Research On The Technique Of Process Printing Detection Based On Machine Vision

In the process of colored newspaper printing, register accuracy is one of the most important standards to evaluate newspaper quality. In order to mensurate register accuracy, a way named localization marker is usually used to detect the change of registers at any moment in order to take relevant action to ensure the register accuracy. At present, traditional detection way which depended on manual work is still used in domestic many printing enterprises.It is known that manual detection results in low efficiency and accuracy. Furthermore, it usually results in lots of waste paper. Therefore, it will make important sense for advancing and enhancing the tecninque in print to research and use register detection technique which is faster, more accurate and more automatic.As a new technique, machine vision is already used in many filed.And it's proved that the technique can not only enhance efficiency but also accept high accuracy. On the basic of that background, research on the register detection of colored newspaper printing combined with machine vision is made in this paper.Firstly, by analyzing the principle of register detection and machine vision, a holistic frame based on machine vision is designed.Then the CLS (Cross Localization System) is programmed by VC++ and Matlab. At last, an experiment is made with homemade experimental model, and the result is given. Key technologies in this paper include real-time capture processing, technology of subpixel localization and camera calibration. They are also the most challenging parts of this paper. In light of this, real-time capture is realized through VFW technology, and a method of cross subpixel central localization based on centroid is developed. A camera calibration Technique based on homemade planar template is developed based on Tsai's two-stage technique. The methods are validated by experiments and can achieve an excellent precision under normal experimental situation.Limited to research condition and time, the research on register detection based on machine vision is still on the stage of laboratory. However, the principle and the method of realizing as well as CLS have stimulative effect on the technique of register detection developing towards high accuracy, high efficiency and high automatization.