An Optimised Region-Growing Algorithm for Extraction of the Loess Shoulder-Line from DEMs

The positive and negative terrains (P–N terrains) of the Loess Plateau China are important geographical topography elements for measuring degree surface erosion distinguishing types landforms. shoulder-lines an terrain feature in often used as a criterion P–N terrains. extraction shoulder lines is predicting recognising gully head. However, existing algorithms loess areas with insignificant slopes need to be improved. This study proposes regional fusion (RF) method that integrates slope variation-based region-growing algorithm extract based on Digital Elevation Model (DEM) at spatial resolution 5 m. RF introduces different factors into growth standards loess-shoulder lines. First, we employed slope-variation-based build initial set difference between smoothed real DEMs N terrain. Second, improved was generate complete area candidate region terrain, which were fused integrate contours eliminate discontinuity. Finally, identified by detecting edge integrated contour, results exhibiting congregate points or spurs, eliminated via hit-or-miss transform optimise final results. Validation experimental ridges hills Shaanxi Province showed accuracy Euclidean distance offset percentage within 10-m deviation range reached 96.9% compared manual digitalisation method. Based mean absolute error standard values, Zhou’s snake model bidirectional DEM relief-shading methods, proposed extracted highly accurately.