The Research of Automatic Registering Detection of Rotary Screen Printing Machine Based on MeanShift and Fast Block-Matching Algorithm

With the register detection problem, the color textile image segmentation algorithm based on MeanShift and the block matching algorithm based on Harris corner detection were proposed. The extended MeanShift algorithm was used to segment the textile image, and then different color regions were extracted from the segmented standard image; then the feature points were detected by Harris operator, and with these feature points as the centers, the standard matching blocks were selected; finally the best matching block in the dealt target image were found to calculate register errors. The experiment results show that MeanShift algorithm is not sensitive to the texture of the fabric image, so the algorithm is good for the segmentation of it. And the matching accuracy and speed are improved, because the feature block-matching algorithm combines the point pattern matching with the feature matching algorithm. The research in this paper will lay the foundation for the closed-loop control of online register detection.