Object-Oriented Road Information Extraction From High Resolution Remote Sensing Images

Roads,as an important part of the traffic structure,are an essential transportation facility.With the advancement of technology,satellite technology is developing more and more rapidly,from which high-resolution remote sensing images containing more and more information can be obtained,and the information in high-resolution remote sensing images can be reasonably utilized,which has social progress and scientific development.Among them,the road information extracted from the high-resolution remote sensing image can be used for urban planning,data acquisition and target detection.It has a very important meaning in how to extracted road information from the many information contained in the remote sensing image with high efficiency and high precision.Therefore,how to obtain road information from high-resolution remote sensing images with high efficiency and high precision is one of the research hotspots for many years.As can be seen from the relevant literature,the methods for extracting road information from high-resolution remote sensing images include not only self-acting methods and semi-automatic methods,but also pixel-oriented and object-oriented methods.The object-oriented automatic method has become the mainstream of road information extraction research.In this context,the object-oriented road extraction method is selected in this paper.Based on eCognition,the road information in high-resolution remote sensing images is extracted.The main work content is from the following aspects:Firstly,the purpose and significance of this paper are determined.The object-oriented road extraction method is used to extract the road information in high-resolution remote sensing image.Then the characteristics of the remote sensing image are analyzed in detail,and combined with the unique characteristics of the target object road.For example,if the road brightness is high and the shape is slender,the knowledge base for extracting road information is constructed;Secondly,establish the method system of road extraction in this paper and analyze the segmentation method and classification method in the system.The object-oriented region growing segmentation and multi-scale segmentation method in different image segmentation methods are analyzed and compared and get the conclusion that the object-oriented multi-scale segmentation method is choice to segment the image.Then,based on the segmentation,different classification methods are adopted:the threshold method and the nearest neighbor method are used to classify and merge the segmented objects,and finally the road is extracted,and the extraction results are evaluated for accuracy,and the quantitative analysis of road extraction accuracy is realized.Then,using the above-mentioned object-oriented extraction road method,the extraction experiments of urban roads and country roads in high-resolution images are carried out.As a result,it can be found that the object-oriented method is more operability in the road extraction process,which can make full use of the relationship between the image pairs and extract the road information with higher precision.Finally,it is concluded that this study can extract road information from the image with high precision,and provide basis and reference for future urban development and transportation planning,which is of great significance for promoting traffic progress and social development.