Automatic recognition method of urban underground silt based on remote sensing image—a case of Anqing City

Underground silt, due to complex and loose compound, is a potential threat in urban infrastructure.Compared with the traditional methods of detecting underground silt, such as geophysical and micro-motion detection technology, remote sensing monitoring has the advantages of wide monitoring range, high efficiency and repeatability. The detection method of remote sensing image change was used to extract the spatial location and area of underground silt in urban area of Anqing. The method was mainly based on the object-oriented image analysis method, first splitting the multi-temporal images separately, and then using the SVM algorithm to classify the land cover. Based on the classification results, the spatial distribution of underground silt was extracted by change detection analysis, which could be defined as the target area or key area for the implementation of physical exploration, so as to detect the depth of the underground silt. Based on the results of two phases of image classification, change detection analysis was carried out to extract the spatial distribution and range of underground silt, and select typical areas for field verification using microtremor detection. The proposed method can provide decision support for urban engineering construction and urban planning. It can delineate the target area or key area for geophysical exploration, and improve the efficiency of geophysical exploration.