Real-Time Tobacco Width Detection Based on Main Axis Extraction Algorithm
Zanwu Xie, Yunhao Li · 2025
Tobacco width measurement plays a pivotal role in cigarette manufacturing. As the cut width changes, the proportion of medium-long, short, and broken tobacco cuts also correspondingly adjusts. Overwide tobacco cuts can lead to loose rolling and reduce the flammability and burning speed of the tobacco. Therefore, quickly and accurately measuring the width of each batch of tobacco and adjusting the cutting machine accordingly is an important method to enhance tobacco quality in the modern tobacco processing industry. Currently, the mainstream method of measuring the width of tobacco still relies on manual sampling combined with offline detection on laboratory equipment. This method involves significant human intervention and cannot adjust the cutting machine on the production line in real-time. Tobacco width measuring under situations of uneven form, chaotic distribution, and mutual blockage is thoroughly examined in this research in order to increase its effectiveness. Based on the main axis extraction approach, we provide a way to measure the width of tobacco. This method can automatically extract the points suitable for the width measurement of tobacco based on the visual characteristics of the tobacco, without manual intervention, and measure the width of the tobacco at these points. Therefore, it reduces the influence of tobacco posture on measurement. With this technology, we can achieve online real-time monitoring of the width of tobacco on the assembly line, and our experiments with samples of varying widths reveal that it ensures measurement accuracy, increases detection speed, and decreases the degree of human intervention.