Design And Implementation Of Urban Remote Sensing Image Statistical Analysis System

Urban spatial planning is of great significance to the construction of ecological civilization and sustainable development in China.At present,spatial planning work is mainly based on manual interpretation and semi-automatic analysis of multi-band images.Due to the limitation of image resolution,the accuracy of analysis results is not enough to meet the needs of current planning work.In order to improve the accuracy of remote sensing image(RSI)interpretation and realize automatic analysis,an urban RSI statistical analysis system based on image semantics segmentation technology has been designed.The design and implementation of the system are as follows.Firstly,the extraction and unification of the original RSI’s geographic information is realized,and images are sliced into tiles to provide basic data and coordinate information for subsequent work.Then,the objects of research are roughly classified and subdivided according to the targets of the subject.Training data is selected and labeled according to the requirements of data balance.The architecture of Unet is improved based on the actual demand to realize image semantics segmentation of construction land,ecological land and cultivated land in map tiles.The segmentation results provide data support for the subsequent calculation.Finally,based on the segmentation results and related calculation formulas,some algorithms are implemented to analyze the scale,morphology and vectorization of Wuhan.The results are visualized to visualize the changes of urban space and land-using pattern in Wuhan in recent years.The thesis completes comprehensive functional and performance testing of the system.Results show that all modules of the system can work satisfactory,and geographic information and position are accurate.The MPA and MIOU of the semantic segmentation model is above 89% and 71% respectively.All the results of calculation,visualization and analysis are in line with expectations.