Online auto-detection method and system of presswork quality

An online auto-detection method and system of presswork quality is studied in this paper. An evaluation criterion for real-time-gathered images of multicolor printing is proposed. Integrated designing and optimizing make it transparent the key factors influencing the real-time performance of a detection algorithm. A subspace-based registration algorithm that accelerated the search speed is presented, and the dynamic refreshing of a reference image avoided the cumulative departure and improved the registration precision. To obtain the position and geometric characteristics of various defects in real time, a method of thresholding by cluster and line-by-line scanning of a dynamic array is proposed to extract and merge defects. Finally, geometric characteristics and statistics parameters are adopted to classify and identify the massive defects, discrete dots defects, and linear defects. A motion image acquisition system and distributed computer processing system are designed. Experimental study and practical running indicates that defects including stria, splash ink, blowby, foreign bodies, etc. could be detected, displayed, stored, and warned in real time. The detection speed reached 250 m/min and the precision reached 0.375 mm in the running direction and 0.15 mm in the crossing direction.