J.-H. Zhang and J.F. Böhme (Germany)
Multi-channel Wavelets, Filter Bank, Palm-print Image,Geometrical Deformation, Image Denoising, WaveletDenoising
In this paper, a 3-channel wavelet filter bank algorithm is developed for recovering the palm-print image from its seriously deformed version. The proposed method for correction of geometrical deformation of the palm-print image is essentially akin to the digital re-sampling technique. The so-called decomposition filter bank, consisting of 3 filters, is constructed for decomposition of the palm-print image in the wavelet (time-frequency) domain, and then binary interpolation is performed after the image's reconstruction by the so-called reconstruction filter bank. The QMF method is used to design both types of multi-channel wavelet filter bank. Since Morie stripe introduces large amount of noise in the measured palm print image, the image is de-noised by using an adaptive thresholding algorithm after its geometrical deformation is corrected. Acceptable results have been obtained in a real life experiment on the correction and denoising of severely deformed palm-print image contaminated by unknown source of noise.
Important Links:
Go Back
