Coastline Monitoring

Objective

Advancing coastal management and climate change monitoring through improved coastline extraction from satellite imagery

Team

This is a collaborative work amongst:

Description

Coastline extraction from satellite imagery is a crucial task that is relevant to several fields such as environmental monitoring, disaster management, and urban planning. The accuracy and efficiency of this process depend on the effectiveness of the methods used for extraction. To improve these methods, this research aims to evaluate and compare different approaches for extracting coastlines from satellite images.


The research will consider various edge detection algorithms and spectral bands/indices to evaluate their effectiveness using a range of metrics such as correlation, mutual information, and figure of merit (FOM). Additionally, this study will investigate pre-processing steps such as histogram equalization and Gaussian blur, as well as feature engineering methods to improve the robustness of the algorithms.



(From left to right) Sample satellite image, along with the corresponding binary map showcasing the distinction between water and land. We also show the various channels including, RGB (red green blue), NIR (near infrared), NDVI (Normalized Di erence Vegetation Index), and other water indices (NDWI, MNDWI, AWEIsh, WI2015, S2WI, WI2).


Moreover, this research will release open-source code and a new dataset of high-resolution satellite images to enable future research in this field. The dataset will contain images of different coastal environments and will be labeled with ground truth information, enabling supervised learning approaches.


The findings of this research have the potential to make a significant impact on various fields, such as coastal management, climate change monitoring, and disaster response planning. By contributing to the development of more accurate and efficient methods for coastline extraction from satellite imagery, this research can enable improved decision-making, planning, and response in these fields.

Results

Please refer to the publications.