Application of iterative Jacobi method for an anisotropic di usion in image processing
محورهای موضوعی : Applied Mathematics
M. Knanian
1
(Department of Mathematics, Khorasgan (Isfahan) Branch, Islamic Azad
University, Isfahan, Iran.)
A. Davari
2
(Department of Mathematics, Faculty of sciences, University of Isfahan,
Isfahan, Iran.)
کلید واژه: Image restoration, Anisotropic di usion, iterative numerical method,
چکیده مقاله :
Image restoration has been an active research area. Di erent formulations are e ective in high qualityrecovery. Partial Di erential Equations (PDEs) have become an important tool in image processingand analysis. One of the earliest models based on PDEs is Perona-Malik model that is a kindof anisotropic di usion (ANDI) lter. Anisotropic di usion lter has become a valuable tool indi erent elds of image processing specially denoising. This lter can remove noises without degradingsharp details such as lines and edges. It is running by an iterative numerical method. Therefore, afundamental feature of anisotropic di usion procedure is the necessity to decide when to stop theiterations. This paper proposes the modied stopping criterion that from the viewpoints of complexityand speed is examined. Experiments show that it has acceptable speed without su ering from theproblem of computational complexity.
Image restoration has been an active research area. Di erent formulations are e ective in high qualityrecovery. Partial Di erential Equations (PDEs) have become an important tool in image processingand analysis. One of the earliest models based on PDEs is Perona-Malik model that is a kindof anisotropic di usion (ANDI) lter. Anisotropic di usion lter has become a valuable tool indi erent elds of image processing specially denoising. This lter can remove noises without degradingsharp details such as lines and edges. It is running by an iterative numerical method. Therefore, afundamental feature of anisotropic di usion procedure is the necessity to decide when to stop theiterations. This paper proposes the modied stopping criterion that from the viewpoints of complexityand speed is examined. Experiments show that it has acceptable speed without su ering from theproblem of computational complexity.
[1] P. Perona, J. Malik, Scale space and edge detection using anisotropic
di usion, IEEE Trans. Pattern Anal. Mach. Intell.,12,629-639 (1990).
[2] I. Capuzzo Dolcetta, R. Ferretti, Optimal stopping time formulation of
adaptive image ltering, Appl.Math. Optim. 43, 245-258 (2001).
[3] G. Gilboa, N. Sochen, Y.Y. Zeevi, Estimation of optimal PDE-based
denoising in the SNR sense, IEEE Trans. Image Proc. 15, 2269-2280 (2006).
[4] H. Molhem, R. Pourgholi, M. Borghei, A numerical approach for
solving a nonlinear inverse di usion problem by Tikhonov regularization,
Mathematics Scienti c Journal, Vol. 7, No. 2, 39-54(2012).
[5] P. Mrazek, M. Navara, Selection of optimal stopping time for nonlinear
di usion ltering, Int. J. Comput. Vision. 52, 189-203 (2003).
[6] A. Ilyevsky, E. Turkel, Stopping criteria for anisotropic PDEs in image
processing, J. Sci. Compute. 45, 337-347, (2010).
[7] Z. Wang, A.C. Bovick, H.R. Sheikh , E.P. Simoncelli, A novel kernel-
based framework for facial-image hallucination Structural Similarity, IEEE
Transactions on Image Processing. 13, No. 4, pp. 600-612 (2004).
