Optical Character Verification of Print on Pharmaceutical Capsules

In this paper, the problem of print quality inspection on pharmaceutical capsules is addressed. Good print quality, a property that makes the print legible, is of utmost importance for identification and preventing mix-ups among various types of capsules. Spatial distortion obtained at image formation process, where a 3-D surface gets projected onto a 2-D image plane, causes print appearance variations. Elimination or compensation of print appearance variations is necessary for improving defect detection performance of inspection systems. For that reason, a novel appearance-based method for print quality inspection on capsules is proposed. The performance of the proposed method was evaluated on a real capsule image database of printed capsules, where a "gold standard" was established by manually classifying the capsules into non-defective and defective class. The obtained results indicate that the proposed method is more effective than the standard defect detection method.