Remote Sensing Image Retrieval And Location Method Based On Knowledge Graph

As the advancement of remote sensing technology,more and more remote sensing images with high resolution and high quality can be captured easily.Facing the TBlevel remote sensing image data growth rate every day,how to make full use of these high-quality remote sensing images has very important research significance.In this paper,a remote sensing image retrieval method based on knowledge graph is proposed for the current remote sensing image retrieval which only retrieves several images of the same or similar categories from the massive remote sensing image database.This method will efficiently determine the geographic information such as latitude and longitude of the shooting area according to the content of the remote sensing image,as new attempts and ideas for the storage management and retrieval of remote sensing images.The main research contents of the paper are as follows:(1)Aiming at remote sensing image feature extraction,different local feature extraction algorithms and feature dimensionality reduction models are studied because of the scale and imaging angle changes of remote sensing image.Compare and analyze which feature can better characterize the remote sensing image,which is more suitable for the field of remote sensing image retrieval.(2)In order to improve the efficiency of retrieving resources in the massive remote sensing image database,a remote sensing image knowledge graph was built to reduce the retrieval space.Using the variance of each feature of remote sensing image,the features are rearranged to form a new feature string for remote sensing image,and then the feature string of remote sensing image is sorted.Nodes are used to represent remote sensing images,a chain structure knowledge graph is constructed,and images are retrieved through binary search.Compared with the traditional linear retrieval,this method can effectively improve the efficiency of image retrieval.(3)In order to improve the retrieval efficiency and accuracy of remote sensing images,a knowledge graph of graph structure is proposed.Compared with the general image,each remote sensing image can correspond to a definite geographic coordinate.Therefore,according to the geographical position of remote sensing images and the degree of similarity between remote sensing images,the relationship between nodes is added in the knowledge graph.Retrieval in the graph instead of linear retrieval improves the efficiency of remote sensing image retrieval,and at the same time,it can also locate the geographic information of the area shown in the remote sensing image.(4)In addition,we also complete the remote sensing image knowledge graph through the method of graph node embedding to improve the accuracy of image retrieval.For the knowledge graph node,the new attribute is formed by combining the node’s attribute and the neighborhood information.According to the newly formed node attributes,relationships are added or removed from the knowledge graph.The retrieval accuracy of remote sensing images in the updated knowledge graph is improved to some extent.