Computational Photography for Dust and Scratch Detection on Transparent Photographic Material

This article pertains to the digital restoration of motion picture films. A new method for the automatic detection of blemishes on any kind of transparent photographic material (still and moving images, silver-based and dye-based material, etc.) is presented. It consists of an innovative combination of different optical techniques and computational photography. The image layer is a random dispersion of microscopic elements (e.g., silver particles in black-and-white material), and its interaction with light is isotropic. Dust, scratches, and other irregularities of the film surface produce shadows and reflections that depend strongly on the provenance of light. The acquisition of multiple images with different geometries of illumination, and the analysis of the differences among them, is found to be an effective method to emphasize irregularities in the film surface. Moreover, the analysis with polarized light is found to improve the blemish detection. The experiments show that the innovative method effectively detects flaws on photographic film surfaces.

Print ISSN

1545-0279

Electronic ISSN

2160-2492

Published

2014-04

Content type

Original Research

DOI

10.5594/j18399XY