| Autor: |
Venkatraman, S., Begum, S. Sabarunisha, Nithya, K., Sujatha, M., Jayasankar, T., Prakash, N. B., Srinivasan, S., Vivek, S. |
| Zdroj: |
Ocean Science Journal; Dec2024, Vol. 59 Issue 4, p1-17, 17p |
| Abstrakt: |
This study explores the detection and classification of floating marine debris in the Nagapattinam area using Sentinel-2 satellite imagery. Data were enhanced through histogram stretching, facilitating the identification of marine plastic debris. Various indices were derived to facilitate the creation of training sets for six models. Each model underwent five tests using a support vector machine and random forest classifiers. Key spectral bands such as B6 (red-edge 2), B8 (near infrared), and B11 (SWIR 1) were pivotal in the classification process. This study illustrates the effectiveness of SVM and RF algorithms in detecting plastic debris, showcasing true and false detection rates. Spectral profiles of various plastic items, including bags, fishnets, and bottles, were analyzed, revealing strong near-infrared reflectance and low red-edge reflectance. Plastic pixels were categorized based on the percentage of plastic coverage, influencing reflectance intensity. Accuracy assessments for the six tests in both the models are presented, highlighting three significant plastic debris groupings near the coast and one near Vedaranyam, with the remaining detections not classified as plastic debris. An automatic floating plastic recognition system was developed and tested in the Nagapattinam area using the best-performing model, identifying floating plastic with an 80–95% accuracy. The Floating Debris Index was the most crucial for identifying marine floating plastic. The study concludes that the distinct reflectance characteristics of plastic facilitate its identification among marine debris, with model accuracies. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink