Approach for palm print recognition based on two-dimensional Empirical Mode Decomposition and ICA

A novel method based on Two-dimensional Empirical Mode Decomposition(2-D EMD) and Independent Comment Analysis(ICA) is proposed to solve palm print recognition.The adaptive time-frequency localization of 2-D EMD and higher-order statistical independency of ICA II are utilized to extract the palm print features.Firstly,the preprocessed palm print image is decomposed into some Intrinsic Mode Functions(IMFs) by 2-D EMD,and then the palm print feature subspaces of IMF subimages matrix are obtained by a fast fixed-point algorithm for ICA II(Fast ICA II),before which Principal Component Analysis(PCA) is used to decrease the dimensions of input images matrix.Finally,the recognition performance of the integrated method(2-D EMD+ICA II) is tested on the Hong Kong Polytechnic University palm print database.Experimental results show that,compared with ICA II,the proposed method not only can more effectively and accurately extract the palm print features,but also achieves superior Signal-noise-ratio(SNR)of the reconstructed image and higher recognition rate.