The Research Of Automatic Inspection Algorithm Of Multi-shape Defects Of Printing Plate Based On Machine Vision

Machine vision is an important component of industrial automation and is widely applied in each segment of the industrial automation production with the proposal of "Industrial 4.0".As so far,manual quality inspection,which is an inefficient and inaccurate method,is still the mainstream choice of the most manufacturers of printing plate in printing industry.On the other hand,this method rises the cost of production with the increase of personnel costs.An automatic quality inspection by means of machine vision will not only reduce the personal costs,but also improve the efficiency and accuracy of inspection,that will promote the generalization of industrial automation.In this paper,we choose reasonable hardware and elaborately design a PC-platform of machine vision inspection in order to accommodate the production environment of printing plate.One of the advantages of this platform is agile and changeable,which is in favor of algorithm updating.Generally,automatic inspection is achieved by the image processing algorithms,which are realized by a personal computer.The major part of this paper is the proposal of the Homocentric Squares filter(HSF),which is designed based on the difference of gray scale for detecting defects with various shape.For the sake of the inspection of defects with different sizes,multiscale analysis is a necessary part.Then,extracting the feature of the gray level co-occurrence matrix from the result of HSFand adopt support vector machine to reduce the false positive of the inspection results.Finally,makes an interface for users to interact with computer by QT.The system is achieved by the development tool VS2015 and open source database OpenCV.In order to detect defects real-time,the integral is adopted in HSF to calculate the pixel sum of any rectangle regions.As the experimental results showed,the combination of HSF and SVM performs better than other algorithms in defects inspection.Comparing the HSF to other edge-detectors,the former misses less defects.The usage of SVM reduce the error rate of inspection.In a word,this scheme not only improves the accuracy,but also reduces the missed defects,and the interface bring convenience to the user,all of that brings us an innovative and functional system.