NI assists in restoration of Holocaust testimonials

Boston, MA. National Instruments chose The Vision Show held here this week to highlight a smart camera inspecting candles, an FPGA-based system counting match heads at high frame rates, and an NI Compact Vision System offering tight I/O synchronization in an industrial environment.

One application, however, wasn't highlighted in the exhibit booth, but NI spokesperson Carlton Heard called attention to it nonetheless: the restoration of archival film with testimonials from Holocaust survivors.

Writing in “Sweet Apps: Restoring Holocaust Testimonials with NI Vision Builder,” NI's Laura Arnold writes, “In the 1990s, the USC Shoah Foundation’s Institute for Visual History and Education recorded over 52,000 interviews with Holocaust survivors and eye-witnesses. Unfortunately, when they went back to the tapes in 2012 to digitize them, about 5% were damaged and unwatchable. The videos were littered with artifacts that hid the faces of the people speaking, though the audio remained.”

She writes that the Shoah IT team tried to use expensive film-restoration tools used in the motion-picture industry without success. Gizmodo quotes Ryan Fenton-Strauss, Shoah's Information Technology Services video archive and post-production manager, as saying, “We went to some of the big names with our sampling of problems and let them take a crack at it, but no one was able to correct them. We brought in some of the most powerful and expensive film restoration tools to set up demos, and they just weren't able to do anything with the videos.”

Consequently, reports Arnold, the team decided to eliminate frames with unacceptable artifacts and leave the rest. The problem was how to detect artifact-laden frames automatically. She writes that a half-hour interview includes 50,000 individual frames, so manual identification of bad frames would not be practical.

Fortunately, the team discovered it could use NI Vision Builder to develop 10 to 20 templates that could search for the patterns found in bad images. The team hopes to complete the project this year.

Read more in Gizmodo.

More in Applications