Application of Remote Sensing for Automated Litter Detection and Management

Abstract

The Clean Europe Network (CEN) estimates that cleaning litter in the EU accounts for €10–13 billion of public expenditure every year. The annual budget for managing roadside litter alone, is approximately €1 billion. While local authorities in Northern Ireland and elsewhere have legal requirements to monitor and control litter levels, requirements for compliance are unclear and frequently ignored. Against this background, the overall objective of this research is to develop an integrated management system allowing remote discrimination and quantification of roadside litter. As such, the intention is that local authorities can more effectively meet their statutory requirements with regards to litter management. The research aligns with objectives outlined by the UK Government and CEN in terms of improving litter-related data levels. As plastic containers of type RIC1, Polyethylene terephthalate (PETE), represent one of the most common components of roadside litter, its identification in the natural environment via remote sensing is a key objective. By combining published US Hyperspectral library data and experimental field study results, the initial findings of this research indicate that it is possible to discriminate PETE plastic samples in a grass background using a low-cost multispectral sensor primarily designed for agricultural use. While at an initial phase, the research presented has the potential to have a significant impact on the economic, environmental and statutory implications of roadside litter management. Future work will employ image processing and machine learning techniques to deliver a methodology for automatic identification and quantification of multiple roadside litter types.