Research Of Defect Detection About Aluminum-Plastic Blister Package Based On Machine Vision

The aluminum-plastic blister packaging, has been widely used and welcomed by themarket, because of the advantages of simple manufacture, convenient carrying. It’s alsonamed press through packaging(PTP).The traditional defect inspection of PTP by labor istime-consuming，tiring，inaccurate. In this paper, a set of solution about defect detection ofthe aluminum-plastic blister package based on machine vision, is developed,comprehensive used of a variety of image processing technology, combined with thehardware features with implementation of the relevant software design, and performanceoptimization. The overall system obtained ideal effect.Firstly, the hardware system and the software algorithm is demonstrated, finallychoosing TMS320DM6437as core processor for machine vision applications, Beside，the video decoder driver software is designed in this paper. The completion of SAA7111Adriver of operating system solved the problem of image acquisition. In addition, the publicinfrastructure software framework is completed with task coordination of networkcommunication, video acquisition and processing, key operation and so on. This baseproject is the foundation of the top level application of machine vision. Finally, pill defectdetection algorithm and the corresponding core software is decrypted step-by-step withexperimental pill image, combining software optimizes and the results analysis.The experimental results show that the system algorithm, can realize the pill breakagedetection accuracy of12.5%, speed rate of8frames/sec, the accuracy rate93.8%, withthe condition of image resolution720x480and using natural light. The overall system isrun in self-learning way, without human intervention, not effecting by the pill shape andarrangement. It reached good stability, high robustness.