Multi-directional illumination for automatic flaw detection on photographic film: The importance of being many

Dark-field illumination is a useful optical method to emphasize structures and impurities in a sample; hence, it can be useful to pinpoint flaws on photographic film. However, in case of silver-based film, the scattering due to the metal particles limits its effectiveness as optical method for flaw detection. The framing of dark-field illumination in a computational photography method based on a set of acquisitions with different illumination directions, improves the performances of flaw detection. The present paper reports the results of the experiment that answers the question “how many different directions of illumination are needed for an effective flaw detection?”. The answer found is “the more numerous is the set of images used, the better is the result”.