Abstract: Based on the demand for automatic monitoring of Lasioderma serricorne （Fabricius） in tobacco storage and cigarette production workshops, combined with the actual image characteristics of Lasioderma serricorne （Fabricius）, a Lasioderma serricorne （Fabricius） automatic monitoring system based on image processing was designed by overlaying segmentation image algorithms on the basis of labeled watershed algorithms. Fully automatic photography equipment was used to collect images of the corresponding traps at regular intervals, the images were transmitted to the server via wired network, and functions such as image recognition and counting, real-time display, over limit alarm, and historical curve viewing were completed on the server end. Many interference factors such as insect board smoke, insect body overlap, and light during the actual monitoring process of Lasioderma serricorne （Fabricius） were solved, and precise automatic image recognition and counting functions for Lasioderma serricorne （Fabricius） were implemented. Practical applications had shown that the system worked stably, with an average counting accuracy greater than 94.00%, and had good application prospects in tobacco pest monitoring in cigarette production workshops.

Key words: image processing, Lasioderma serricorne （Fabricius）, automatic monitoring system, cigarette, tobacco, design and application