Study On Extraction Of Land Cover Information In Northeast Border Based On Texture Features Of YG-24 Satellite Data

With the continuous development of human society,the depth and breadth of human land development and utilization are also expanding,resulting in the change of land cover nature faster and faster,and the impact is more and more,so the research on land cover has gradually become the focus of research.As a technical means,remote sensing has great advantages in large-scale land cover monitoring and texture features can reflect the characteristics of remote sensing image spatial structure,so it has been paid more and more attention in the research of land cover information extraction.As a new type of domestic high-resolution military satellite,the remote sensing satellite series has high image resolution and high image quality.From yg-01 launched in 2006 to yg-30 launched in 2020,it has formed a satellite series with a huge amount of data,which has been applied in the scientific research,general survey of natural resources,prevention and control of geological disasters.This article takes the northeast border area on both sides of Yalu River in Linjiang City,Baishan City,Jilin Province as a research area,uses the new type of domestic high-resolution military satellite data yg-24 image,uses the most widely used gray level co-occurrence matrix,differential box dimension method and wavelet packet analysis in texture analysis,by selecting a experimental area,to extracting the texture features of yg-24 image.Texture feature is used to assist the extraction of land cover information,and human-computer interaction interpretation results are used as the standard for accuracy evaluation,the best combination of three methods for land cover information extraction in the new type of domestic high-resolution military satellite data yg-24 image is determined.The best combination of parameters is used to extract land cover information of the whole study area of northeast border.The main achievements are as follows:(1)Using MATLAB,the texture features of yg-24 image are extracted by the method of gray level co-occurrence matrix,differential box dimension method and wavelet packet analysis method,and the texture feature image are formed,which are used for land cover information extraction.(2)Based on yg-24 image,by calculating the classification accuracy of feature value,energy value and land cover information extraction under different parameter combinations,the optimal parameters combination of gray level co-occurrence matrix,differential box dimension method and wavelet packet analysis method in texture feature extraction are determined when the texture features are used to assist the extraction of land cover information,including the best extraction band,the best extraction window and the extraction direction.(3)By using the best combination of parameters and the texture features obtained by three different texture feature extraction methods,this paper extracts the land cover information of the whole study area in the northeast border,The classification accuracy and kappa coefficient are calculated by using the human-computer interaction interpretation corrected by field verification points and high-resolution Google Earth image,then makes the corresponding accuracy evaluation and comparative analysis,and studies the application effect of yg-24 image in the land cover information extraction.