Hyperspectral image classification based on Empirical Mode Decomposition

This paper proposes hyperspectral image classification based on EMD (empirical mode decomposition). Each hyperspectral image band is decomposed to its intrinsic mode functions (IMFs) using EMD and classification is done over these intrinsic mode functions. After EMD is performed for each band, new values of each band is expressed as sum of the IMFs which are obtained in high level. Support vector machine (SVM) is used to show the performance of the proposed algorithm. Experimental results show that, using first three IMFs and first four IMFs significantly increases the SVM classification accuracy results compared to original SVM.