*** apologies for crossposting ***
Has someone experience with automated control of the visual content of still images?
We have some old data from digitization (cca 1,5 million of pages) stored in JPEG and DjVu we need to check.
We are able to detect corrupted headers, but we would like to check also visual damage of the images (color shifts, missing/dismissed pixels, one color screens, etc.). The images typically would have correct headers but the visual information might damaged.
Do you have suggestions on a methodology, "workflow" solution (software; tool; script etc.) to check visual damage of images automatically? Do you know or you are using any software to correct image corruptions? Somebody has experience with similar problem he would like to share with us (and others also)?
Digital preservation department, National Library of the Czech Republic, Prague