Research And Implementation Of Flaw Detection Algorithm For Filtered Cigarette

Automatic inspection of product quality is an effective way to improve the efficiency of production.Tobacco is a pillar industry of Yunnan Province,and the production of cigarettes has been highly automated.However,there will be a certain proportion of unqualified cigarettes on the high-speed cigarette production line that the speed totally between 120 and 200 sticks per seconds.if we test these cigarett es manually,the production efficiency of the production line will be reduced seriously.So,designing a simple cigarettes appearance detection algorithm to testing the unqualified cigarette is a necessary way to improve the efficiency.Based on the basic method of digital image processing,this thesis design and achieve a cigarette flaw detection algorithm,the algorithm is capable of distinguishing the flawed cigarettes and the type of flaw of defective cigarette can be determined roughly.The work of this paper will be described as follows:(1)Summarizing some method of flaw detection about product appearance.(2)Making a research about the pretreatment method of cigarette image.In the preprocessing stage,we did a filtering process for cigarette image in order to reducing the effect of noise;and then,separate the single cigarette picture from the image for the convenience of later treatment;As we know,the color difference between the filter and the tobacco stick is large,so finally,for the convenience of flaw distinction,canny algorithm was used to extract the outer edge,and the filter and smoke sticks in the cigarette were separated.(3)Making a research about the algorithm of the flaw detection in tobacco products.For the common flaw types of the tobacco,such as wrong branches of cigarettes,partial wrinkles,dark spots of filters,spots of tobacco rods,punctures of tobacco rods and so on.firstly we performed spot detection,then classify the flaw by analyzing the area ratio of the incomplete part of the grayscale image;and a method of multiple detections that separate the trademark was proposed,usually the trademark or a pattern in filter will be misidentified as a dark spots of filters,the new algorithm we proposed can reduce the false recognition rate and come true the detection of cigarette flaw.(4)Using matlab2014 a,we designed and implemented the algorithm of the flaw detection,which provided a reliable basis for high-speed automated cigarette production line.