Research And Application On Building Extraction Of High Resolution Remote Sensing Image Based On Object-oriented

High resolution remote sensing images with high resolution,can fully display the details of the surface information.The efficient and accurate acquisition of surface information from high-resolution remote sensing images has become a hot-spot in the field of remote sensing.The building is an important feature in the urban area,and its identification and extraction are of great significance and application value to the digital city construction and urban planning.Traditional pixel-based target recognition and extraction methods are no longer applicable in high-resolution remote sensing images.Object-oriented classification technology has become a major trend.Aiming at the problem of object-oriented's image over-segmentation,the blindness of feature selection and the high-resolution remote sensing image building extraction based on different data sources,firstly,this paper studies edge detection algorithm based on the Canny,detects the boundary of the boundary with obvious contour,and takes the edge detection result as a spectral layer to assist multi-scales segmentation.then,the Maximum Correlation Minimum Redundancy feature selection algorithm based mutual information and SVM classifier are used to select and analyze the image object characteristics.Finally,the spatial relationship between buildings and shadows on high-resolution remote sensing images and the thematic features of buildings are analyzed,and the characteristics of spatial relations and nDSM feature are introduced,and combined the Maximum Correlation Minimum Redundancy feature selection algorithm with SVM classification technology,compared with different data sources in the case of high resolution remote sensing image building extraction method.The experimental results show that the multi-scale segmentation method and the Maximum Correlation Minimum Redundancy feature selection method can improve the over-segmentation of the image and overcome the blindness of the object feature selection in object-oriented classification,At the same time,by introducing spatial relations and nDSM The thematic features can solve the problems of buildings and squares and parking lot and other places to a certain extent,can effectively improve the accuracy of building extraction.