【作者】 陈晨；

【导师】 何炳蔚；

【作者基本信息】 福州大学 ， 测试计量技术与仪器， 2017， 硕士

【摘要】 快递物流业的急速发展以及机器视觉技术的进步开启了智能物流的新时代,特别是当前国内快递物流行业自动化装备差、信息化水平低,迫切需要能够缓解快递分拣配送压力、改善快件收发时效的智能分拣识别装备。快递物流的分拣一般通过快件编号信息进行分类,而当前国内绝大所数分拣中心仍依靠手持式扫描仪对条码进行识别。该方式劳动强度大、工作效率低。因此开发能够自动识别快递条码的智能识别装备已迫在眉睫。近年来机器视觉自动检测技术蓬勃发展,基于图像处理技术的条码检测识别技术也逐渐得到关注。本文根据实际需求,设计和实现了基于机器视觉的快递物流条码智能识读系统。其主要研究内容如下:1、结合快递物流业的作业环境,对比手持式扫描仪,阐明了利用机器视觉技术进行条码识别具有低成本、高效率等优势。依据相应的设计要求,确定了静态称重读码系统的设计方案。然后对系统的结构组成进行了简要分析,并对各硬件的选型给出了详细的选型依据,最终确定了硬件系统的结构布局。2、对于快递物流条码的检测识读,在分析条码的结构特征之后,以条码条纹相间的纹理特征为基础,避开二值图像操作易产生干扰项的缺陷。首先,采用图像分解算法分解出条码的纹理图谱;然后,利用图像灰度形态学运算对条码区域进行图像增强,再利用改进的自适应阈值分割算法进行候选区域的分割;最终通过区域几何特征筛选出粗定位的条码区域。3、将条码相互平行的边缘属性作为细分项,对候选区域进行精选。有别于传统的直线段提取算法,文中采用Line Segment Detector(LSD)直线段检测算子完成复杂背景下条码“条”与“空”边缘的提取,并设定相应的参数筛选出条码区域。4、对于条码的解码问题,文中首先对快递物流条码的结构组成及编码方式进行了详细分析。为了解决大件包裹在固定焦距工作模式下可能出现的图像模糊以及条码本身缺损、畸变等因素造成的解码困难的问题,提出利用图像梯度与块平均方法求取解码扫描线方程,并联立边缘方程求取边缘位置点进行自适应迭代验证的解码方法。最后,设计了校验与纠错算法保证了解码的精度。5、设计与实现了基于MFC及OPENCV的条码软件识别系统。更多还原

【Abstract】 The rapid development of e-commerce and express logistics and the progress of machine vision opened up a new era of intelligent logistics.It is urgent to equip an intellectual device which can alleviate the pressure of express sorting and distribution and improve the timeliness of express delivery sending and receiving under the logistics of our country.The express sorting usually circumstance of poor automatic facilities and low information technologies in based on the information of courier number.Presently,most of the sorting centers in our country depend on the hand-held scanning device to recognize the linear code which is a way of time-consuming and labour-intensive.So,it is imminently to develop an instrument that can auto discern the logistic bar code smartly.The machine vision detection technology has flourished in recent years.The linear code detection techniques based on image processing is also gradually achieved attention.According to the actual demanding,this thesis designed and implemented the express intelligent bar code recognizing system based on the machine vision.The main research contents are as follow:Firstly,compared to the hand held scanner,the advantages of low cost and high efficiency using machine vision technology to identify the linear code in present working environment are illustrated.According to the corresponding design requirements,the framework of the semi-automatic static weighing and bar code reading system is determined.Then,analysing the structure of the system in brief and describing the evidence of selecting appropriate hardware are carried out.Finally,the layout of the whole system are determined.Secondly,in the aspect of linear code detection and recognition,after exploring the structure characteristics of the bar code,the method of fast cartoon texture decomposition is utilized to enhance the area of linear code and to avoid the disadvantage of using image binarization operation which is prone to generate disturbances on the basis of texture feature that the stripes alternate arranged in the linear yards.Then,the way of grey morphology open operation and the improved adaptive threshold segmentation algorithm are taken advantage to extract the candidate region.Ultimately,the bar code region is coarse localized by using the geometric feature.Thirdly,the attribute that the edges of the bar code are parallel to each other is adopted to subdivide the linear code.Being different from the traditional line segment extraction algorithm,the LSD operator is used to extract the parallel edges between the stripe and space of a linear code under the complex background.Finally,some critical parameters are set to pick up the accurate linear code region.Fourthly,as respect to the problem of decoding,the structure of bar code is first to be introduced and the coding scheme for current domestic express is followed to analyzed in detail.Then,to solve the decoding difficulties that caused by the factors of image blur,linear-code defects and bending distortion,the adaptive iterative decoding method which needs to solve simultaneous equations to acquire the edge position coordinates is proposed.And to ensure the accuracy of decoding,a procedure of calibration and correction is designed.Finally,a linear-code recognition system is designed by the tools of MFC and OPENCV to make the system implementation.更多还原