Discriminative wavelet packet filter bank selection for pattern recognition
IEEE Transactions on Signal Processing
Hi-index | 0.00 |
A novel de-noising method based on BEMD (Bi-dimensional Empirical Mode Decomposition) and wavelet packet transform-wiener filter was proposed. Firstly, BEMD was applied to decompose the preprocessed palm print image including noise into a group of IMFs (Intrinsic Mode Functions) with different intrinsic time scales, and then the first several IMFs corresponding to high frequency information and noise were de-noised by means of wavelet packet decomposition integrated with wiener filter; finally, the image was reconstructed through adding the processed IMFs and the residual component. Simulation results show that compared with BEMD, wavelet packet threshold de-noising and BEMD integrated with wavelet threshold de-noising, this proposed method can achieve more superior de-noising performance with the lowest MSE and the highest PSNR, which provides a basis for the accurate extraction of palm print features.