Research On The Visual Inspection System For The Appearance Quality Of Cigarette Rods

Cigarette appearance quality inspection system is based on machine vision technology,specifically for the detection of defective smoke of an automated equipment,including a combination of optical,electronic,mechanical,computer,including a variety of disciplines,its use not only help improve Production line automation,improve productivity and production efficiency,but also to ensure product quality,has a good prospect.In this paper,the work flow of the system and the hardware structure of the system are studied.The advantages and disadvantages of the system hardware structure are analyzed,and a hardware system is designed for smoke appearance inspection.Then the image smoothing algorithm and the image sharpening algorithm are studied.Aiming at the image smoothing algorithm,the advantages and disadvantages of the mean smoothing filter,the bilateral filter,the non-local mean filter are analyzed,the non-local mean filter is improved,and the image smoothing operation is realized under the condition of preserving the edge information.And then the operator of the LoG operator is improved and the sharpening operation of sharpening edge natural transition is realized.Secondly,the detection algorithm of tobacco smoke image localization,smoke rod inspection and smoke end detection is studied.In the end detection of tobacco rod,the method of artificial neural network is introduced.The method of sample training is used to study the characteristics of the end face,and the structure of neural network is constructed.The method has wider applicability.Finally,based on the above research,this paper develops a software system that matches the hardware device,analyzes the main interface of the system,the loading grade module,the image recording module,the work record module,the system management work flow and the corresponding composition structure and realizes User-friendly and aesthetics of the software interface.