Online Recognition of Moving Cigarette Cartons on Semi-automatic Sorting Line

improve the sorting efficiency of a vertical cigarette carton sorter, vision technology was adopted to realize the online recognition of moving cigarette cartons on a semi-automatic sorting line in a tobacco distribution center. A vision system platform was configured for image acquisition and preprocessing. A "two-step" algorithm was proposed to extract the image features, including color, texture and shape. An image feature database was established with the obtained image features. Images were recognized via the criterion of minimum feature distance to check the brand and amount of cigarette cartons in each individual order automatically. The results showed that in the case of a distribution center where the total daily capacity was about 50 000 cartons handled by four lines, 0.5 hours could be saved per day. The system is stable, running at a speed of 6 frames per second, recognition correctness is over 99.99%.