The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLII-2/W13
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-2/W13, 469–473, 2019
https://doi.org/10.5194/isprs-archives-XLII-2-W13-469-2019
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-2/W13, 469–473, 2019
https://doi.org/10.5194/isprs-archives-XLII-2-W13-469-2019

  04 Jun 2019

04 Jun 2019

MAPPING ARTIFICIAL TERRACES FROM IMAGE MATCHING POINT CLOUD IN LOESS PLATEAU OF CHINA

J. Na1,2, X. Yang1, X. Fang1,3, G. Tang1, and N. Pfeifer2 J. Na et al.
  • 1School of Geography, Nanjing Normal University, 210023 Nanjing, China
  • 2Department of Geodesy and Geoinformation, Technische Universität Wien, 1040 Vienna, Austria
  • 3School of Environment Science, Nanjing Xiaozhuang University, 211171 Nanjing, China

Keywords: Loess Plateau of China, UAV, Point Cloud, Artificial Terrace, Image Matching

Abstract. The Loess Plateau of China, as one of the most affected areas in the world, suffers from serious gully erosion due to its fragmented terrains and erosional materials. The farmland is terraced, i.e. artificial terraces are widely constructed in this region from 1960s to improve the food productivity. While from late 1990s, a project “Grain for Green” start to change those built artificial terraces from the agricultural use into ecologic areas, helping to conserve water. Mapping the artificial terraces, both their distribution and boundaries, is the basis of monitoring their extent and understanding their ecological effects. The drone-based image matching technology provides a possible solution. In this study, an automatic extraction method for artificial terraces was proposed based on the image-matching point cloud. Firstly, an image-matching point cloud was generated using the Pix4d software. Then the vegetation index and height difference were applied on the original point cloud for positive (non-gully) – negative (gully area) terrain segmentation. After that, edge detection on normal vector difference was performed in the non-gully area to define the ridges of artificial terraces. The case study was performed in a small catchment Wucheng in Shanxi province. A comparison between the manual delineation result and the automatic extraction result indicates our method has a total classification accuracy of 85.8%. The proposed method considers comprehensively of topography and landcover. We conclude that it has a an optimistic potential in loess hilly-gully region with similarly complex terrains and diverse vegetation covers.