Computer vision detection of glass bottle dimensions

In the process of quality inspection of glass products, with the expansion of production scale, the improvement of production speed and more and more strict quality requirements, the traditional manual inspection methods are no longer competent.In this case, many foreign manufacturers have begun to develop for the quality of glass bottles testing machines.China is relatively backward in the development of glass bottle quality testing machine, at present some domestic manufacturers are also developing for the glass bottle quality testing machine, they are generally copy foreign products, the development work is still in progress.From the point of view of the products developed abroad, in the aspect of the glass bottle size detection, generally use the mechanical contact way, and this way requires a high level of mechanical manufacturing technology.The computer vision inspection system of glass bottle size designed by the author is a subsystem of the computer vision on-line inspection system of glass products developed by the Institute of Electronic Technology of Guangxi Normal University and guilin glass factory.This system avoids the weakness of China’s low level of mechanical manufacturing technology, adopts a non-contact sensing method, and USES computer vision and image processing technology to detect the dimensions of glass bottles.The test contents are: the inside diameter and outside diameter of the bottle mouth, the height of the bottle and the perpendicularity of the bottle.When the detection system detects the dimensions of a bottle, two cameras are needed to collect two images respectively.One is the bottle mouth image, which is taken by the industrial camera perpendicular to the bottle mouth. It is used to detect whether the inside diameter and outside diameter of the bottle mouth and the perpendicularity of the bottle are qualified.The other is a bottle height image, taken by an industrial camera looking horizontally at the top half of the bottle to see if the height of the bottle is correct.The system USES the external trigger mode to control the camera for image acquisition, that is, when the detected bottle arrives at the detection station, the external trigger circuit generates a trigger signal and sends it to the image acquisition card.The computer detects the external trigger signal and immediately controls the camera for image acquisition.The system adopts the method of first calibration and then detection, that is, a standard size is determined by using the external size of the standard bottle. During the detection, the size of the tested bottle is compared with the standard size to see whether the deviation is within the allowed range, so as to determine whether the external size of the tested bottle is qualified.The system software consists of two functional modules, one is the bottle mouth image processing module, the other is the bottle height image processing module.Bottle mouth image processing module includes bottle mouth image acquisition, image edge detection, bottle mouth inner diameter and outer diameter corresponding to the inner circle and outer circle detection, bottle mouth inner diameter and outer diameter dimension analysis and perpendicularity analysis.The bottle height image processing module includes the collection of bottle height image, the detection of bottle contour edge, the determination of the line where the upper edge of bottle mouth is located, and the qualified analysis of height.In the edge detection of bottle mouth image and bottle height image, the method of edge extraction using gray threshold segmentation is adopted instead of edge detection using edge detection operator.When detecting the inner circle and outer circle of the bottle mouth in the bottle mouth image, the author puts forward two methods of finding the center of the circle by the vertical bisector of the half-split chord, and decides to use the half-split method to detect the inner circle and outer circle of the bottle mouth through experimental comparison.In the whole process of software development, the author designs algorithms and writes programs from two aspects of speed and effect.The manufacturing cost of the detection system is low and the precision of the mechanical manufacturing is low.And the detection speed of the system can be improved with the increase of CPU speed.The author USES Visual C++ to complete the software development of glass bottle size detection.The detection system has successfully realized the detection of glass bottle size in the experimental stage.


Post time: 11月-25-2020
+86-180 5211 8905