Sensor Fusion Based Plastic Bottles Classification System: Machine vision, signal processing and artificial intelligence techniques

Plastic bottles are increasingly becoming a serious environmental problem. They are a major part of the municipal waste due to their low density to volume ratio and non-biodegradable nature. Therefore, intensive research efforts have been made to provide solutions to improve the efficiency of plastic bottles recycling techniques. However, many of these solutions focus on the plastic chemical composition or color classification. This book presents the design and implementation of an automatic plastic bottles classification system based on the combination of color and chemical property. The system has hardware and software architectures. The hardware architecture consists of a computing device, a CCD camera and an NIR spectrometer. The software architecture is developed based on statistical classifiers and sensor fusion techniques. The proposed system testing showed major enhancements in plastic bottles sorting accuracy based on the fusion of camera images and the spectrum signature of plastic bottles. This study should shed some light on a new classification system solution and should be especially useful to professionals in plastic recycling and environment protection industry.