"Evaluation of subtle irregularities in roll to roll coatings has historically been a job for the human eye. Unfortunately, with modern manufacturing being a 24/7 operation, the repeatability and reproducibility of grading is heavily impacted by different operators on different shifts and fatigue caused by long hours. To help eliminate this variability, we looked to technology. The primary defect used for this study was flow mottle on x-ray film. The coating is less than one mil thick and must be very uniform to ensure consistent imaging. Using an off the shelf film scanner, a high resolution digital copy of the sheet was created. This was then processed using image analysis software and run through an algorithm in Excel to return a flow mottle grade. The entire process takes ~2 minutes.
The model takes the image and averages pixel intensity in the downweb direction over a two inch region. Too long of a distance results in a smoothing of the intensities while too short of a distance results in excessive noise. This process provides a series of crossweb optical density values for the sheet in question. From this series a long distance moving average is subtracted to remove baseline drift in intensity over the sheet. Finally, a height threshold is created, so that every time the intensity of the film changes by a fixed amplitude over a fixed width, a tally is taken. The number of these intensity variations per sheet is then positively correlated to the human graded flow mottle. This is just the start of Carestream’s exploration into augmenting the manual inspection system with a digital system which is more consistent and not subject to the variability of fatigue or technician. There is potential for this technique to also apply to quantifying white spots, black spots, streaks and many more.