Film Restoration 

Aims and constraints

I am involved in a digital film restoration research project since 1994. The aim is to design an automatic restoration system which can significantly reduce the most frequent artifacts damaging old films. 
We thank our industrial partner Centrimage (Laboratoires Neyrac Films, Paris, France) for supplying film material for our research. All such material is subject to copyright.

PhD Students

Laurent Joyeux, PhD, Université de La Rochelle, awarded January 2000.
« Reconstruction de séquences d'images haute résolution. Application à la restauration de films cinématographiques »
Olivier Buisson, PhD, Université de La Rochelle, awarded December 1997.

« Reconstruction de séquences d'images haute résolution. Application à la restauration numérique de films cinématographiques »

Researcher at INA (Institut National de l?Audio-Visuel), Paris, since 1999.

Results

Dust removal
*Techniques : hierarchical recursive bloc matching, polynomial interpolation.
*Results : the clip dustCompare.mpeg (233K) shows the effects of dust removal on a sample of the film ?La belle et la bête? (Jean Cocteau, 1946). You can view an image part of this sequence before Belle22.gif (241K) and after correction Belle22C.gif to see the effects of ourrestoration.

Further samples : 

1.beforedust.mpeg (233K) and after correction dustC.mpeg

2.beforeBelle44.gif (241K) and after correction Belle44C.gif

Line scratch removal

*Techniques : extrema detection, Kalman filtering, Bayesian restoration.

*Results : the clip scratchCompare.mpeg (360K) shows the effects of scratch removal on a sample of the TV film ?Les allumettes suédoises? (1997). You can view an image of this sequence before Allumette300.gif (316K) andafter correction Allumette300C.gif to see the effects of our restoration.

Jitter removal

*Techniques : motion compensation, global motion estimation

*Results : the clip jitterCompare.mpeg (233K) shows the effects of jitter removal on a sample of a ?Frères Lumière? film. Note that the line scratch is also removed from this sequence using a Bayesian technique, see Lumiere30.gif (239K) before and after correction Lumiere30C.gif.

*Further samples : before jitter.mpeg (233K) jitter.avi (807K) and after correction jitterC.mpegjitterC.avi

Flicker removal

*Techniques : motion compensation, global intensity estimation

*Results : the clip flickerCompare.mpeg (233K) shows the effect of flicker removal on a sample of a TV film.

*Further samples : before flicker.mpeg (233K) and after correction flickerC.mpeg

Publications

L.Joyeux, S.Boukir, B.Besserer, O.Buisson. 
Reconstruction of degraded image sequences. Application to film restoration, to appear in Image and Vision Computing Journal.
L.Joyeux, S.Boukir, B.Besserer. 

Film line scratch removal using Kalman filtering and Bayesian restoration, to appear in WACV?2000, IEEE Workshop on the Application of Computer Vision, Palm Springs, California, December 2000.

B.Besserer, S.Boukir. 

La restauration numérique des films cinématographiques, invited paper, Les dossiers de l?audiovisuel, Nb. 93, September-October 2000.

L.Joyeux, S.Boukir, B.Besserer. 

Digital tracking and removal of line scratches in motion pictures, JTS'2000, Fifth Joint Technical Symposium, Paris, France, January 2000.

L.Joyeux, O.Buisson, B.Besserer, S.Boukir. 

Detection and removal of line scratches in motion picture films, CVPR'99, IEEE Int. Conf. on Computer Vision and Pattern Recognition, Vol.1, pp 548-553, Fort Collins, Colorado, USA, June 1999.

O.Buisson, B.Besserer, S.Boukir, F.Helt. 

Deterioration detection for digital film restoration, CVPR'97, IEEE Int. Conf. on Computer Vision and Pattern Recognition, Vol.1, pp 78-84, Puerto Rico, USA, June 1997.

O.Buisson, L.Joyeux, B.Besserer, S.Boukir, F.Helt.

Restauration des documents cinématographiques par des méthodes numériques,3èmes Journées Internationales d'Etudes de l'ARSAG (Association pour la Recherche Scientifique sur les Arts Graphiques), Paris,France, pp.175-184, April 1997.

O.Buisson, B.Besserer, S.Boukir, L.Joyeux

Deterioration Detection in a Sequence of Large Images, IWISP'96, 3rd International Workshop on Image and Signal Processing, pp 593-596, Manchester, UK, November 1996.