The Inspection System Of Cigarette Label Based On The Machine Vision

Machine vision has been the outstanding field in industrial automation after decades' development. Image measurement and image recognition are playing more and more important part in on-line inspection system in industry.The design of the on-line inspection system of cigarette label is illustrated in this paper, which includes three parts: the hardware and software design of dynamic image acquisition system;the algorithms software of recognition of cigarette label;how to realize cigarette label and how to output the result of the inspection.Theory of the color image processing is systemically introduced aiming at the color images of cigarette label in this paper. At the same time, the design of the algorithms of the image processing is emphasized, which is the key to the inspection system of cigarette labels. Many experiments have been made to validate the different result of the image of the cigarette case when it was lighted by the different lamp-houses and the different illumination modes. According to the diversity of cigarette labels, the characters of all kinds of cigarette labels have been summed up in order to set up the right algorithms of cigarette label inspection. Moreover, the author has attempted four methods to inspector the cigarette label: (1) the inspection method based on the weight of RGB for recognition of the labels which have the different color from background;(2) the inspection method based on the S weight of HIS for recognition of the color labels on argentate background;(3) the inspection method based on the H weight of HIS for recognition of the fuscous labels;(4) the inspection method based on pattern for recognition of the labels which have similar color with the background.The result of many experiments indicates that the algorithms of cigarette label inspection are efficient and the stability and accuracy of the inspection system are wonderful.