Detection of product container defect using automatic visual inspection: System design and performance evaluation

The fourth industrial revolution brings the capability of manufacturer to improve efficiency in quality control. Enhanced technology of computer and camera enables automatic visual inspection to fully replace human task in quality control. A quality control of container is examined by using automatic visual inspection to determine the defected container based on defected criteria. The challenge of visual inspection is inconsistency in image acquisition due to lighting, inconsistent adjustment position, or mechanic speed in both camera and container object. A system of automatic visual inspection is designed to cope with image consistency using match pattern reference coordinate. The coordinate is used as an anchor point to make measurement algorithm in container image feature. The result shows that using reference coordinate minimize probability of false alarm up to 0.5%. It has tradeoff in increased computational complexity which occupies up to 63% of overall execution speed.