Research On Automatic Register Detection Algorithm In Color Printing

With the rapid development of printing technology, post-press finishing equipment get deeper toward to automatic, multi-functional, and auto-detecting instruments. The register detection is an important working procedure in color printing. The detection accuracy and precision have a direct influence on the printing quality. As a significant parameter in color printing, many international printer manufacturers use automatic equipment to check it. But in our country, the printer manufacturers still check it manually with low accuracy. For improving our country's printing industry automatic level, and for meeting the real need of the industry, we devote ourselves to the research on the Register Detection System.This thesis applies machine vision technology to the register detection and develops a register detecting instrument. At first, research background and the project target is presented. This article also introduces the content and application of computer vision. The characteristic of register is involved too. According to this characteristic, the area filter is adopted to filter the noise of original image. Then it lay stress on the color segment and positioning the cross of register. On the basis of the characteristic of register, the register is segmented in the HSI color space. The different colors are clustered by the normal eigenvector (i, r, g). Then the edge is detected by gradient. The experiment has proved the method is stable and reliable. In this thesis, the Hough and the least square method are combined to position the cross. The method has very strong interference rapid and accuracy. Later, the system hardware is presented. In the end, the experiment data and operation in factory is referred to show the correction of the register detection system, and its accuracy is better than the national standard.