Study On Hyperspectral Remote Sensing Image Classification Based On Multiple Feature Fusion

With the rapid development of remote sensing image technology and information technology,the amount of remote sensing image data is obtained rapidly.It is a challenging topic that how to use the computer automatically according to a certain model of the image classification.The traditional method is based on the visual interpretation method to classify the remote sensing image,which requires abundant professional experience knowledge and sufficient outdoor field survey information.This identification method is based on a certain prior knowledge,so it is very difficult to identify with low efficiency.Hyperspectral remote sensing technology is also known as the imaging spectroscopy remote sensing technique that has the characteristics of image information and spectral information fusion,which means each pixel on the hyperspectral image corresponds to a unique spectral curve and applying the unique properties can identify the types of remote sensing images based on spectral reflectance object.The hyperspectral image has the ability of optical properties and spectral recognition,so it has become the focus of extensive research and application in the field of remote sensing image.The application of remote sensing image classification is significant in the research of remote sensing image.Remote sensing image classification is based on single feature classification algorithm which is relatively limited,so in order to improve the accuracy of hyperspectral remote sensing image classification,this paper advocates hyperspectral remote sensing classification method that is based on multi-feature fusion.This paper is divided into two parts.The first part applies Indian Pines hyperspectral images as experimental data.Firstly,the features are extracted from the image,the spatial features and spectral features are normalized,and then Ada Boost classifier algorithm is used to classify the features.The algorithm mainly combines the characteristics of hyperspectral data,texture features and histogram characteristics.Finally,the three classification algorithms of AdaBoost are used to classify single-photon,hyperspectral remote sensing data fusion,and compare the accuracy of classification results.The second part of the experiment is to select two parts of the data to carry out experiments: Indian Pines hyperspectral remote sensing image and the Sand Lake of Ningxia Hui Autonomous Region wetland image.This algorithm is based on RVM algorithm.Firstly,the LBPC algorithm and morphological algorithm are improved to obtain the L-M Algorithm algorithm,then the improved algorithm is embedded into the RVM algorithm,and finally,the Fusion feature classification results are obtained.In order to make the experiment more fully,the SVM and AdaBoost algorithm are used to classify and compare experiments.The experimental results show that the classification accuracy of the fusion algorithm is higher than that of the single feature classification.Therefore,it can be concluded that the fused feature classification results are more reliable and accurate.