The Research Of Sample Database And Interpretation Methods Based On Remote Sensing Images

With the rapid development of geographic big data and artificial intelligence,the intelligence and automation of geographic information mining have been widely used in various fields and have been paid attention by researchers.Remote sensing data has become an indispensable source of data for GIS because it has the advantages of wide space coverage,high spatial and temporal resolution,and easy accessibility.With the number of satellite data sources increased,remote sensing images have undergone great changes.The concrete manifestation is that remote sensing images have more extensive spatial coverage and a smaller time span.Compared with the traditional geographical survey and artificial visual interpretation of geographic information extraction,automatic and intelligent image recognition is the future of remote sensing technology and geographic information mining.However,the automatic extraction of machines has a strong dependence on the sample data,and the quality of the samples directly affects the results of machine automation extraction.In some bad quality images,results of computer extracted show disadvantages such as low precision and poor classification.Based on the existing theoretical knowledge,we built a sample knowledge base using remote sensing images and explored the methods of interpretation.The main contents of the study are as follows:(1)The classification system of the remote sensing image database was determined to facilitate the management and retrieval of the sample database.Based on remote sensing images with high spatial resolution,the interpretation of feature elements was assisted by human interpretation and multiple remote sensing image processing methods to obtain vector samples of training samples.Finally,the accuracy of interpreting samples was verified.(2)Geospatial Data Abstraction Library Open source grid space data conversion library was used for system implementation.Establish the search conditions to determine the system search range,and then generate index files based on the search range search sample data.Finally,match the feature categories listed by the text search criteria to obtain sample slicing data.(3)Application experiment was designed to evaluate the accuracy of the results generated by sample database search.Based on the requirements of the project and the design of the system,we selected Yuqiao and Wuhan as two study areas.The accuracy of interpreting samples was evaluated by using truth ground ROI.The experiment consists of two parts:verified the classification result of the sample in the source remote sensing image;verifiea the classification result of the sample in the remote sensing image of adjacent phases.And results were analyzed.Our research combined with artificial intelligence with remote sensing interpretation in the era of large data,and expects to find the correlation between the spectrum and texture of the various objects in the mass remote sensing image samples.Based on this,we can more efficiently and accurately interpret remote sensing images.The research results show that the sample database based on remote sensing images was very effective in storage,querying,retrieval and management.The samples retrieved by the sample database have high classification accuracy.