Land-Use Mapping Using Orthophoto Obtained From SenseFly EbeeX UAV Images and Object-Based Support Vector Machine

Akar A.

The 13th International Scientific Research Congress, Ankara, Türkiye, 11 - 12 Mart 2022, ss.147-154

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Ankara
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.147-154
  • Erzincan Binali Yıldırım Üniversitesi Adresli: Evet

Özet

Land-use maps are widely used in many fields, especially in agriculture, soil management, urban

planning, environmental research. In especial, unmanned aerial vehicles (UAVs) provide advantages in

land-use mapping in terms of low mapping costs, obtaining images at the desired time and period, and

providing images with high spatial resolution. This study aims to produce a land-use map with the

orthophoto obtained from the SenseFly eBeeX UAV images and the object-based Support Vector

Machine (SVM) classifier. Firstly, the orthophoto image was segmented with multiresolution

segmentation. Then, the segments were classified using object-based SVM and seven land-use classes.

According to a result of the accuracy analysis, the overall classification accuracy of the thematic image

was calculated at 98.10%. In addition, kappa values, quantity, and allocation disagreement/agreement components supported this result. The results showed that a land-use map can be obtained with high accuracy from the orthophoto produced with the SenseFly eBeeX UAV.