With the comprehensive improvement of remote sensing data acquisition technology and ability, remote sensing data shows obvious characteristics of big data. To develop the intelligent analysis and information mining technology for remote sensing big data has become the research front of remote sensing technology. The GF-2 satellite is the first self-developed satellite with submeter resolution in China. Furthermore, it has the advantages of width observation, short revisit period, high radiation precision and high positioning accuracy, which provides high precision, stable and reliable data source for the long-term dynamic monitoring and research of geological hazards in China. This paper selects the GF-2 satellite images of Anhui Xieqiao coal mine collected on January 8th in 2015 as the research data, based on the remote sensing geographical analysis of major geological hazards in coal mine area, the object-oriented image analysis method is used to automatically extract geological hazard induced by coal mining activities. The results show that the spatial distribution characteristics of geological hazard bodies can be effectively identified on GF-2 satellite data, such as the location, scope, shape; the object-oriented automatic extraction method has high accuracy, over 90%, for large seeper subsidence basins, small scale collapse pit and linear ground fissures in the coal mine area; The extraction rules constructed based on the concept of layer-by-layer rejection provide good technical support for the application of GF-2 data in geological hazard investigation and big data analysis in coal mine area, and also provide reference for the extraction of other objects. Howerver, the selection of features and the setting of thresholds need specific analysis according to the situation.