Research On Light Treatments And Feature Matching Methods Of Blister Drug Package Defect Detection

As State Drug Administration pays more attention to therequirement of sterilisation for the pharmaceutical production workshop,the mechanization and automation for the tablets production is needed tobe strengthened. And the way of manual inspection of the lacking unitand the pill completeness is far from enough to satisfy the need ofindustrialization. Furthermore, the lacking in the statistic lie in thecurrent pill checking system and medicine check efficiency. However,blister and intelligent machine vision detection system can solve thisproblem. The system which can take the place of eyes detection withoutthe disadvantage of fatigue is more of high speed and accuracy, whichcan make the production more consistent with GMP certification.This issue is based on the current check system of blister pills,studying how to receive and store the valid data from the intelligentcamera, then analyze and process the data. The server is based on theWindows PC server, and the hardware platform is account of the blisterchecking system that lie in the intelligent camera. The detected contentincluding pills lacking, pill incompleting, blister crack, surface foreignmaterial and so on. On the foundation of the error data，analyzing thedata in a multidimensional way, which will result in the solution to dealwith the problems of medicine production, improving enterprisemanufactur-ing management level and product quality.This paper proposes a application by SQI method,which aims atimproving its detection affect affecting by indoor light interference. Weput forward a Matching method of the aluminum-plastic blister drugspackaging.The method is based on LBP histogram and SIFT for thedefects detection. First, extract the feature points in the image ofaluminum-plastic blister drugs packaging, and then calculate the LBPhistogramon the surrounding area. At last match the key points to judgeif there is defect on the medicine packaging.The results show that SQI and feature matching method in thisissue have apparent effect on the blister detection, which can indicate the phenomenon of reflect light, lacking pill, pill incompleting, pill crack inthe blister.