Object-based image analysis for extracting regional slope units and its application

The selection and quality of the evaluation unit directly affect the rationality of the evaluation results of regional landslides, in which the slope unit as the evaluation unit can better describe the landform characteristics and topographic changes, and can better reflect the natural environmental conditions of landslides. However, there is currently no optimization method for slope unit partitioning, which restricts the application of slope units in regional landslide disaster evaluation. Therefore, this paper proposes a R-MSS (Refined Multi-Scale segmentation) method based on object-based image analysis (OBIA). Typical mountainous areas in southwest China and hilly areas in southeast China were selected and compared using hydrological analysis, curvature watershed, and MSS methods. The results show that the R-MSS method can effectively extract large-area slope units without a lot of manual modification. The divided slope units have uniform size, small spatial difference and good morphological effect. The overall effect is significantly better than other methods, especially in the high and steep mountainous areas with complex and changeable terrain, its segmentation performance is especially outstanding. The research results can serve for regional landslide hazard evaluation and improve the reliability of regional landslide hazard analysis results to a certain extent.