Research on Halcon-Based Quality Inspection Algorithm for Bottom Coding of Beverage Cans

To address the limitations of traditional coding quality inspection methods, including low character-region localization accuracy, poor adaptability to complex environments, and insufficient character-integrity verification, this study designed a Halcon-based machine vision inspection system. A novel algorithm named MIRIAM-DotQA (Multi-stage inscribed-circle calibrated ring-constrained inspection algorithm with morphological optimization for dot-matrix quality assessment) is proposed. First, the reference coordinate system of the can bottom is precisely calibrated through dynamic threshold segmentation and inscribed circle alignment, while morphological optimization eliminates reflective interference and fits a minimum enclosing circle. Second, under annular region of interest (ROI) constraints, multi-stage morphological processing segments the bottom region, with character regions identified via dual thresholds of area and geometric dimensions. Finally, affine transformation corrects rotational deviations, dot-matrix features are enhanced, and dynamic adhesion compensation is applied. The pre-trained DotPrint model in Halcon integrates these operations to verify character integrity, enabling quality assessment based on character quantity and topological distribution relationships. Experimental results demonstrate a 98.5% detection rate for defects (misprints, omissions, and chaotic arrangements) with an average processing time of 11.1 ms. The MIRIAM-DotQA algorithm exhibits high robustness, rapid detection speed, and precise defect identification capabilities.