Machine Vision Based Detection And Location Technology For Filling Of Drum Oil

The recognition and localization of eccentric bungholes in cylinder barrels based on machine vision has been widely used in heavy industrial applications like petroleum and cement shipping.This thesis aims to employ machine vision technology to deal with the problem of automatic filling of edible oil in food industries.Aims for that,eccentric filling ports recognition and location of edible oil drums in the automatic filling systems is performed using machine vision technology.It is well-known that sanitary condition is a high requirement in food industries.In order to make sure that the sanitary conditions during the entire oil filling process,a cover is needed on the filling mouth to make the oil drum be sealed.The purpose of this work is to recognize and locate the cover for its get and place using machine hand to realize automatic filling of edible oil.For this purpose,a method of recognition and location of eccentric port in edible oil drums based on grey image and color image processing.The main contents are as below.(1)Edge detection of oil drum images is performed using Canny operator.The circle center of oil drum,the circle center of white plastic cover and the circle center of yellow metallic cover are recognized and located.The rotating direction of the metallic cover snap joint is determined.(2)The circle centers of white plastic cover and yellow metallic cover are recognized and located based on their image HSV(hue,saturation,value)color components.In this stage,a dynamic threshold of image coordinate and the angle direction of the minimum enclosing rectangle of the image pixels are extracted.The vector of two pixels which are close to parallel are regarded as the rotating direction of the cover snap joint.(3)In order to deal with the problem of low accuracy of metal cover recognition due to the low contrast between the cover and the oil drum,machine vision technology and image processing algorithms are used to recognize the filling mouth of the oil drum.To this end,the influence of low color contrast on object detection is analyzed based on the color image characteristics.Then the image is enhanced to improve its quality using convolution operation that enhances the high frequency component and weakens the low frequency component.(4)Experiments are carried out to evaluate the approach performance.The experimental results indicate that the proposed method is effective to recognize the filling mouth of oil drums in the automatic edible oil filling systems,which provides a new technology for automatic production of edible oil filling and has great values for engineering applications.