论文信息 - SPATIO-TEMPORAL POINTWISE ADAPTIVE DENOISING OF VIDEO : 3 D NON-PARAMETRIC REGRESSION APPROACH

SPATIO-TEMPORAL POINTWISE ADAPTIVE DENOISING OF VIDEO : 3 D NON-PARAMETRIC REGRESSION APPROACH

An adaptive spatio-temporal algorithm for video denoising is presented. The local polynomial approximation (LPA) is exploited in order to design 3D directional Þltering kernels. For each speciÞed direction in the 3D space-time domain, an optimal scale (size of the kernels support) is selected using the intersection of conÞdence intervals (ICI) rule. In this way a pointwise adaptive spatio-temporal estimator is constructed. Experimental results show an advanced performance of the proposed method with a good noise attenuation and nearly perfect edges and change-point preservation.

Alessandro Foi | Vladimir Katkovnik | Karen Egiazarian | Chiara Ercole

[1] J. Astola,et al. A novel anisotropic local polynomial estimator based on directional multiscale optimizations , 2004 .

[2] KatkovnikVladimir,et al. Adaptive Window Size Image De-noising Based on Intersection of Confidence Intervals (ICI) Rule , 2002 .

[3] A. Goldenshluger. On Spatial Adaptive Estimation of Nonparametric Regression , 2004 .

[4] E. Mammen,et al. Optimal spatial adaptation to inhomogeneous smoothness: an approach based on kernel estimates with variable bandwidth selectors , 1997 .

[5] Jaakko Astola,et al. Anisotropic local likelihood approximations: theory, algorithms, applications , 2005, IS&T/SPIE Electronic Imaging.

[6] Jianqing Fan,et al. Local polynomial modelling and its applications , 1994 .

[7] Vladimir Katkovnik,et al. A new method for varying adaptive bandwidth selection , 1999, IEEE Trans. Signal Process..

[8] Jaakko Astola,et al. Directional varying scale approximations for anisotropic signal processing , 2004, 2004 12th European Signal Processing Conference.