Research And Design Of Intelligent Machine Vision Infusion Foreign Matter Detection System

Large Volume Parenteral(LVP) is one of the five important classes ofpharmaceutical products. Due to the particularity of LVP, all countries made strictproduct quality inspection on the infusion. The traditional method is to check directlythrough artificial vision.Not only does it consume a lot of manpower, resulting inincreased production cost, but also restricts the efficiency of the entire productionline.Based on this background,the article studied smart infusion foreign objectdetection systems based on machine vision. With the integration of image processingtechnology, electronics and communication, mechanical design, control technologysystematically, according to the characteristics of foreign material in LVP,the articledesigned mechanical detection system and supporting imaging light source system,analyzing foreign body imaging characteristics, presenting a strong robustness targetrecognition algorithm and the completion of defective separation with foreign bodyelectrical control system. The system has the characteristics of non-contact,high-precision, detection speed and other.The main research contents are as follows:1、Current research of LVP foreign material detection domestic and abroad isintroduced, smart infusion foreign object detection system is proposed based onmachine vision against the problems in medical infusion detection, machine visiontechnology is introduced briefly, the difficulty in identifying large infusion of foreignbodies is also analyzed.2、According to the requirements of pharmaceutical production line, the overalldesign of intelligent infusion light inspection machine is proposed，artificial lightinspection movement and large infusion "rotation-stop" stepping type detectionmethod is also proposed. Key mechanical locations such as bottling unit,bottle-wiping device, brake device，collecting device and dividing device are designed,scheme of light source and imaging system is studied.3、Foreign object image detection method in smart infusion monitoring systemis elaborated in detail. Processing the collected image sequences in advance, usingimage registration to eliminate the influence of mechanical jitter caused by imageacquisition, using weighted average filtering algorithm value-based to eliminateimage noise, by choosing a reasonable threshold calibration region of interest andpartial image enhancement after inter secondary differential operation to extract the characteristics of foreign body,and ultimately measuring the particle size of foreignmatter to determine whether the product is qualified.4、According to the principle of large infusion foreign body recognition,intelligent infusion of light inspection machine electric control system is designed andimplemented, ensuring the accuracy and stability of product testing and the separationof the defective goods.5、A software interface of foreign object detection system is developed, LVP ofintelligent light inspection machine prototype is researched, Knapp-Kushner andrepetitive experiments are applied to the prototype to detect the system’s stability andaccuracy, and the results are analyzed.The article analyses the intelligent infusion light inspection machine theoretically,designs the prototype through the reasonable machine vision system, electric controlsystem as well as mechanical structure. Test results has showed that the medicalinfusion foreign body recognition system can solve part of the actual productionproblems, has a good application prospect.