| With the development of printing technology and production,the higher request about printing quality and diversity has been put forward by the customers. And the modern detection technology of machine-vision-based has become the main trend in the printing industry. In this paper, lots of printing surface quality testing theory and detection algorithm has been studied and researched, which is used for corresponding wide printing format and high precision detection. And then, a machine-vision-based inspection system for full screen printed matter has been presented.(1) In this paper, the detecting system framework is researched and composed by image acquisition, image processing, data management and signal control functional modules. Meanwhile, through cooperation between the modules, the defect detection task has been completed.(2) In the image matching process about the position space between the standard image and tested image, this paper presents an improved image matching method, in which the extractive objects and matching range will be specified. By use of this algorithm, the amount of algorithm calculation is reduced, the matching time is shortened, and the performance of accuracy and real-time has been guaranteed.(3) In respect of defect feature analysis, the detection system respectively presents defect feature analysis methods for surface defects and color deficiency. By using the CIE Lab chromaticity space and CIE 1976 Lab color diference formula, the estimate of whether the pixels in image have color difference has been done. Based on using the Blob analysis method , this paper present an improved per-pixel scanning marker methods for surface defects, only after one scan, that can compute the marker work of all connected domain.(4) In respect of surface defect recognition, different recognition rules have been defined for point, Line or surface defect detection. And then, verified by experiment system, the system can now realize the common defect detection. |
PDF Full Text Request