目的 针对圆角包装袋传统裁切工艺中存在的毛刺和废料问题，设计一种基于机器视觉的高精度制袋机裁切定位系统。方法 利用LED背光源、低畸变镜头、CMOS黑白相机和工控机搭建视觉系统硬件平台，在Visual Studio开发环境中，基于OpenCV函数库设计视觉软件，用基于最小二乘法检测图像边缘，进而标定相机，提取包装袋圆角轮廓为特征信息。系统以特征信息为匹配模板对圆角位置进行精确定位，得到亚像素级角点位置，最终控制伺服系统执行误差补偿裁切。结果 该定位算法可实现16 ms内精确定位圆角坐标，绝对误差小于±0.02像素点。结论 基于机器视觉的定位系统具有运算速度快、定位精度高等特点，有效改进了传统制袋工艺，能够满足生产需求。

The work aims to design a location system for cutting of high-precision bag making machine based on machine vision, aiming at the problems of burrs and wastes in the traditional cutting process of fillet packaging bags. LED backlight, low distortion lens, CMOS black-and-white camera and industrial personal computer were used to build the hardware platform of the vision system. In the Visual Studio development environment, the vision software was programmed based on OpenCV function library. The image edge was detected based on the least square method for the camera calibration. The package fillet contour was extracted as the feature information. The system took the feature information as the matching template to accurately locate the fillet position, obtain the sub-pixel level point, and finally control the servo system to perform error compensation and bag cutting. The location algorithm could achieve precise location of fillet in 16 ms, and the absolute error of location was less than ±0.02 pixels. Featured by fast operation speed and high precision location, etc., the location system based on machine vision effectively improves the traditional bag making process, and can meet the production requirements.