Study On Changes Information Extraction Method And Dynamic Changes Of Land Use

In recent years,the change of land-use has been one of focus of the research on global environmental change.Extracting the information of the change of land-use and analysis it,you can better understand the processes and mechanism of the change of the land-use,then according to these to suggest effective and targeted decisions to adjust human activities,so that land-use is more reasonable and sustainable use.Xinjian County as one of the key development of Nanchang city,changes of land-use in Xinjian County is necessary to study,Xinjian County is divided into up_Xinjian County and down_Xinjian County.This article uses part of Xinjian County(down_Xinjian County)as the study area.Using the method of maximum likelihood,support vector machine method and the hierarchical classification method on classifying the image of Xinjian County in 2013 and 2016,and qualitative analysis and accuracy evaluation of the three types of classification results,and comprehensively compare the results of qualitative and quantitative analysis.Selecting a method that is more suitable for the study area.Analyzing the change of land-use of the classification results.The main research contents are summarized as follows:(1)Image fusion is an important step of the classification in the remote sensing image,the accuracy of fusion seriously affect the accuracy of image classification,so it is very necessary to choose a more suitable fusion method.This paper adopts three methods of image fusion which include Gram_Schmidt Pan sharpening(Gram),HSV sharpening(HSV),PC Spectral sharpening(PC)and selects indicators to evaluate fusion results.Fusion evaluation indexes mainly include mean value,standard deviation,correlation coefficient,the average gradient and the information entropy.By using MATLAB software to compute the index value,the results show that PC Spectral sharpening(PC)is better for the study area;(2)The Paper according to remote sensing image of Multi-spectral information,texture features,normalized difference vegetation index NDVI,modified normalized difference water index MNDWI and so on,using maximum likelihood,support vector machine method and hierarchical classification method to classify two period image of Xinjian County,then qualitative analysis and accuracy evaluation to classification results,filter out a method that is more suitable for the study area.Experimental results show that the classification accuracy of hierarchical classification method is the highest,followed by is support vector machine method,the classification accuracy of maximum likelihood is the lowest,the overall accuracy of the hierarchical classification results in two phases respectively were 95.36% and 95.62%,their Kappa coefficient respectively were 0.9247 and 0.9413.Comprehensive comparison multiple methods of qualitative and quantitative analysis,eventually using the hierarchical taxonomy divides the class into waters,vegetation,bare land and building land categories,followed by support vector machines divided the vegetation into forest land,cropland and grassland.(3)Finally,this article analyzes the range of the change of land-use,moving attitude of single land-use type changes,the change of using spatio-temporal pattern,the rate of land-use,and makes a forecast of dynamic analysis of land-use,by analyzing can get following conclusions: the changes of construction land between 2013 with 2016 is the most fast,changes of bare land and cropland is the biggest,cropland,water and grassland show a downward trend,building land,bare land and forest land show an upward trend.By Markov model found out that land-use in Xinjian County is unreasonable,and the level of development land is low.