Homotopy Curve Tracking for Total Variation Image Restoration

Year:    2012

Journal of Computational Mathematics, Vol. 30 (2012), Iss. 2 : pp. 177–196

Abstract

The total variation (TV) minimization problem is widely studied in image restoration. Although many alternative methods have been proposed for its solution, the Newton method remains not usable for the primal formulation due to no convergence. A previous study by Chan, Zhou and Chan [15] considered a regularization parameter continuation idea to increase the domain of convergence of the Newton method with some success but no robust parameter selection schemes. In this paper, we consider a homotopy method for the same primal TV formulation and propose to use curve tracking to select the regularization parameter adaptively. It turns out that this idea helps to improve substantially the previous work in efficiently solving the TV Euler-Lagrange equation. The same idea is also considered for the two other methods as well as the deblurring problem, again with improvements obtained. Numerical experiments show that our new methods are robust and fast for image restoration, even for images with large noisy-to-signal ratio.

Journal Article Details

Publisher Name:    Global Science Press

Language:    English

DOI:    https://doi.org/10.4208/jcm.1107-m3423

Journal of Computational Mathematics, Vol. 30 (2012), Iss. 2 : pp. 177–196

Published online:    2012-01

AMS Subject Headings:   

Copyright:    COPYRIGHT: © Global Science Press

Pages:    20

Keywords:    Image restoration Total variation Newton method Homotopy method Correction and curve tracking.

  1. An Iterative Lagrange Multiplier Method for Constrained Total-Variation-Based Image Denoising

    Zhang, Jianping | Chen, Ke | Yu, Bo

    SIAM Journal on Numerical Analysis, Vol. 50 (2012), Iss. 3 P.983

    https://doi.org/10.1137/110829209 [Citations: 27]
  2. Hybrid Variational Model for Texture Image Restoration

    Ma, Liyan | Zeng, Tieyong | Li, Gongyan

    East Asian Journal on Applied Mathematics, Vol. 7 (2017), Iss. 3 P.629

    https://doi.org/10.4208/eajam.090217.300617a [Citations: 4]
  3. Integral Methods in Science and Engineering

    A Finite Element Formulation of the Total Variation Method for Denoising a Set of Data

    Harris, P. J. | Chen, K.

    2013

    https://doi.org/10.1007/978-1-4614-7828-7_12 [Citations: 0]
  4. Efficient Homotopy Method for Total Variation Image Registration

    Zhang, Jin | Chen, Ke | Yu, Bo

    2013 International Conference on Computer Sciences and Applications, (2013), P.655

    https://doi.org/10.1109/CSA.2013.159 [Citations: 0]
  5. A relaxed fixed point method for a mean curvature-based denoising model

    Yang, Fenlin | Chen, Ke | Yu, Bo | Fang, Donghui

    Optimization Methods and Software, Vol. 29 (2014), Iss. 2 P.274

    https://doi.org/10.1080/10556788.2013.788650 [Citations: 17]