Machine Vision for Detecting Defects in Liquid Bottles: An Industrial Application for Food and Packaging Sector

Abstract

The quality control of liquid packaging, such as cooking oils and beverages (bottled water, soft drinks, juices, etc.), is crucial due to the inherent risk of leakage. This process involves inspecting bottles for cap and seal ring defects and addressing issues arising from the gradual degradation of filling machines, leading to variations in the surface level of liquid bottles over time. Additionally, proper label placement significantly contributes to the customer-friendliness of a product. This research aims to introduce an automated vision-based rating system designed for the online inspection of defects in liquid bottles. The system is versatile, applicable to both academic and industrial settings, and can be easily adapted for use with various types of transparent liquid bottles. The defect detection metrics include three measures of distance determination and pattern matching. The equipment used in this study includes a Complementary Metal Oxide Semiconductor (CMOS) camera with a USB connection, a laptop, and a 14-speed conveyor belt, among other components. The system demonstrated an average accuracy of 95.6%, with specific accuracies for surface level, cap, and label placement at 100%, 95%, and 92%, respectively.