Role of Zernike Moments in Hyperspectral Image Classification

Classification of heterogeneous classes present in the Hyperspectral image is one of the recent research issues in the field of remote sensing. This work presents a novel technique that classifies Hyperspectral images that contain number of classes by making use of the image moments. Recently, researchers have introduced a number of neural network models and structured output based methods for classification of these Hyperspectral images they however suffers with the problem of confusion between the classes that are having similar characteristics and hence provides imbalanced solution for the classes with less number of pixels. The polynomial features such as Zernike moments are extracted from the Hyperspectral image and is used for classification. Support Vector Machines with Binary Hierarchical Tree is used for classification of the Hyperspectral data by One Against All methodology. Then, the performance of Zernike moments in Hyperspectral image classification is evaluated.