Defect Detection Of Curved Brim Glass Based On Machine Vision

With the progress of intelligence,the number of robots is increasing rapidly,artificial intelligence has been fully developed.Machine vision is a technology that USES a high-performance camera and other acquisition devices to obtain images that need to be analyzed,and then separates the required information from them,so as to realize the function of r eplacing human eyes.Glass is widely seen in the current society,edge grinding is the technical requirement for the high-level development of glass processing.The high precision and long duration of machine vision satisfy the requirements of quality inspection and realize the rapid and efficient development of glass manufacturing industry.The machine vision image acquisition system is signed according to the defect characteristics of the curve brim glass.From the processing mechanism and grinding method of glass,the forming principle of glass defects(bright spot,white line,chip edge)is analyzed.Because the detection surface is a symmetric surface,in order to achieve uniform irradiation of light source,the structure design of strip LED light source parallel to the detection surface is adopted.The features of white line and bright spot can be completely displayed when they are in the same plane with the camera through the irradiation of the light source.In the detection of the blasting edge,since the blasting edge is small,when the camera is at a 45o angle with the glass,the blasting feature can be expanded and the detection process can be simplified.Using a mature image processing software HALCON recognition,defect images were collected by artificial vis ion and machine vision of comprehensive analysis to determine the edge of the glass defects detection.On the basis of the existing algor ithm,bright spots are separated by threshold values,histogram equalization enhances the characteristics of chip edge,double smoothing combined with white line features.The feature extraction methods of bright spot,white line and chip edge are designed.The object-oriented design idea is adopted to realize the operationalization of the detection system through the MFC database of VC++,and the data is saved to facilitate the maintenance and update of the system in the future.Through the combination of theory and practice,the overall design ideas are constantly adjusted to increase the feasibility and advancement of the design.