Rapid Identification of Seismic Landslides Combining with Object-Oriented and Independent Component Analysis Transformation :A Case of the Ms6.5 Earthquake in Ludian, Yunnan

Taking the typical landslide triggered by the 2014 Ms6.5 earthquake in Ludian County of Yunnan Province as an example, the high-resolution remote sensing image of two phases before and after the earthquake is used, and the object-oriented segmentation technique and change detection based ICA transform are combined. Identify the earthquake-triggered landslide, and obtain the information such as the range and scale of the landslide after the earthquake through the superposition analysis of the feature and the change information of NDVI and slope. The results show that the landslide recognition accuracy by using the integrated method can reach 93.3%, and the error extraction and miss extraction rate are low. The integrated method is simple, fast, with less human intervention and a higher degree of automatic extraction, it can meet the needs of post-disaster emergency and rescue work. Compared with traditional change detection and object-oriented classification algorithms, the integrated method further improves the recognition accuracy of new and old landslide, and human activities such as mining, and can be used for risk analysis, disaster management and disaster relief decisions for earthquake-induced landslides.