›› 2013, Vol. 26 ›› Issue (11): 71-.

• Articles • Previous Articles     Next Articles

An Improved Method for Multiplicative Noise Removal Based on the Nonconvex Regularization Model

YU Ting,ZHANG Zhenshan   

  1. (School of Science,Xidian University,Xi'an 710071,China)
  • Online:2013-11-15 Published:2013-11-19

Abstract:

In view of the fact that non-convex regularization model is sensitive to noise in removing multiplicative noise and the suppression capability is weak for greater intensity noise,an improved image denoising model is proposed.In the new model,multiplicative noise is first transformed into additive noise by Logarithmic transformation,and Gaussian convolution is then introduced into the regularization term and the fidelity term for smooth image preprocessing and more details of edge so that the edge is located accurately.The new model is robust and can smooth according to the characteristics of the image,and therefore the edges are better preserved.Numerical experiments show that the new method of denoising results has a greatly increase on quantitative indicators and a remarkable improvement on the visual effect.The new method has distinct advantages especially for large noise.

Key words: image denoising;nonconvex regularization term;gaussian convolution;alternating minimization

CLC Number: 

  • TN911.73