Study On Image Registration Technology Of Offset Quality Automatic Detection

The offset printing quality checking is an important part of printing process. The automatic detection with high speed and precision to presswork is the inevitable trend of printing quality checking technique. The efficiency of traditional manual inspection is low and cost is high. So, the system of automatic offset printing quality checking is needed. Based on a lot of references, the existing algorithms were analyzed. The image was registered by synthesizing corner registration and template matching and the SUSAN corner detection algorithm was improved to meet the demand of automatic offset printing quality checking, reduce the laboring intensity and improve the productivity.The printing image was matched by rotation and translation synthesizing corner matching with template matching. Based on cluster theory of pattern recognition, the SUSAN algorithm was improved according to the corner detection method. The quality of image was enhanced by the means of double line spline. The experimental projects were designed according to black box theory. The program was complied with VC ++ 6.0.The SUSAN detection algorithm has been improved and the fixed threshold't'in the former algorithm can change with the difference of contrast in different areas of the image. The threshold increases along with the contrast. The parameters of defection were obtained by using the Blob technique to analyze print defection. The standard images were detected using alterable threshold SUSAN algorithm and the algorithm was evaluated according to detection results.The result shows: The alterable threshold corner algorithm can distill various corners exactly and it can apply in the low contrast area. The image matching time is 20% of template matching method by synthesizing the corner orientation and template matching. According to the parameter of detective image, the detection precision can be set and the line or point defections can be judged.