Land Cover Mapping from Colorized CORONA Archived Greyscale Satellite Data and Feature Extraction Classification

Athos Agapiou

doi:10.3390/land10080771

Abstract

Land cover mapping is often performed via satellite or aerial multispectral/hyperspectral datasets. This paper explores new potentials for the characterisation of land cover from archive greyscale satellite sources by using classification analysis of colourised images. In particular, a CORONA satellite image over Larnaca city in Cyprus was used for this study. The DeOldify Deep learning method embedded in the MyHeritage platform was initially applied to colourise the CORONA image. The new image was then compared against the original greyscale image across various quality metric methods. Then, the geometric correction of the CORONA coloured image was performed using common ground control points taken for aerial images. Later a segmentation process of the image was completed, while segments were selected and characterised for training purposes during the classification process. The latest was performed using the support vector machine (SVM) classifier. Five main land cover classes were selected: land, water, salt lake, vegetation, and urban areas. The overall results of the classification process were then evaluated. The results were very promising (>85 classification accuracy, 0.91 kappa coefficient). The outcomes show that this method can be implemented in any archive greyscale satellite or aerial image to characterise preview landscapes. These results are improved compared to other methods, such as using texture filters.

Highlights

Land cover mapping is considered one of the most well-documented research areas of remote sensing science [1,2]
At the end of a classification process, the satellite image is labelled into the pre-defined thematic land use classes, while the overall results are evaluated via different classification metrics
This paper aims to present a methodology where historical land cover maps can be retrieved from CORONA imagery using deep learning colourisation techniques, feature extraction, and vector machine classification

Summary

Introduction

Land cover mapping is considered one of the most well-documented research areas of remote sensing science [1,2]. Numerous applications have been presented in the past dealing with this topic [3,4,5,6]. The majority of these studies focus on the exploitation of multispectral and hyperspectral data sets to generate land cover maps [7,8,9,10]. Land cover maps are considered essential for studying diachronic landscape changes that can promote sustainability. Land cover maps can be used for monitoring natural hazards like floods and soil erosion. [19] used Landsat images to deliver land cover maps to characterise the 2014 flood of the Indus River in Pakistan

Objectives

Methods

Results

Discussion

Conclusion