【作者】 陈路；

【导师】 刘昕； 谢天保；

【作者基本信息】 西安理工大学 ， 印刷工程， 2002， 硕士

【摘要】 本文在传统印刷图像质量检测方法的基础上，把数字图像处理学方法原理与印刷原理相结合应用到画面质量检测的新领域中，建立了基于画面的印刷图像质量检测与控制系统，提出了该系统的理论设计与模型。 主要对所提出的画面印刷质量检测算法进行了实验验证研究。1)使用模板匹配方法实现待检测图像和标准图像的定位；2)用几何形状和纹理特征对印刷文字部分的各种缺陷进行检测；3)针对网点图像区域，提出了在低分辨率下的纹理特征变化和颜色相关的检测，以及高分辨率下各色网点面积率大小变化的检测方法；4)用神经网络模式识别方法对多色网点图像实行分割和面积率计算，应用基于灰度共生矩阵和基于分形维数进行纹理特征检测，根据待检图像和标准图像对应区域的相关系数进行颜色相关分析。 实验数据表明：1)模板匹配定位算法能实现待检图像和标准图像的精确定位；2)基于几何形状和纹理特征的文字质量检测算法能正确识别出文字的各种缺陷；3)在高分辨率下，使用神经网络识别方法能正确的识别各色网点；在低分辨率下，基于灰度共生矩阵和分形维数的图像纹理特征分析，能精确描述待检图像和标准图像的差异。4)新系统与传统的质量检测方法相比，更精确，效率更高。更多还原

【Abstract】 In this thesis, the checking methods for printing quality based on image are taken as research objects. On the base of the traditional checking methods of printing quality, combining the method and theory of digital image processing with printing theory in the new domain of image quality checking, We constitute the printing quality checking and controlling system based on image, and expound theory design and model of this system.The checking algorithms of printing quality put forward have been mainly researched by experiments. 1) achieving the location of the image waiting for checking and standard image by template matching method; 2) checking the limitations of printing letter by geometry shape character and texture character; 3) for dot image, two kinds of checking methods have been proposed , one is the checking method using texture character and color correlation in low resolution .another is the checking method using dot area percent of printing colors in high resolution. 4) segmenting dot image and accounting dot area percent using Neural Network pattern recognition method, checking texture character based on Co-occurrence feature and fractal dimensional feature, and checking color correlation based on correlation coefficient of the image waiting for checking and standard image in the corresponding field.Experiment results show that 1) template matching location algorithms can achieve exactitude location of the image waiting for checking and standard image ;2)the checking algorithms for printing letter based on geometry shape character and texture character can detect the limitations of printing letter;3) in the high resolution, Neural Network pattern recognition method can exact-ly recognise-the dots of printing colors; in the low resolution, checking method ,of .texture character based on Co-occurrence feature and fractal dimensional feature can exactly describe the difference between the image waiting for checking and standard image;4)the checking method of the new system is more exactly and efficiently than the traditional checking method.更多还原