On-line Inspection Research Of Rotary Screen Printing Based-on Machine Vision

The registration system for rotary screen printing machine is one of the most important aspects determining the quality of the printed fabrics, but it was mainly controlled manually in the traditional process of fabric printing. But the manual control of the on-line inspection is inefficient and depends on many human factors, so it can not give an objective judgment about whether there had been a "off pattern" phenomenon, and the operator will be easy to be tired. With the development of computer technology and digital image processing technologies, it is possible to develop an automation registration inspection system based on machine vision. The on-line inspection is widely applied in modern industrial automation production.In this thesis, the background of this application is Rotary Screen Printing Machine. Firstly, the design of the on-line inspection system based on machine vision is dissertated in detail, a system which includes PC and linear CCD was introduced on the basis of preceding research. Then, the image preprocessing technology and the image segmentation technology used in machine vision are discussed, median filter method is used to filter noise, and histogram equalization method is used to enhance image. After the basic image segmentation algorithms are introduced, a new thresholding segmentation algorithm which is suitable in this system is applied. Mathematical morphology is used to process the binary image. Besides, the two high-precision positioning algorithms which includes point pattern matching algorithm and block matching algorithm are also discussed, and then a fast matching algorithm is presented considering the requirements of real-time system. Finally, the software design of registration system developed in Labwindows/CVI platform is introduced, which includes a match based and NI Vision based approach.