Fast removal of line scratches in old movies

In this paper a fast algorithm for removing line scratches in old movies is presented. It is strongly based on exploiting the defect visibility in the image. To this aim the Weber's law can be applied to coefficients of an over-complete wavelet representation of the degraded image. The intensity of the defect, which is represented as a light diffraction effect, is then attenuated in the vertical and approximation sub-bands till the minimum threshold of visibility is reached. The experimental results are very satisfying: the image is completely recovered without local artifacts or annoying smoothing effects.

[1]  S. Mallat A wavelet tour of signal processing , 1998 .

[2]  Lucia Maddalena Efficient methods for scratch removal in image sequences , 2001, Proceedings 11th International Conference on Image Analysis and Processing.

[3]  Vittoria Bruni,et al.  Scratch detection via underdamped harmonic motion , 2002, Object recognition supported by user interaction for service robots.

[4]  Anil Kokaram,et al.  Line scratches detection and restoration via light diffraction , 2003, 3rd International Symposium on Image and Signal Processing and Analysis, 2003. ISPA 2003. Proceedings of the.

[5]  Bernard Besserer,et al.  Film line scratch removal using Kalman filtering and Bayesian restoration , 2000, Proceedings Fifth IEEE Workshop on Applications of Computer Vision.

[6]  Lonnie C. Ludeman,et al.  Fundamentals of Digital Signal Processing , 1986 .

[7]  Guillermo Sapiro,et al.  Image inpainting , 2000, SIGGRAPH.

[8]  Anil C. Kokaram,et al.  Motion picture restoration - digital algorithms for artefact suppression in degraded motion picture film and video , 2001 .

[9]  Vittoria Bruni,et al.  A generalized model for scratch detection , 2004, IEEE Transactions on Image Processing.

[10]  Ali Tabatabai,et al.  Sub-band coding of digital images using symmetric short kernel filters and arithmetic coding techniques , 1988, ICASSP-88., International Conference on Acoustics, Speech, and Signal Processing.