Research On Vegetation Change Detection Of High-resolution Remote Sensing Images Based On Map Spot

In recent years,with the rapid development of China’s economy and society,the urbanization process has been accelerating,and the changes in land cover have become more frequent.This has also caused a series of problems such as frequent natural disasters,an increase in the proportion of illegal land use,and serious waste of land resources.The annual national geographical conditions monitoring can grasp the changes of national geographical conditions comprehensively and accurately,and analyze the reasons for its changes and development trends,which is of great significance for the government departments to make scientific decisions and implement sustainable development.The current national geographical conditions monitoring workload is large,the degree of automation is not high,and the work efficiency is low.To this end,in this paper,combined with actual needs,taking vegetation area as an example,exploring the technology and method of surface vegetation coverage change detection integrating multi-source data,and developing change detection tool software to provide a reference for improving the production efficiency and automation of national geographical conditions monitoring.The article selects a certain area of Chongqing,and takes the map spot as the basic unit,takes into account the results of the previous national geographical conditions monitoring vector,uses the two previous and second phases of high-resolution remote sensing images,and conducts automatic change detection research on the vegetation area,and tests the results in the form of tool software.The basic idea of the thesis is:data preprocessing;then research on vector-based image segmentation technology to obtain segmentation patches with strong homogeneity and consistent early and late effects;on this basis,carry out feature optimization research and select The optimal feature group;then the correlation coefficient is used as the similarity measurement criterion to determine the change of the patch,and the pseudo-change of the detection result is deleted by the band ratio feature;finally,the change detection processing software is developed.Experiments show that the missed detection rate of the technical method proposed in this paper is within 10%,which satisfies the primary accuracy requirements of change detection in geographic national conditions monitoring.The main contents of this paper are as follows:(1)Vector-based remote sensing image segmentation technology.In national geographical conditions monitoring,in addition to remote sensing images,the data that can be used for change detection also includes the results of the census data of the geographic conditions of the previous images.It contains data such as surface coverage classification data,geographic national conditions elements,and geographic units.Prior knowledge is helpful to obtain more information for change detection.Therefore,this paper proposes a method for obtaining patches based on the results of geographic national census,which makes full use of the boundary and category information of the vector data.First,the vegetation area on the image is extracted based on the category information for image cropping and segmentation The boundary information is used to multi-scale segmentation of the later image to obtain a more homogeneous segmentation patch;on this basis,the checkerboard segmentation algorithm is used to map the multi-scale segmentation results to the two images to obtain consistent segmentation in the early and late periods segmentation spot.(2)The intergration of multi-feature change detection technology.After completing the vector-based image segmentation,this article first extracts the features of the formed image object,and considers its spectral and texture features,through theoretical research,and selects a certain number of ground features samples to compare their characteristics,and select the best feature group from it,and calculate the correlation coefficient between the image feature groups in the early and late period,determine whether the corresponding patch has changed by setting an appropriate threshold,and finally use the band ratio method to remove the pseudo change information.(3)Software development for change detection tools based on map spot.The software uses C# language,combined with Arc Engine collection tools for development.This tool implements functions such as loading and overlay display of raster and vector data,calculation of feature similarity measure,setting of change threshold,and detection and editing of change information.Combined with four sets of high-resolution remote sensing images in Bishan District of Chongqing in 2017 and 2018,the experimental results show that the method proposed in this paper is used for change detection,and its detection accuracy meets the expected target,and the tool software is also verified The practicability of can effectively improve the degree of automation and work efficiency of detection.