Quality inspection of color printing products constitutes a very important QC task in printing industry. Many defects on such products may be inspected by current commercially available vision systems. However, for the inspection of complex color prints, the capabilities and performance of available systems are limited. The present practice in the printing industry is to inspect complex color products manually. This process is labor intensive and the results are not reliable as they vary with the time, mood and personal skills of inspectors. In this paper, we propose a novel algorithm to automate the color prints inspection process, which incorporates color histogram-based techniques for color image analysis and a neural network for image classification. Preliminary results have shown that this algorithm is able to inspect defects of complex color prints under varying illumination conditions.
展开▼