crucial problems with municipal solid waste produced by households and industry activities. Integrated waste management and advanced technologies for waste segregation are needed. Imaging methods and artificial intelligence are preferable for faster and more reliable waste sorting. Waste segregation has some purposes, to reduce waste reaching landfills and incinerations, to maximize recycling waste such as metals and plastics, and to collect organic waste for further use. The efforts to segregate plastic and organic waste from municipal solid waste are crucial. Plastic types such as PET, HDPE, and PP are valuable recycling materials, while organic waste is a potential biogas source. This study aimed to use a multispectral imaging system with ten wavelengths to identify and characterize the reflectance intensities of plastic and organic wastes. Principal Component Analysis (PCA) analyzed and visualized the waste image datasets. The samples were plastics of PET, HDPE, and PP types and organic waste of cartons, paper, and wet organic or vegetable. Each type had 35 pieces. The results show significant differences in reflectance intensities of the sample types, especially at 710 nm. From the average percentage difference in reflectance intensities over all wavelengths, HDPE plastic has a 23% lower in reflectance intensity than PET, 41% higher than colored PP, and 35% higher than transparent PP plastic. HDPE intensities are higher than PET at certain wavelength. The labeled plastic bottles have 11% higher in reflectance intensities than the unlabeled. The organic samples showed higher reflectance intensities than the plastics but varied between the organic waste types. In reflectance intensities, paper is 26% higher than carton waste, 48% than wet organic. PCA results show that multispectral imaging has the potential application in an automatic waste segregation system.