Volume XLII-4/W16
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W16, 621–626, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W16-621-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W16, 621–626, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W16-621-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

  01 Oct 2019

01 Oct 2019

UAV-BASED HYPERSPECTRAL DATA ANALYSIS FOR URBAN AREA MAPPING

N. A. Suran1, H. Z. M. Shafri1, N. S. N. Shaharum1, N. A. W. M. Radzali1, and V. Kumar2 N. A. Suran et al.
  • 1Department of Civil Engineering and Geospatial Information Science Research Centre (GISRC), Faculty of Engineering, Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysia
  • 2ORNET Sdn. Bhd.47-4, Jalan SP 2/1, Taman Serdang Perdana, Seksyen 2, 43300 Seri Kembangan, Selangor, Malaysia

Keywords: Airborne, UAV, Hyperspectral, Urban Area, Artificial Neural Network, Support Vector Machine

Abstract. A recent development in low-cost technology such as Unmanned Aerial Vehicle (UAV) offers an easy method for collecting geospatial data. UAV plays an important role in land resource surveying, urban planning, environmental protection, pollution monitoring, disaster monitoring and other applications. It is a highly adaptable technology that is continuously changing in innovative ways to provide greater utility. Thus, this study aimed to evaluate the capability of UAV-based hyperspectral data for urban area mapping. In order to do the mapping, Artificial Neural Network (ANN), Support Vector Machine (SVM), Maximum Likelihood (ML) and Spectral Angle Mapper (SAM) were used to classify the urban area. The classifications involved seven classes: concrete, aluminium, flexible pavement, clay tile, interlocking block, tree and grass. Then, the overall accuracies obtained from ANN, SVM, ML and SAM for 0.3 m spatial resolution images were 92.33%, 85.86%, 83.41% and 46.55% with the kappa coefficient of 0.91, 0.83, 0.80 and 0.38 respectively. Thus, the classification results showed that the powerful and intelligent ANN algorithm produced the highest accuracy compared to the other three algorithms. Overall, mapping of urban area using UAV-based hyperspectral data and advanced algorithms could be the way forward in producing updated urban area maps.