Research On Defect Detection Technology Of Capsules Cigarette Based On Machine Vision

At present,in the cigarette market,the market share of innovative cigarettes represented by "capsules cigarette and fine cigarettes" is rising year by year.the trend breaking through the sales bottleneck of cigarette market.In order to ensure the quality of capsules cigarette,it is necessary to control the quality of capsules before entering the filter rod,and to control the surface quality of cigarette during production.Traditional artificial detection mode cannot meet the high efficient production at the same time,also some problems such as leakage inspection by mistake,because cigarette surface quality detection are greatly influenced by workers’ subjective,not conform to the requirements of the national standard,therefore,in this paper,based on machine vision and image processing technology,the combination of the key technology of defect detection in capsules cigarette were studied.The following contents are mainly studied:According to the related technical requirements of the defect detection of capsules cigarette,an overall intelligent defect detection scheme of capsules cigarette based on machine vision was designed.Completed the construction of the defect detection experimental platform,and designed the input module,light source lighting module,image acquisition module,unqualified product elimination module in the detection platform.According to the related detection requirements of surface defects of capsules and cigarette sticks,an intelligent detection platform for defects of capsules and cigarette sticks was built.Aiming at the defect characteristics of the shape and appearance of the capsules,a defect detection algorithm based on image processing technology was studied.In order to ensure the single particle extraction in the processing process,a method based on improved inter-class variance was proposed to achieve the single particle image segmentation extraction.In order to complete the discoloration defect detection algorithm based on gray analysis was proposed for the capsules discoloration defect detection.Aiming at the bubbles in the capsules caused by the capsules breakage and leakage,a detection algorithm of capsules breakage defect based on nonlinear transformation was proposed.Aiming at the unqualified size defect of capsules,a detection algorithm of unqualified size defect based on Bayesian estimation was proposed.Through the synchronous detection of multiple defects of capsules,the experimental results show that the multi-defect detection algorithm based on image processing proposed in this paper can efficiently and accurately realize the intelligent defect identification and detection of the quality of exploding beads in the production of capsules cigarette rolls,and the accuracy meets the production requirementsIn order to detect the creases,smudging,damage,wrong teeth and other defects on the surface of capsules cigarette,we proposed a detection algorithm based on Fourier Transform(FFT),and the features of local micro-defects on the surface of cigarette were extracted and detected by FFT.In order to further improve the detection speed and intelligence of cigarette surface,a cigarette defect classification model algorithm based on deep learning was proposed.Through the classification model training of defective cigarettes and qualified cigarettes,the efficient intelligent and accurate detection of cigarette surface defects was realized.Experimental results show that the accuracy of the detection algorithm based on deep learning cigarette defect classification model is above 98%,and the detection speed of each image can reach 7.89 ms.According to the defect inspection of capsules cigarette,develop and design an intelligent defect detection system of capsules cigarette based on HALCON and Visual Studio software development platform,the analysis and design the relevant test procedure,to the humanized design of software interface,in a state and satisfy the real-time display detection at the same time,the results of complete preservation and start-stop operation functions such as design,In experiment and production application has good results,and it has significant industrial popularization value.