Dead Pixel Detection on Liquid Crystal Displays using Random Forest, SVM, and Harris Detector

Machine learning for soft and liquid molecular materials

“…The support vector machine algorithm proved to be the most effective in detecting dead pixels with an accuracy of about 92% compared to random forest. 86 In contrast, mura (a variation in local brightness with no distinct contour on a uniform LCD surface) was predicted using the random forest algorithm with a detection rate above 99% and a processing time of 27 ms per image, which is a competitive result for industrial systems. 87 4 Machine learning for bubbles…”

“…Specific tasks, such as dead pixels and mura detection during manufacturing process were also adequately addressed by ML. 86,87 Typically, such defects are detected by an operator, which is not always performed reliable and increases the cost of production. The support vector machine algorithm proved to be the most effective in detecting dead pixels with an accuracy of about 92% compared to random forest.…”