基于数字图像分析技术的印刷墨斑与纸张匀度检测方法的.pdf

申请工学硕士学位论文
论文题目
基于数字图像分析技术的印刷墨斑与纸张匀
度检测方法的研究
学科门类:工学
级学科:轻工技术与工程
培养单位:轻工与能源学院
硕士生:王倩
导师:徐永建
2014年5月
Research on Detecting Method for Print Mottle
and Paper Formation using Digital Image Analysis
Technolog
Y2554502
A Thesis(or Dissertation )Submitted to
Shaanxi University of Science and Technology
in Partial Fulfillment of the Requirement for the Degree of
Master of Engineering Science
By Qian Wang
Thesis(or Dissertation)Supervisor
Yong-iian Xu
May,2014
基于数字图像分析技术的印刷墨斑与纸张匀度检
测方法的研究
摘要
采用数字图像分析技术分析印刷墨斑和纸张匀度,构建了印刷墨斑和纸
张匀度的评价模型。纸张匀度和印刷墨斑是影响印刷质量的两个重要因素。
两者均是从均匀性的角度出发,印刷墨斑反映了油墨的不均匀性,纸张匀度
则反映了纸张的匀度信息。实际上,在造纸、印刷工业中,大多数论文都写
了关于印刷塁斑和纸张匀度机理、起因以及解决办法,本课题主要是提供评
价印刷墨斑和纸张匀度的数学模型,更加精确的定义了印刷墨斑和纸张匀
度,为改善印刷墨斑和纸张匀度提供了指导意见。
课题采用传统方法、小波去噪理论、小波细节能量理论以及小波去噪和
梯度法构建评价印刷黑斑的数学模型,并与人眼视觉评价体系进行相关性分
析,最终,通过实验的方法,验证这些结果的正确性和可行性。结果发现:
本课题所构建印刷墨斑的数学模型均优于传统方法构建印刷墨斑的数学模
型。其中,基于小波去噪理论构建印刷墨斑的数学模型在处理四种纸张时,
新闻纸的处理结果最优,其次是喷墨纸,铜版纸处理的结果最差。新闻纸和
喷墨纸的最优参数是1级~3级,双胶纸为3级、4级,铜版纸则只能选择1
级:;基于小波细节能量理论构建印刷墨斑的数学模型,喷瞾纸处理的结果最
优,依旧是铜版纸处理的结果最差,新闻纸的最优参数为3级~7级,喷墨
纸、双胶纸和铜版纸的最优参数为4级7级,与人眼视觉评价体系的相关
系数大于0.78;对双胶纸应用小波去噪和梯度法构建印刷墨斑的数学模型,
其最优的分解级数为1级~4级,与人眼视覚评价的结果大于0.8,处理结果
明显优于传统方法、小波去噪理论和小波细节能量理论构建印刷墨斑的数学
模型。
另外,对于纸张匀度的处理,采用相同的思路,利用标准匀度指数、梯
度法、 Jordan比周长及傅里叶变换和梯度法结合构建纸张匀度的数学模型,
利用 MATLAB对这些模型进行实现,最后采用实验的方法,验证这些结果
的正确性和可行性。结果发现:基于傅里叶变换和梯度法构建纸张匀度的数
学模型优于传统方法构建纸张匀度的数学模型。
关键词:印刷墨斑,纸张匀度,数字图像,小波变换,傅里叶变换
Research on Detecting Method for Print Mottle
and Paper Formation using Digital Image
Analysis Technology
ABSTTRACTT
The model is based on digital image analysis techniques to analyze
the print mottle and paper formation. Print mottle and paper formation
are two crucial factors in the assessment of the printing quality of
papers and boards. Print mottle can reflect ink unevenness, while the
paper formation can reflect the information of evenness of paper. In fact
much has been written about the mechanisms, causes and cures for
print mottle and paper formation. This work extends that understanding;
offering a mathematical evaluation model of print mottle and paper
formation
The model is based on the theory of wavelet image denoising
analysis and details of energy by using the wavelet multi-scale fast
algorithm and wavelet image denoising and gradient. It was then
applied to four business papers (inkjet papers, newsprint papers, art
papers and double coated offset printing papers). This portion of the
work checked the validity of the proposed method by analyzing the
correlation between the method and a human visual evaluation system
(HVS). The results showed that the experimental results showed that
good agreement with HVS, in which the correlations with visual
evaluation on printed newsprint papers were over 0.76 using eigh
different wavelet base functions (such as haar, sym4, bior3.7, etc)
decomposing at the first, second and third level And the results on the
other three papers were also better matched with analysis results of
human eyes using this model. But the correlation of the art paper and
visual model was not as strong as others. The optimal parameters of
print mottle model for the four different structured p