Landslide Extraction from High-Resolution Remote Sensing Imagery Using Fully Convolutional Spectral–Topographic Fusion Network

Abstract

Considering the complexity of landslide hazards, their manual investigation lacks efficiency and is time-consuming, especially in high-altitude plateau areas. Therefore, extracting landslide information using remote sensing technology has great advantages. In this study, comprehensive research was carried out on the landslide features of high-resolution remote sensing images on the Mangkam dataset. Based on the idea of feature-driven classification, the landslide extraction model of a fully convolutional spectral–topographic fusion network (FSTF-Net) based on a deep convolutional neural network of multi-source data fusion is proposed, which takes into account the topographic factor (slope and aspect) and the normalized difference vegetation index (NDVI) as multi-source data input by which to train the model. In this paper, a high-resolution remote sensing image classification method based on a fully convolutional network was used to extract the landslide information, thereby realizing the accurate extraction of the landslide and surrounding ground-object information. With Mangkam County in the southeast of the Qinghai–Tibet Plateau China as the study area, the proposed method was evaluated based on the high-precision digital elevation model (DEM) generated from stereoscopic images of Resources Satellite-3 and multi-source high-resolution remote sensing image data (Beijing-2, Worldview-3, and SuperView-1). Results show that our method had a landslide detection precision of 0.85 and an overall classification accuracy of 0.89. Compared with the latest DeepLab_v3+, our model increases the landslide detection precision by 5%. Thus, the proposed FSTF-Net model has high reliability and robustness.

MAST: An Earthquake-Triggered Landslides Extraction Method Combining Morphological Analysis Edge Recognition With Swin-Transformer Deep Learning Model

“…In addition to SVM and RF, various enhanced machine learning approaches have found application in automated ETLs extraction. For instance, neural-networkbased methodologies, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), can adeptly discern spatial and temporal features of landslides from seismic data, significantly enhancing the precision of landslide extraction [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]. For instance, Ramdhoni et al [14] employed the Smorph method to perform landslide extraction.…”

EMR-HRNet: A Multi-Scale Feature Fusion Network for Landslide Segmentation from Remote Sensing Images

“…Zhang [ 37 ] introduced a landslide detection network model by combining change detection and Multiple Instance Learning (MIL), using the MIL framework to reduce the need for pixel-level samples. Xia [ 38 ] enhanced landslide detection models in retaining high-resolution remote sensing image details by introducing multi-pore pyramid pooling. Ghorbanzadeh [ 39 ] used Dempster–Shafer theory combined with CNN models trained on multiple datasets to improve landslide detection accuracy.…”

ShapeFormer: A Shape-Enhanced Vision Transformer Model for Optical Remote Sensing Image Landslide Detection

“…Chen et al [44] introduced a pretrained CNN for feature fusion and landslide detection. Xia et al [45] introduced atrous spatial pyramid pooling (ASPP) to improve the landslide detection network's ability to preserve details in high-resolution remote sensing images. CNN models can also be combined with multi-temporal images and change detection methods for landslide detection.…”