Abstract: A new method for constructing the nonlinear update-lifting morphologic wavelet is presented. The information of the detail signal is used to modify the scale signal, which also guarantees the perfect reconstruction feature of wavelet transform. The update operator of the update-lifting morphologic wavelet is extended. The generalized update operator is presented, which consists of a series of mathematical morphologic operators filering the detail signals in spatial space. The update-lifting wavelet the using generalized update operator is applied to image denoising. Experimental results show that the lifting morphologic wavelet has better denoising performnace and less edge loss in detail image compared to the traditional wavelet thresholding method, expecially in a low signal to noise ratio.

Key words: mathematical morphology, morphological wavelet, lifting, image denosing