chambolle.chambolle¶

chambolle.chambolle(ut, Dx=None, Dy=None, mu=1e-05, dt=0.25, itmax=10000, tol=0.001)[source]¶

Solve the nonlinear TV-denoising problem using Chambolle’s projection algorithm

Parameters:

ut : NumPy 2darray Raw grey-scale image to denoise. Note that ut has to be square! Dx : NumPy/SciPy matrix Discrete derivative operator in direction x (forward differences are recommended). Note that if ut is N-by-N then Dx has to be N**2-by-N**2! Dy : NumPy/SciPy matrix Discrete derivative operator in direction y (forward differences are recommended). Note that if ut is N-by-N then Dy has to be N**2-by-N**2! mu : non-negative float Regularization parameter in the TV functional. Note that mu >= 0 has to hold! dt : positive float Pseudo time step. Note that dt has to satisfy 0 < dt < 1. itmax : positive integer Maximal number of Chambolle iterations. Note that itmax has to be > 0! tol : positive float Error tolerance in Chambolle iterations. Note that tol has to satisfy 0 < tol << 1!

Returns:

u : NumPy 2darray Denoised Chambolle-image. Has the same dimension as the input image ut. p : NumPy 3darray Dual variable (“Chambolle-edge-set”). If ut is N-by-N then p is 2-by-N-by-N, i.e. a tensor.

References

[R5]	A. Chambolle. An algorithm for total variation minimization and applications. Journal of Mathematical Imaging and Vision, 20(1-2):89-97, 2004.