UPRE method for total variation parameter selection

Total variation (TV) regularization is a popular method for solving a wide variety of inverse problems in image processing. In order to optimize the reconstructed image, it is important to choose a good regularization parameter. The unbiased predictive risk estimator (UPRE) has been shown to give a good estimate of this parameter for Tikhonov regularization. In this paper we propose an extension of the UPRE method to the TV problem. Since direct computation of the extended UPRE is impractical in the case of inverse problems such as deblurring, due to the large scale of the associated linear problem, we also propose a method which provides a good approximation of this large scale problem, while significantly reducing computational requirements.

[1]  Colin L. Mallows,et al.  Some Comments on Cp , 2000, Technometrics.

[2]  M. Hutchinson.  A stochastic estimator of the trace of the influence matrix for laplacian smoothing splines , 1989 .

[3]  D. Anderson,et al.  Algorithms for minimization without derivatives , 1974 .

[4]  P. Hansen.  Rank-Deficient and Discrete Ill-Posed Problems: Numerical Aspects of Linear Inversion , 1987 .

[5]  Victor Solo.  Selection of regularisation parameters for total variation denoising , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[6]  Gene H. Golub,et al.  Calculation of Gauss quadrature rules , 1967, Milestones in Matrix Computation.

[7]  郑洪斌.  COMPUTING TRACE OF FUNCTION OF MATRIX , 2000 .

[8]  L. Rudin,et al.  Nonlinear total variation based noise removal algorithms , 1992 .

[9]  A. N. Tikhonov,et al.  REGULARIZATION OF INCORRECTLY POSED PROBLEMS , 1963 .

[10]  C. Atkinson.  METHODS FOR SOLVING INCORRECTLY POSED PROBLEMS , 1985 .

[11]  Youzuo Lin.  Application of the UPRE Method to Optimal Parameter Selection for Large Scale Regularization Problems , 2008, 2008 IEEE Southwest Symposium on Image Analysis and Interpretation.

[12]  Per Christian Hansen,et al.  Rank-Deficient and Discrete Ill-Posed Problems , 1996 .

[13]  G. Golub,et al.  Generalized cross-validation for large scale problems , 1997 .

[14]  Gene H. Golub,et al.  Generalized cross-validation as a method for choosing a good ridge parameter , 1979, Milestones in Matrix Computation.

[15]  Yehoshua Y. Zeevi,et al.  Estimation of optimal PDE-based denoising in the SNR sense , 2006, IEEE Transactions on Image Processing.

[16]  Rosemary A. Renaut,et al.  A Newton root-finding algorithm for estimating the regularization parameter for solving ill-conditioned least squares problems , 2009 .

[17]  Guo Hongbin.  COMPUTING TRACE OF FUNCTION OF MATRIX , 2000 .

[18]  Misha Elena Kilmer,et al.  Choosing Regularization Parameters in Iterative Methods for Ill-Posed Problems , 2000, SIAM J. Matrix Anal. Appl..

[19]  Dianne P. O'Leary,et al.  The Use of the L-Curve in the Regularization of Discrete Ill-Posed Problems , 1993, SIAM J. Sci. Comput..

[20]  T. Gasser,et al.  Residual variance and residual pattern in nonlinear regression , 1986 .

[21]  C. L. Mallows.  Some comments on C_p , 1973 .

[22]  Søren Ingvor Olsen.  Noise Variance Estimation in Images , 1993 .

[23]  G. Wahba.  Practical Approximate Solutions to Linear Operator Equations When the Data are Noisy , 1977 .

[24]  C. Mallows.  More comments on C p , 1995 .

[25]  Aggelos K. Katsaggelos,et al.  Parameter Estimation in TV Image Restoration Using Variational Distribution Approximation , 2008, IEEE Transactions on Image Processing.

[26]  Tony F. Chan,et al.  Structure-Texture Image Decomposition—Modeling, Algorithms, and Parameter Selection , 2006, International Journal of Computer Vision.

[27]  C. Vogel.  Computational Methods for Inverse Problems , 1987 .