【作者】 段峰；

【导师】 王耀南；

【作者基本信息】 湖南大学 ， 控制理论与控制工程， 2002， 硕士

【摘要】 基于机器视觉的智能空瓶检测机器人以其高速度、高精度、智能自动化的特点成为了当今高速啤酒生产线上必不可少的高科技设备之一。而我国机器视觉的研究和应用还处于起步阶段，对智能空瓶检测机器人的研究几乎是空白，远远落后于国外发达国家的水平。我国啤酒生产的大规模发展，急需国人的努力来填补这一空白。为此，本文对智能空瓶检测机器人进行了深入、系统的研究。 论文综述了当今世界机器视觉技术的发展与应用现状，介绍了与空瓶检测相关的重要机器视觉技术，分析了空瓶检测的技术特点和难点，提出了直线式和旋转式两种机器人的传动结构，设计了智能空瓶检测机器人的机械和电气控制方案，深入研究了多种数字图像检测算法，研制了智能空瓶检测机器人实验系统，完成了所有机械、电气控制部分和软件的设计、安装和调试。 主要工作包括以下几方面： 1、对啤酒行业中空瓶检测的生产需求和工艺流程的要求进行了完整而详细的描述，并在此基础上，通过理论分析和实验总结了机器视觉应用与空瓶检测的特点和难点，得到整个智能空瓶检测机器人的技术需求，提出了智能空瓶检测机器人是集光、机、电一体化的高度综合复杂的工业机器视觉应用系统。 2、在总结了目前与空瓶检测机器人相关的机器视觉技术和国外相关成果的基础上，结合空瓶检测的实际需求，提出了智能空瓶检测机器人系统的整体方案。在机械结构方面，提出直线式和旋转式两种机械传动结构，并详细介绍了直线式空瓶检测机器人的机械布局以及光学成像系统和照明方案。对于系统的电气控制部分，本文提出了基于PC机、基于DSP和基于视觉传感器的三种结构，分析了三种结构的特点，并设计了系统的方案。 3、对数字图像处理技术在空瓶检测中的应用进行了详细地研究，并做了大量的实验。针对空瓶检测的高速、高精度、实时性特点，在总结分析传统数字图像边缘检测技术基础上，提出了用于瓶底、瓶口、瓶身检测的实用算法，并通过试验验证了算法的有效性。 4、对智能信息处理技术在空瓶检测中的应用进行了分析研究和试验。将模糊信息处理技术和神经网络技术用于图像检测，将多传感器信息融合技术用于提高CCD摄像机的动态范围。通过实验发现用模糊信息处理技术进行边缘检测具有可行性，但效果一般，没有明显优越性。而BP神经网络在图像检测中效果较好，有一定的优点，但计算耗时多，在基于PC机的应用系统中，实现高速度的实时检测比较困难，必须借助专用运算器。而采用多传感器图像融合算法，能大幅度地提高CCD摄像机的动态范围。 5、研制开发了智能空瓶检测机器人实验系统，完成了整个系统的机械、电气控制部分的设计、安装和调试，用VC6.0和VB6.0开发了基于PC机的一整套测试和研究的软件平台，建立了空瓶图像数据库，完成了本文所有算法的测试、研究工作，为智能空瓶检测机器人的进一步研究奠定了很好的基础。本文还针对智能空瓶检测机器人实验系统的不足， 拍要 提出了新一代开放型机器视觉实验系统的设计方案。 通过理论分析研究和大量实验证明了本文所提出的针对空瓶瓶口、瓶底、瓶身的数字 图像检测算法等有效性和可行性，解决了空瓶检测机器人应用中的大部分问题，为进一步 的研究开发奠定了4for的基础。更多还原

【Abstract】 Characterized with auto inspecting ability and high inspecting speed and accuracy, intelligent empty bottle inspector (IEBI) is one of the most necessary high technology equipments in high-speed beer production. The research and application of machine vision have been introduced to our country just several years before. Compared with developed country, we achieve fairly limited accomplishment in the research on EBI. At present, ft is necessary for us to develop EBI due to the large and increasing beer production scale in our country. Under such situation, this paper put forward a deeply and systematically research on IEBI.This paper gives an overview of current machine vision and introduces the key technologies applied in EBI. Based on the analysis of characteristics and difficulties in the realization of EBI, this paper presents line and rotary transfer structures, the mechanical and electrical system scheme of EBI. This paper focuses on the application of digital image processing technology to the defect detection in empty bottle inspection. An experimental equipment including all mechanical, electrical instruments and software is developed. The main contributions of this paper are the followings:l.An explicit description of empty bottle inspector in the beer production is givea Based on theoretically analysis and experiments, this paper concludes the characteristics and difficulties in the application of machine vision to empty bottle inspection and technology requirements for EBI. It is emphasized that EBI is an integration of optical, mechanical, electrical and digital image processing technology.2. According to the technology requirement of EBI and referring to the current advanced achievements, a scheme of EBI is proposed. The line and rotary transfer mechanical system are suggested. This paper introduces line empty bottle inspector in its mechanical structure, optical and lighting system. Three type of electric control system are discussed in this paper which are PC-based, DSP-based and vision sensor-based respectively.3. This paper concerns on the application of digital image processing technology to the detection of defect in empty bottle inspection. After concluding the traditional edge detection method, severalnovel and useful algorithms for the inspection of empty bottle bottom, finish and wall are put forward. Experiments have proved their efficiency and feasibility.4. This paper makes an attempt to using intelligent information processing technologies for empty bottle inspection. Fuzzy information processing, BP neural network and multi-sense information fusion are studied. The results of experiments show that Fuzzy technology is not better than traditional processing method in edge detection while BP neural network better but is time consumption. Multi-sense information technology succeeds to improve the dynamic scale of CCD camera, which is useful in empty inspection.5.An experimental EBI equipment is developed, which is a useful tool including the whole mechanical, electrical system and software developed with VC6.0 and VB6.0. By the aid of this experimental equipment, an empty bottle image database is built and all algorithms for inspection are tested. Another advanced experimental equipment is devised, which is more capable in various machine-vision experiments.Theoretical study and practical experiments have proved the efficiency and feasibility of thealgorithm for inspection provided in this paper. Further study is bond to benefit from theachievements of this paper.更多还原