A Detection And Identification System Based On Machine Vision For Glass Ball Impairment

Glass balls are widely used in many fields due to their good chemical stability,high mechanical strength and good electrical insulation.At present,the domestic glass production capacity is huge,but the detection and identification technology of glass ball impairment is backward,mostly using the traditional manual method.Traditional detection method results in visual fatigue and damage,low efficiency and high error rate of detection and reducing the quality of the products,in urgent need of an automatic detection and identification system based on machine vision for glass ball impairment.This paper proposes an detection and identification system based on machine vision for glass ball impairment.The system is mainly divided into two parts of hardware and software,the software part includes designing the detection and identification algorithm,obtaining the glass ball impairment information;the hardware part mainly completes the image acquisition,glass ball transmission and remove defective products.The main innovative works of this paper are as follows:1.Glass ball image acquisition.We use a model MV-EM120C/M array CCD camera to take photos.A LED lamp is light source with high illumination-angle.We propose a method that combinate the position of the light source and brightness with camera angle and its filed depth to take photos.We use black velvet as background,which increases the measured object and background contrast,and play a role on fix the glass balls prevent rolling.2.Image preprocessing.We combine following methods to preprocess the glass ball image: the image enhancement with wavelet transform,Hough transform method of plane circle recognition,LOG operator edge detection and mathematical morphology filling function.These methods can highlight the outline edge and get rid of background of the glass ball image.3.Feature extraction and defect detection.We use an area measurement method based on region marking to extract the defects of the glass ball image.Then calculate the defect threshold and get the detection results.4.Create GUI visualization operating system interface.More intuitive operation to avoid lengthy procedures,processing results is clear,simple way to operate and easy to interact.