Semi-Automated Shoreline Extraction in Satellite Imagery and Usage of Fractals as Performance Evaluator

Abstract

Abstract-Analysis of shoreline detection has importance in many investigations undertaken by coastal zone and coastal management studies. These studies require tracking changes in shorelines to reach many objectives such as detecting erosion and land mass movements, discriminating land and sea and etc. At the same time shorelines are important features to display dynamic nature of Earth's surface.In this paper a novel shoreline extraction method and use of fractals as a performance evaluator are proposed. As a first step of shore line extraction, blurring is done on shoreline image to reduce noise. Then variance map calculation and thresholding are applied. In second stage, a series of morphological binary image processing techniques are performed. After user feedback, boundary of the resulting connected component is extracted. Performance evaluation of the proposed method is done by using fractal values. Evaluation is done by matching calculating fractal values of extracted lines and fractal values of handrawn shorelines. A high correlation has been seen between fractal values of computed and handrawn shorelines.Considering the ability of fractal geometry to express natural entities, fractal dimension is contributed as a performance metric.

Tracking the effects of the long‐term changes on the coastal archaeological sites of the Mediterranean using remote sensing data: The case study from the northern shoreline of Nile Delta of Egypt

“…For the shorelines that were extracted from the Landsat and Sentinel2 images, the thresholding method is used depending on the near-infrared (NIR) band due to its efficiency and high accuracy in mapping the position of wetlands (Custom-Scripts, n.d.). Furthermore, for every satellite image separately, the threshold values extracted from the NIR band were used to separate the land cover in the satellite image into two classes: the class that contains the pixels of higher values, which were classified as land area (based on the infrared bands in the sensors; Landsat4, Land-sat5 and Landsat7 [B4]; Landsat8 [B5]; and Sentinel2 [B8]), and the pixels with fewer values, which were classified as waterbodies (Altinuc et al, 2014). Once the bands were converted into land and waterbodies, the separator line between them is performed as the shoreline.…”

Análise Espaço-Temporal E Modelagem Preditiva Da Linha De Costa Do Estado Do Ceará

“…O banco de dados construído com o MNDWI, para cada ano, foi utilizado para extração semiautomática das linhas de costa a partir do método de binarização de Otsu (Otsu, 1975). O método de Otsu permite determinar o valor ideal de threshold que separe os elementos de fundo e de frente da imagem em dois clusters principais, nesse caso é possível determinar o limiar de separação ideal entre os valores correspondentes a terra e água (Altinuc et al, 2014;Yang et al, 2014;Ghorai & Mahapatra, 2020). Como resultado tem-se uma imagem binarizada, que pode ser convertida para um arquivo vetorial de polígono ou linha de forma automática utilizando o ArcGIS 10.5.…”