Performance of Landsat 8 OLI and Sentinel 2 MSI Images Based on MNF Versus PCA Algorithms and Convolution Operators for Automatic Lithuanian Coastline Extraction

• Published in: SN computer science Vol. 5; no. 3; p. 308
• Main Authors: Gadal, Sébastien, Gloaguen, Thomas
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 09.03.2024; Springer Nature B.V; Springer
• Subjects: Advances on Geographical Information Systems Theory; Aerosols; Algorithms; Annual variations; Applications and Management; Archives & records; Coasts; Computer Imaging; Computer Science; Computer Systems Organization and Communication Networks; Data Structures and Information Theory; Digitization; Environmental Sciences; Geography; Humanities and Social Sciences; Image enhancement; Image filters; Information Systems and Communication Service; Landsat satellites; Machine Learning; National parks; Operators; Original Research; Pattern Recognition and Graphics; Principal components analysis; Remote sensing; Satellite imagery; Sediments; Signal and Image Processing; Software Engineering/Programming and Operating Systems; Spectral resolution; Statistics; Vision; Lithuania; Lithuania; Convolution operators; Coastline recognition; Minimum noise fraction; Principal component analysis; Baltic sea region; Convolution Operators; Spatial statistic; Machine learning; Coastline Recognition; Minimum Noise Fraction; Principal Component Analysis
• ISSN: 2661-8907; 2662-995X; 2661-8907
• DOI: 10.1007/s42979-024-02623-9

Faced with the increasing coastalization in the world, exposing populations and activities to coastal risks, decision-support tools based on the extraction of coastlines by remote sensing have become essential for measuring coastal dynamics and developing future local planning. However, these tools are constrained by factors such as the choice of data or the extraction methods, etc. which influence the reliability of the data produced. The aim of this study is to evaluate the automatic coastline extraction methods using transformation algorithms and image enhancement operators from Landsat 8 OLI and Sentinel 2 MSI satellite data. The estimates are based on the average distances and the average differences in annual variation rates between the automatically extracted coastlines and the reference coastlines digitised manually from orthophotographs. Additional validation data are used, such as the prediction rate of coastal dynamics, the digitisation rate of the study area and an overall margin of error including georeferencing errors. The results show that coastlines modelled using the Principal Component Analysis transformation method with High Pass and Laplacian image enhancement filters generate the best performances. The good accuracy of coastline recognition with Sentinel 2 MSI satellite images coastline could be related to spatial and spectral resolution factors.