关键词: 印刷星标, 二维Gabor滤波, 水墨平衡, 支持向量机, 主成分分析

Abstract: Printing label is the main way to monitor the state of ink balance. As the control for printing ink balance mainly depends on human experience, there are some problems such as strong subjectivity and low efficiency. In this study, the digital image processing and pattern recognition algorithm are used to solve the problem mentioned above and realize the automatic monitoring and identification of the balance of ink and water. The feature of printing label is abstracted with a 2D-Gabor filter and analyzed with a principal component analysis. SVM nets are trained to identify the statement of water and ink with the result of PCA and it has a qualified accuracy in application. This study has a significant in reducing the debugging time and cost of equipment. Both the quality and automation degree of printing will be improved.

Key words: printing label, 2D-Gabor filtering, ink-water balance, Support Vector Machine（SVM）, Principal Component Analysis（PCA）