Research And Design Of Pharmaceutical Packaging Detection System Based On Machine Vision

The detection of the drugs packaging is extremely important,its accuracy and efficiency is not only directly related to the drug classification,storage,and security of the drug taking,but also affect the economic benefits of the manufacturer.The existing means of the drugs packaging are mostly rely on manual operation,the miss rate and the efficiency are both very low,and it is easy to produce false detection.Meanwhile it needs a lot of labor costs and the work is boring.Aiming at the problems above,this dissertation is based on machine vision technology,the image denoising method and the extracting method of the fault features face to the drug packaging are put forward,a set of the intelligent drug packaging detection system is designed,and a common platform of defect detection system for the liquid packaging is developed.According to industrial environment of drugs packaging production line,the quantity problems of mis-packaging,miss packaging,the dose of soup,the label defect and the code error can be detected accurately and real-timely.At the same time,the mechanical design of the machine vision system and the whole detection system are made,and a reference scheme is put forward for their organic integration.This dissertation introduces theory of the machine vision and the image processing technology,including its history,present situation,characteristics,application field and developing direction in the future.The environment of the drug packaging production was analyzed,and the features of the device was studied.The drug packaging detection strategy based on the machine vision was proposed,the system framework was designed,and the hardware was chosed.,the related mechanical principle has carried on the detailed analysis.For liquid packaging defects monitoring,the machine vision processing method was further studied,the defect images were analyzed and processed,as well as the realization of the mechanical separation after image processing steps.The neighborhood averaging method and median filtering method were applicated to suppress the image noises,and the smooth image edge was obtained;Combining the threshold method and edge detection for image segmentation,the binarization images were gained;After the intensive study of traditional two-dimensional threshold segmentation method,a fast Otsu algorithm based on two-dimensional histogram was proposed,two-dimensional threshold was converted to one dimensional threshold,and preliminary search first through compressedhistogram,then go back to the original two-dimensional histogram precise fast algorithm,and computational complexity was reduced significantly;Finally,the judgment criterion based on tags and level threshold is proposed.Visual programming language VB 6.0 was applicated for modular design,aiming at the real-time requirements,the programme framework and process is optimized.