Study On Land Cover Classification Method Based On Spatial-temporal Remote Sensing Data Fusion

Land are closely related to all aspects of human life,the land use and land cover is the important component of surface biological survival and development system,it is also one of the hot spots and trends in global scientific community.Therefore,it is very important to be able to obtain land cover classification in a timely and accurate manner,which is of great significance.However,land cover classification is a quite complex process.At present,remote sensing technology with its superior conditions that is high efficiency,low cost,in the land cover classification and resource monitoring has been widely used.Land cover classification is also a key task in the field of remote sensing technology.With the rapid development of remote sensing platform,sensor,computer and other related technologies,how to make the land cover classification of remote sensing image can be carried out quickly,accurately and in real time.How to improve the land cover classification accuracy is the hot topic of current research.In fact,we can think that the study of land cover classification can be seen as a study of the classification of remote sensing images.using the multi-band,high temporal resolution and easy-to-obtain advantages of MODIS images combine the high spatial resolution of Landsat-8 images,the data fusion based on Adaboost algorithm is used to classify the land cover in the study area-northeastern Thailand.The main research contents include:1.As the rainy season in northeastern Thailand is rainy and cloudy,the image noise pollution is serious.This paper first deals with the problem of cloud coverage.The MODIS image are denoised by the S-G filtering method respectively to optimize the image quality.In addition,the influence of the Landsat image on the atmospheric radiation is briefly described.In order to improve the image definition and improve the classification accuracy,then made atmospheric radiation correction for it.2.Based on the reconstructed MODIS NDVI timing data,this paper uses the Euclidean distance to measure the similarity of the standard time series curve,so as to classify the MODIS data fuzzily.In this paper,MOD09Q1 NDVI time series data are used to classify the land cover in northeastern Thailand in 2010 and 2015,and the results of classification and area were obtained.Then for the Landsat-8 image,the object-oriented nearest neighbor classification method is classified by fuzzy classification,The membership degree maps of the categories in northeastern Thailand in 2015 are obtained and compared with the results of MODIS classification.3.Based on the Adaboost algorithm,the Adaboost algorithm is used to fuse the MODIS image and the Landsat-8 image,that is,the fuzzy classification accuracy is the basis of the image fusion classification.After two rounds of fusion classification,completed the land cover classification of the study area.The results show that the high temporal phase MODIS data and high-resolution Landsat-8 image fusion,make full use of their advantages,the classification accuracy than a single use of a sensor image obtained by the high accuracy,can achieve a better Classification effect.This study provides effective technical means for the classification of land cover in the areas with uneven distribution of cloudy and rainy land,and got the better performance and prospect.