At Delft University, Faculty of Applied Earth Sciences, Department of Raw Materials Processing, research is carried out to develop sensor based quality control and sortation systems for recyclable materials and wastes. We develop advanced multi-sensor quality control and sorting systems. A flow of particulate solid materials is observed continuously by means of sensors. The main applications are the sorting and cleaning of:

• Non-ferrous metals
• Glass cullet
• Plastics
• Building rubble

Signal processing theory learns that classification of particles fundamentally improves as the number of independent features increases. The number of detectable features is a compromise between desired accuracy, throughput, and costs of the system.

Sensors extract several features from the detected particles. We apply cameras in combination with dedicated sensors for metal detection, light transmission, and spectral reflection. Classification features are for example size, texture, edge-structure, and colour intensity.

Our approach:

• State-of-the-art sensors, components and parts. Additional customised development when necessary. Examples are cameras, lighting systems, sensors, and computers.
• Modular software system for detection, segmentation and classification of particles. Our system design shortens development time and suits a wide range of materials. (see X-ray image of non-ferrous metals, and scheme of a glass quality control device)
• High processing capacity by optimal use of computing power. (see software)
• Sensors, data-processing, and ejection mechanisms are developed simultaneously.
• The system is prepared for industrial application. (See glass quality control device)