Classification Research Of Surface Objects Based On EO-1 Hyperion Hyperspectral Remote Sensing Images

The classification and recognition of surface objects have important change characteristics,and the classification and recognition of surface objects can reflect the important signs of ecological environment changes in the target area.With the development of remote sensing technology,hyperspectral remote sensing technology has embodied its important role in various fields of society.Hyperspectral remote sensing has many spectral channels and high spectral resolution of nanometer scale,which contains nearly continuous and huge spectral information.Using hyperspectral remote sensing images can classify and study the types of surface objects,and realize the application of hyperspectral remote sensing image data in real life.EO-1,launched in November 2000,is NASA's first Earth Satellite to observe the ground in the NMP program.On this Earth Observation Satellite,there are three kinds of sensors to image the earth,namely,Advanced Land Imager(ALI),Hyperion and Atmospheric Corrector(AC).This paper chooses Hyperion's hyperspectral image as the research object,extracts the features of the local objects,and uses the hyperspectral remote sensing image data to study the classification of the surface objects in this area is of great significance.Based on the current research on the classification of surface objects in hyperspectral remote sensing images,The hyperspectral remote sensing image of Hyperion in a specific period of time is selected to be recognized by supervised and unsupervised classification methods.The whole process consists of three basic steps:firstly,the hyperspectral remote sensing image is preprocessed,which mainly includes the removal of the bad channel information and water vapor band of the image,the repair of bad line,the atmospheric correction in the hyperspectral remote sensing image and the removal of smile effect.Then extract characteristic pixels based on minimum noise transformation and pure pixel index from the preprocessed results,and use spectral library comparison to establish spectral library of ground object classification,and apply different methods to classify and analyze the accuracy of ground object types in the processed images.By comparing the accuracy of four different classification methods,it is determined that the classification method based on spectral Angle has the highest classification accuracy,and the classification accuracy can reach 80%,which can better meet the classification requirements of hyperspectral images.