文档介绍：论文题目:基于机器视觉的印刷品缺陷识别系统研究
专业:模式识别与智能系统
硕士生:刘冰清(签名)
指导教师:黄梦涛(签名)
摘要
机器视觉主要是用计算机来模拟人的视觉功能,从客观事物的图像中提取信息进行
处理并加以理解,最终用于实际检测、测量、控制和识别。随着计算机及图像处理技术
的发展和机器视觉技术在工业在线检测中的广泛应用,利用机器视觉技术来进行印刷品
缺陷的自动识别已切实可行。
本文把机器视觉、图像处理、识别等技术与印刷工业相结合,系统地研究了基于机
器视觉技术的印刷品缺陷识别系统。设计并建立了一套基于机器视觉的印刷品缺陷识别
实验系统。该系统由 TMS320DM642、D 相机、显示器、照明光源、印刷品模
拟实验台以及采集同步控制装置(旋转编码器)等组成,为进行印刷品缺陷在线识别研
究提供了平台。针对采集到的印刷品图像,提出了图像灰度化,滤波,增强,边缘提取
及边缘差分的预处理流程。研究了缺陷识别算法,选择了 BP 神经网络算法实现缺陷的
分类。为软件开发平台,以 C、C++语言为工具,开发出了一套印刷品
缺陷识别系统软件,能实现图像定位采集,图像预处理,图像识别,图像显示等功能。
本文设计的印刷品缺陷识别系统基本满足了实时性要求,系统功能较全,使用方便。
但还有许多地方有待改善,才能满足实际应用的需要。在本文的最后,给出了一些系统
改进的想法。

关键词:TMS320DM642;机器视觉;缺陷识别;BP 神经网络
研究类型:应用研究
Subject :Research of Print Defects Recognition System based on Machine
Vision
Specialty :Pattern Recognition and Intelligent System
Name :Liu Bingqing (Signature)
Instructor:Huang Mengtao (Signature)
ABSTRACT
Machine vision means simulating the human visual function puters. It can
extract information from image of objective things, and the information can be processed and
understood. Ultimately it will be used for practical test, measurement, control and
identification. With the development puter, image processing and machine vision
technology is probably to identify the print defects with machine vision.
Combining with machine vision, image processing, recognition and other technologies
in the printing industry, it was studied to print defects recognition system based on machine
vision in this paper. Firstly, an experiment system of print defects recognition based on
machine vision was designed and set up. The system consists of TMS320DM642, area array
CCD cameras, monitors, lighting, simulated print station and the acquisition synchronized
control device (rotary encoder) etc. The system provided a platform for the research of print
defects on-line identification. Secondly, the collected images were analyzed and the image
pre-processing process, including graying, filteri