Screen-Printed Ink Protrusion Defect Detection Method for Mobile Phone Cover Glass

During the screen-printing process of mobile phone cover glass, factors such as ink accumulation, non-uniform squeegee scraping, and environmental vibration can easily induce tiny ink-protrusion defects along the edge region. These defects are characterized by small size and weak grayscale contrast, and are often accompanied by glass specular reflections and screen-printing texture interference, posing significant challenges for online inspection. To address this problem, this paper proposes a screen-printed ink protrusion defect detection method based on image enhancement and row-wise scanning statistics. First, bilateral filtering is applied to enhance the screen-printed edges, suppressing noise while preserving edge structures and improving the discriminability of low-contrast defects. Next, the bottom contour is extracted using Canny edge detection, and a bottom reference line is selected via Hough line fitting; this line is then used to partition the inspection region and establish row-scanning baselines. Finally, leveraging the local grayscale anomaly of protrusions along the row direction, a row-wise pixel accumulation and ratio-thresholding strategy is designed: edge feature values are computed for each row, abnormal rows whose responses rise above a certain proportion of the row-wise mean are selected, and a consecutive-row triggering mechanism is introduced to achieve stable identification and localization of protrusion defects. Experimental results demonstrate that the proposed method can effectively detect screen-printed ink protrusion defects and exhibits good robustness under complex conditions involving reflections and texture interference, satisfying the requirements of practical production lines for online inspection of screen-printing defects.