Extraction of abndoned farm land using multi spectral satellite images

2021 Fiscal Year Final Research Report

• Project/Area Number: 19K06307
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 41040:Agricultural environmental engineering and agricultural information engineering-related
• Research Institution: Ibaraki University
• Principal Investigator: Kinoshita Tsuguki 茨城大学, 農学部, 教授 (10313008)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: リモートセンシング / 画像分類 / 土地利用

Outline of Final Research Achievements

The objective of this study was to develop a method for extracting abandoned farmland using satellite imagery from multiple time periods. To achieve this objective, two tasks were set. The first was to establish a correction method for satellite images taken at different times by different sensors, and the second was to develop a classifier suitable for this task. For image correction, we developed a method that combined several existing methods with improvements. The resulting images were used to extract abandoned farmland, and the results were more accurate than those obtained in previous studies. In the development of a new classifier, a classifier based on rough set theory was developed, but it became clear that the improvement in accuracy was small compared to existing classifiers when multiple bands were used.

Free Research Field

農業工学

Academic Significance and Societal Importance of the Research Achievements

本研究では、目的達成のために二つの課題を設定した。第一の課題である異なるセンサによる異なる時期の衛星画像の補正方法を確立においては、単年度の画像を用いて、高解像度かつ高精度の分類を可能とした。今後の改良を行うことで、実用的な分類となることが期待され社会的意義があるものと考えられる。ラフ集合理論に基づいた分類器の開発では、多バンド化による分類精度の向上が小さいことが判明した。しかし、その原因も明らかとなったことにより、今後の精度向上が見込まれる。この分類器は他の分野にも適用可能と考えられ、学術的な意義を有すると考えられる。