Research On Printing Character Quality Detection Technology Based On Halcon

The modern print of product package is executed in an assembly line, which is completed by industrial printing machine. Since this is large scale, various defects, such as incomplete printing, garbled sequence of characters, leakage of characters and dislocation of characters, may emerge during the printing process. In order to reduce economic loss induced by these defects, it is necessary to examine the printing quality and reject defective products.Thus, a rapidly and accurate detect system for printing quality should be found.This study is devoted to researching segmentation and identification of characters at high speed, including digital, letter and Chinese characters. According to our research result, these technological problems, such as irregular front defect, prolonged image processing, which affect the actual productivity and accuracy rate, have been solved. What is more, this technology will meet the requirement of online and real-time detection from printing industry.Pattern recognition and image processing and understanding is a hot spot in current research on the inside of the machine vision, which is more popular in the popular image processing. This shows that the image processing to play a key role in the modern computer applications also showed in this area there are a lot of not overcome. In the application of image processing, will choose different way of handling according to different situations, such as the selection of image enhancement, image smoothing, sometimes, sometimes to the image segmentation, with the deepening of the scientists research new research will also occur.Machine vision is the core part of the image processing. Image processing can be a process,such as the use of certain methods of image preprocessing, such as binarization, such as gauss-filtering, the purpose of the pretreatment is to remove the noise in the process.During this study, we setup the digital camera on the production line to take pictures by triggering camera with PC. The pictures are acquired, read and called real pictures. Another obtained pictures, which are called model pictures, will be trained and compared with real pictures to obtain difference with real pictures. These differences will be analyzed to get theinformation of defect. This research applied computer-vision software Halcon(Germany) to found the whole visual system. We applied the mean filter and median filter to processing these pictures to reduce the influence of edge sharpening and image noise. During the detection, template matching method was used to confirm the defect of printing character. If this algorithm comes true, the difficult in identification of fonts defects that produced during further printings will be solved. This algorithm will make it possible to test thousands of products in an hour with the same accuracy and Breakthrough technology bottlenecks between speed and accuracy rate.