利用机器视觉的烟包封签缺陷检测算法  

Algorithm of Fault Supervision of Cigarette Tag Paper Based on Machine Vision

作　　者：何文平[1] 邱超[1] 

机构地区：[1]河南工业大学机电工程学院,河南郑州450007

出　　处：《工程图学学报》2009年第3期131-135,共5页Journal of Engineering Graphics

基　　金：河南省重点科技公关资助项目(072102210013)

摘　　要：构建了一个图像识别及检测的DSP+FPGA硬件平台,采用机器视觉技术,对烟包外包装各种信息进行准确而快速的识别。设计了一个烟包封签缺陷检测算法,选取图像强度分量进行灰度处理,选择合适的二值化和滤噪方案对图像进行二值化及滤噪处理,最终在二值化图像中进行封签的偏移、叠角和缺失的检测。为烟包的高速自动化包装检测提供了一种高效可行的实施方案。A hardware platform with DSP＋FPGA is constructed to identify and supervise images. Machine vision is used to identify the information of cigarette packages. An algorithm of faults supervision of cigarette tag paper is designed. The intensity of images should be firstly processed by shade, then denoise and binarization processes are carried out. The faults of package tag paper such as offset, overlap angle, etc. are identified. This method is efficient in automatic package supervision with high speed.

关 键 词：计算机应用 图像处理 封签检测 机器视觉 烟包 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]