Research on High Precision Detection Method of Register Error Based on Machine Vision

Register accuracy is an important index to measure the quality of printed electronic products. But the register error detection accuracy of photoelectric sensors cannot meet the requirements of printed electronic register control. For this reason, this paper proposes a sub-pixel-level register error detection method based on machine vision for circular matrix color marks. First, the image preprocessing operation was completed according to the application scene and color mark features. Then, sub-pixel edge extraction was accomplished using the improved Canny operator and Zernike moment method, and high-precision registration error detection was accomplished using least squares fitting. The results show that, compared with the traditional detection methods, this method has better rapidity, accuracy and stability in register error detection. It solves the difficulties of image preprocessing and sub-pixel-level detection under the condition of multi-color groups, and meets the demand for high-precision control of complex printing electronic devices.