| Increasingly strict requirements have been raised on the quality of the presswork by the modern printing industry, and the traditional artificial defect detection method can no longer satisfy the demands of the modern industrial production due to its high cost, low efficiency and strong subjectivity. With the rapid development of the computer technology and the high-resolution digital imaging technology, the non-contact product quality detection based on the machine vision has been one of the important measures for detecting the modern industrial quality. Therefore, there will be a great practical value and realistic significance in studying the presswork defect online detection technology based on the machine vision, improving the related algorithms and enhancing the performance detection.In this paper, corresponding improvements in the algorithm have been proposed on the basis of furthering the study on the related principles and algorithm for the presswork defect online detection technology. The research work in this paper mainly includes the following points:(1). Aiming at the large batches of the small target presswork detection tasks on the production line, an improved rapid moving multi-target detection algorithm would be put forward, in which the moving target detection would carried out with the three image difference method at first and the background image would be produced, and then the moving target detection would be conducted for the subsequent frames with the background subtraction.(2). A pre-selected multi-target rapid matching and tracking method based on the geometrical characteristics would be put forward, in which the target region without complete presswork image would be eliminated according to the geometrical characteristics of the minimum bounding rectangle in the target region before matching the templates in all the moving target region.(3). Global registration and part splicing would be conducted for the presswork images waiting for detection based on the SURF-based image characteristics extraction method, and a characteristic matching strategy resisting the interference from repeated characteristics would be proposed.(4). The defect region of the presswork image waiting for detection would be extracted with the image difference and adaptive threshold binaryzation, and the position and geometry information of the defect would be obtained with the Blob connectivity analysis. Finally the presswork image waiting for detection would be marked with the defect.In the hardware integration and software realization of the online defect detection system, the hardware experimental platform based on the computer, machine vision imaging system and the assembly motion work platform has been established, and the presswork defect online detection software system based on the machine vision has been realized by taking advantage of the LabVIEW development platform. The major work for the system realization mainly includes the following points:(1). The composition of the hardware platform and the overall structure of the software system would be analyzed and designed;(2). The CCD camera of the machine vision system would be analyzed and selected, proper light source would be chosen and the lighting solution would be designed;(3). Modular implementation would be conducted for the software system according to different functions, and software system’s drive for the hardware would be realized;(4). Various functional modules would be integrated by the system to realize the normal operation of the complete system by combining the software system with the hardware equipment.Through theoretical study and experimental analysis, the related algorithm for the key presswork defect detection technology, including the moving target detection and tracking, image registration, local image mosaic, etc. has been proved to be effective and feasible. Meanwhile, the defect detection system realized in this paper has been equipped with good performance and reliability. |
PDF Full Text Request