A Real-Time Vision System for Defect Detection in Printed Matter and Its Key Technologies

Abstract

A vision inspection system with high speed and on line is proposed to detect defects in printed matter, such as smudges, doctor streaks, pin holes, character misprints, foreign matters, hazing, and wrinkles, etc. Any tiny defect would be developed by using the high and low illumination angles design and anti-blooming techniques. A new image reference method based on morphological pre-processing eliminates all false defects brought by slight distortion of printed matter and chromatography mistake. The fast objects searching algorithm based on run-length-encoding can locate the coordinates of defects and define the shape of the defects. The C/S parallel network structure was used, image data were processed distributed and quality data is managed centralized. Experimental results verify the speed, reliability and accuracy of proposed system.