Machine vision based recognition and integrity inspection of printing characters on food package
Abstract
Food packaging bags are fast-moving consumer products with large output and fast production speed. Instant noodle bags need to be printed on the production date and number before packaging the food. For the possible date or code printing error, character sticking, incompleteness, etc., the correctness of the characters cannot be judged in real time by the human eye. A set of automatic recognition characters and detection systems are used to binarize, filter and tilt the picture, and use image morphology and horizontal and vertical projection to segment a single character. BP neural network is used to identify the character and compare it to the template to obtain the degree of defect and compare it with the threshold to determine whether the character is qualified. The detection speed of the system can reach 3 images per second, which can realize real-time recognition and detection of characters, so as to eliminate non-conforming products in time.
- Publication:
-
2019 International Conference on Optical Instruments and Technology: Optoelectronic Measurement Technology and Systems
- Pub Date:
- March 2020
- DOI:
- 10.1117/12.2550049
- Bibcode:
- 2020SPIE11439E..1FJ