-
OLD IMAGE RESTORATION
-
(a)
(b)
-
-
(a) Original image scanned as jpeg from Snow E., Red Star over China, Pelican Publisher, 1972 (size image 852×622 pixels).
-
(b) Restored image.
Values of the parameters entering the transformation:
thrpre=190; itpre=20;
Maskpre=5×5;
l=5; it =20.
The restoration process took up a cpu time of 12.5sec using Matlab 7®
on an
Intel® Core™ 2 Duo Processor T9300.
-
TEXT REMOVAL
-
(a)
(b)
-
(c)
(d)
-
-
(a) Original image with superimposed text (size image 512×512 pixels).
-
(b) Restored image by minimization of the total variation over the inpainting domain, as described in T. Chan and J. Shen 'Mathematical models for local nontexture inpaintings' SIAM J. Appl. Math. 62 (2002) 1019-1043 (obtained by running the
MatLab code by Pascal Getreauer that implements the split Bregman algorithm described in 'The split Bregman methd for L1-regularized problems' by T. Goldstein and S. Osher,
SIAM J. Imaging Sciences 2 (2009) 323-343).
-
(c) Restored image as result of a preprocessing step of our transformation. Number of iterations: 5. Mask used for local interpolation: 5×5.
-
(d) Restored image by applying our transformation applied to the image displayed in (c).
Values of the parameters entering the transformation:
l=5; it =10.
-
(a)
(b)
-
(c)
(d)
-
-
(a) Detail of the original image around the pirate mounth without superimposed text.
-
(b) Detail of the restored image by total variation minimization.
-
(c) Detail of the restored image obtained in the preprocessing step.
-
(d) Detail of the restored image obtained by our transformation.