The purpose of a green leaf threshing (GLT) plant is to separate the tobacco leaf from the stem. In the GLT process, there are over a dozen parameters to adjust for optimal performance. The current laboratory measurement technique is destructive, laborious and not as well suited for optimizing processing parameters as an on-line technique. A machine vision technique was developed to non-destructively measure the amount of stem in the stream of processed tobacco leaf. The machine vision technique employs novel lighting to collect images and image processing to measure the amount of stem in the image. The image processing combines fuzzy logic and unique morphological image processing to extract the leaf from the background and then the stem present in the leaf. Performance of the optical technique compared favorably with the lab method. Correlation coefficient (R 2 ) was greater than 0.95 and relative error (%RSD) was less than 6%.