| Remote sensing is used in numerous fields,including geography,land surveying and most Earth Science disciplines(for example,hydrology,ecology,oceanography,glaciology,geology);it also has military,intelligence,commercial,economic,planning,and humanitarian applications.Remote sensing image classification technology is a relatively popular research direction in the field of remote sensing technology.Many researchers spend much time focusing on improving the classification accuracy effectively.In the classification of remote sensing images,there are many kinds of classification methods and different classifiers have different classification principles which lead to different classification results.With the continuous development of multi-classifier combination algorithm,the combination of different classifiers has become a trend.The main contents and achievements of this thesis:1.The two study areas have different distribution of objects.Before image classification,the images were subjected to preprocessing pretreatment such as radiation calibration,atmospheric correction and selecting proper ROI.Using five base classifiers Supervision classifications to process two study areas and comparing the classification accuracy of different algorithms.It is found that the accuracy of maximum likelihood method is the highest in the two classification areas,but the classification accuracy of cultivated land 1 is not the highest.2.In the research,using ENVI and IDL software to process the results of base classifiers with multi-classifier combination algorithm based on decision fusion.Compared with the classification results of base classifiers,the classification overall accuracy of the multi-classifier combination is increased by 2.5%,and the accuracy of the cultivated land 1 in the study area 1 is increased by 15.5%.The method of multi-classifier combination is more flexible than single classifier classification method and can use the characteristics of different classifiers tocompensate the deficiency of single classifiers and improve the accuracy of image classification. |
