Feature Description And Recognition Of High-Resolution Remote Sensing Images In Mine Subsidence Area

Mine mining triggers different types of ground disasters such as ground subsidence,ground collapse,landslides,collapses,and ground fissures,which affects residential areas,structures,water resources,and land resources in the mining area to varying degrees,seriously restricting the ecological environment and sustainable development.In recent years,high-resolution ground observation technology has developed rapidly,it has played an increasingly important role in monitoring and management of land use,environment and geological disasters in mining areas.Study the geometry,texture,spectrum,context and other features of typical collapse targets in high-resolution remote sensing images of mine surface subsidence area,to realize the detection and identification of specific collapse phenomena in mine surface subsidence area scenes,and help mine surface subsidence area scene understanding,it has reference value for the monitoring and investigation of subsidence area and restoration and treatment.Summarizes the typical surface phenomena caused by coal mining subsidence in the surface subsidence area of the mine,which mainly includes surface disasters such as collapses,ground fissures,collapses and landslides of different sizes and forms,as well as the relocation or abnormal distribution of settlements,vegetation degradation or changes in vegetation types,land degradation and other surface changes,and describe their image characteristics on high-resolution remote sensing images.The subsidence area of Jinggong No.1 Mine in Pingshuo Mining Area,Shuozhou City,Shanxi Province is selected as the research area.Field information is obtained through field investigations to describe the features of typical collapse targets.Interpretation and analysis of collapse types such as ground fissures,collapse troughs,collapse pits and collapses.The UAV remote sensing image samples of ground fissures,trough collapses,collapse pits,collapses,road pollution and coal mining subsidence reclamation areas are selected respectively,and their color features(grayscale statistics and color histogram)and shape features(Aspect ratio,compactness,rectangularity and shape index)for extraction and analysis.Combined Hough transform,shadow detection and Harris corner detection methods to achieve detection of typical collapse targets in collapse area.The combined method first uses the Hough transform method to detect the edge shape curve of the collapsed pit and trough collapsed area,and then uses the shadow detection method to extract its internal area,and then uses the Harris corner detection method to locate local features points in the image,and analyze the distribution of the extracted feature points.The results show that the method has good application effect on target detection in collapsed pits and trough collapsed areas.However,this combination method also has certain limitations.For collapsing targets with different edge shapes in coal mining subsidence areas,different shape detection templates need to be established,and the extraction of the internal area of the collapsing target can only be done when the solar altitude angle is small and the formation time of the collapse is relatively short.In a short time,the detection effect is better.Pack the image feature extraction algorithms used in the paper,and add other algorithms to develop image feature extraction software.The functions of the image feature extraction software mainly include color statistics feature extraction,color histogram,LBP texture feature extraction,Harris corner detection,shadow detection,FAST feature extraction,SURF feature extraction,Hough line detection,Hough circle detection,Canny edge detection,Sobel edge detection,Gaussian Laplacian(LOG)edge detection and Prewitt edge detection.