| one publication added to basket [367982] | Evaluation of historic and new detection algorithms for different types of plastics over land and water from hyperspectral data and imagery
Castagna, A.; Dierssen, H.; Devriese, L.; Everaert, G.; Knaeps, E.; Sterckx, S. (2023). Evaluation of historic and new detection algorithms for different types of plastics over land and water from hyperspectral data and imagery. Remote Sens. Environ. 298: 113834. https://dx.doi.org/10.1016/j.rse.2023.113834 In: Remote Sensing of Environment. Elsevier: New York,. ISSN 0034-4257; e-ISSN 1879-0704, more Related to:Castagna, A.; Dierssen, H.M.; Devriese, L.I.; Everaert, G.; Knaeps, E.; Sterckx, S. (2024). Corrigendum to “Evaluation of historic and new detection algorithms for different types of plastics over land and water from hyperspectral data and imagery” [Remote Sensing of Environemnt 298 (2023) 113834]. Remote Sens. Environ. 300: 113916. https://dx.doi.org/10.1016/j.rse.2023.113916, more |   |
| Keyword | | | Author keywords | Hydrocarbons/plastic detection algorithms; Plastics on land and water; Hyperspectral remote sensing; Airborne imaging |
| Abstract | Several spectral indices have been proposed in the last decade for remote detection of macroplastics in the environment, however no comprehensive analysis has been provided on the sensitivity of those algorithms to different plastic types and their application to remote imagery over land and water. In addition, algorithm threshold values for plastic detection are typically variable per scene, and evaluation of how thresholds perform in multiple images is scarce. In this study we use a reflectance database of diverse pure plastic and non-plastic materials to evaluate spectral signatures for algorithm design and specificity evaluation, as well as to define optical classes for plastics. Published and new algorithms proposed in this study were then evaluated on three flight lines, comprising different observation geometries, of the Airborne Prism Experiment (APEX) sensor taken over Ostend, Belgium. Our results show that most common plastics cluster into two different categories based on their absorption features in the near infrared (NIR) and shortwave infrared (SWIR): Type 1 includes polyethylene, polypropylene, and vinyl, while Type 2 includes polyester, polystyrene, and acrylic. Nylon (polyamide) forms a separate spectral signature. Most line height algorithms were specific in the sense of mostly detecting plastic of either Type 1 or Type 2, while the evaluated ratio algorithms provided less distinction. All but one of the evaluated algorithms were found to have a threshold value that provided distinction between plastic and non-plastic materials (e.g., wood, concrete, algae, ice, etc.). Only the thresholds of the newly developed algorithms were consistent between the reflectance database and the imagery. Detection of macroplastic between flight lines was more consistent over land than over water. Floating transparent plastic bottles (Type 2) with plastic labels (Type 1) and floating polyethylene bags (Type 1) were only detected in one of the three flight lines. Our results support previous observations on the importance of high spatial resolution to minimize mixed spectral signals within a pixel, and points to the relevance of the thickness of the plastic layer influencing the detection for a given background and mixed signal. We also show that residual errors from atmospheric correction have a higher impact on the performance of ratio indexes than line heights. Our research suggests that the different plastic-specific algorithms can be used as complimentary metrics for plastic remote sensing, potentially enabling plastic optical type classification. |
|
