基于机器视觉的印刷品全画面缺陷检测系统的研究-机械电子工程专业论文.docxVIP

摘要 摘要 I I 摘要 随着印刷工艺和生产的进步，用户对产品质量与多样化的要求日趋增高，基于机器视 觉的现代检测技术已经成为印刷行业的主流趋势。本文对应用于宽幅面、高精度的基于机 器视觉的印品表面质量检测理论与算法进行了研究， 提出了一种基于机器视觉的印刷品 全画面缺陷检测系统。 （1）对检测系统的结构框架进行了设计和分析，提出由图像采集、图像处理、数据 管理和信号控制等功能模块组成的系统框架，并基于这些模块间的协同完成检测任务。 （2）在表面图像的空间位置匹配定位方面，提出了一种指定特征区域的图像匹配方 法，该方法应用了边缘匹配技术，通过指定提取范围与预设的匹配范围，降低了计算量、 缩短了匹配时间，保证了图像匹配的准确性和实时性。 （3）在缺陷特征分析方面，针对表面缺陷和色彩缺陷分别提出了不同的缺陷特征分 析方法。针对色彩缺陷，论文利用 Lab 颜色空间以及 Lab 色差公式，计算图像中像素点是 否存在色差；针对表面缺陷，采用了基于 Blob 分析的特征分析方法，并提出了一种改进的 逐像素扫描标记方法，只需经过一次扫描，便可完成对所有连通域的标记工作。 （4）在表面缺陷识别方面，设计了点、线、面不同缺陷类型的识别规则，并通过实 验验证了系统已可实现对常见缺陷的检测。 关键词：机器视觉，印刷品表面质量检测，图像处理，Blob 分析 Abs Abstract II II Abstract With the development of printing technology and production，the higher request about printing quality and diversity has been put forward by the customers. And the modern detection technology of machine-vision-based has become the main trend in the printing industry. In this paper, lots of printing surface quality testing theory and detection algorithm has been studied and researched, which is used for corresponding wide printing format and high precision detection. And then, a machine-vision-based inspection system for full screen printed matter has been presented. In this paper, the detecting system framework is researched and composed by image acquisition, image processing, data management and signal control functional modules. Meanwhile, through cooperation between the modules, the defect detection task has been completed. In the image matching process about the position space between the standard image and tested image, this paper presents an improved image matching method, in which the extractive objects and matching range will be specified. By use of this algorithm, the amount of algorithm calculation is reduced, the matching time is shortened, and the performance of accuracy and real-time has been guaranteed. In respect of defect feature analysis, the detection system respectively presents defect feature analysis methods for surface defects and color deficiency. By us