双色印铁机印刷质量控制系统关键技术研究

发布时间：2018-08-26 08:07

【摘要】：随着生活水平的提高,人们对印刷品质量提出了更高的要求,而双色印铁机中印刷质量控制系统是决定印刷品质量好坏的关键因素,因此具有很高的研究价值。针对目前国内双色印铁机印刷质量控制系统存在的不足,本课题采用图像处理和模式识别技术设计了双色印铁机印刷质量控制系统,主要研究内容包括以下几个方面:根据金属薄板印刷品质量的控制需求,搭建一套基于机器视觉的彩色印刷品网点图像采集装置。结合采集彩色印刷品图像的大小和精度要求完成CCD数码相机、图像采集卡等硬件的合理选型。根据不同颜色空间特性的不同,选择了合适的颜色空间f_1f_2f_3;设计并制作了包含9种颜色模式信息的印刷品网点图像的标准样本,在此基础上提出了采用各向异性权重的模糊C均值聚类法(AWFCM)采集各颜色模式类的数据样本;最后,采用了一种基于竞争学习的算法用于图像信息的压缩。针对彩色印刷品网点图像颜色模式的复杂性和多样性,首先利用极限学习机(ELM)构建的粗分器对网点图像颜色模式进行初步识别,然后利用神经网络组构建的精分器对网点图像颜色模式做进一步分类,其中各子网络对像素的颜色模式单独识别,再通过信息融合中的模糊积分法对各子网络的输出结果进行整合,从而进一步提高了分类精度。本文提出了一种经过改进的PID神经元网络算法对印铁机的墨键开度进行调节,并利用MATLAB进行了仿真分析,仿真结果证明该进后的PID神经元网络算法具有较好的稳定性以及收敛速度,能够满足印刷品质量系统控制的要求。最后采用C#语言对印刷质量控制系统进行了开发,初步实现了系统功能。
[Abstract]:With the improvement of living standard, people put forward higher requirements for printing quality. The printing quality control system in two-color iron printing machine is the key factor to determine the quality of printing materials, so it has high research value. Aiming at the shortage of printing quality control system of two-color iron printing machine in our country, this paper designs the printing quality control system of two-color iron printing machine using image processing and pattern recognition technology. The main research contents are as follows: according to the quality control requirements of sheet metal printing, a set of color printing dot image acquisition device based on machine vision is built. The hardware selection of CCD digital camera and image acquisition card is completed by combining the size and precision of collecting color print image. According to the characteristics of different color spaces, the appropriate color space f1fStats is selected, and a standard sample of print dot images containing information of 9 color patterns is designed and made. On this basis, a fuzzy C-means clustering method with anisotropic weights is proposed to collect data samples from each color pattern class by (AWFCM). Finally, a competitive learning algorithm is used to compress the image information. In view of the complexity and diversity of color pattern of color print dot image, the coarse separator constructed by extreme learning machine (ELM) is used to identify the color pattern of dot image. Then the fine splitter constructed by the neural network group is used to further classify the color patterns of dot images, in which each sub-network recognizes the color patterns of pixels separately. Then the output results of each subnetwork are integrated by the fuzzy integration method in information fusion, and the classification accuracy is further improved. In this paper, an improved PID neural network algorithm is proposed to adjust the ink bond opening of the printing machine, and the simulation analysis is carried out by using MATLAB. The simulation results show that the proposed PID neural network algorithm has good stability and convergence speed and can meet the requirements of print quality system control. Finally, the printing quality control system is developed with C # language, and the system function is realized.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TS85;TP391.41

【相似文献】