中文摘要：

为了提高香烟在线生产的效率,提出了一种烟包封条缺陷检测方法,开发了基于图像识别技术的烟包封条缺陷检测系统.所开发系统首先对采集到的烟包图像进行中值滤波和边缘检测;然后对边缘图像进行Radon变换;进而通过检测烟包小标上特征直线的角度和位置来判断烟包小标的粘贴是否合格;最后用PLC控制将判别不合格的烟包剔除出香烟生产线.中烟公司某烟厂长周期的应用结果表明：该检测方法具有判别速度快、易于实现等优点,所开发的系统可行、有效.

英文摘要：

To improve the efficiency of online cigarette production, a detection method for a cigarette packet seal sticker was proposed. Based on the method, an application system was developed. First, the captured image of cigarette packet was executed with median filtering and edge detection in this system. Second, the straight line on the edge image was detected by radon transform. Third, according to the position and angle of the straight line, the qualification of the seal sticker was determined. Finally, the cigarette packets with defective seal sticker were detected and removed from the production line based on PLC control. Results show that this detection method runs fast and easy to implement, and the long term application results in a factory of China Tobacco verified the feasibility and effectiveness of the developed system.