Research On The Detection Of Facial Defect In Resin Glasses Based On Machine Vision Technology

Glasses have become an indispensable part of modern life,not only in correcting vision,but also in decoration.As of 2017,China’s glasses production reached as high as 648.582,accounting for 70% of the world,and the overall trend is on the rise.At present,China’s glasses testing still stays in the manual testing stage whose biggest problem is the superiority and inferiority of the product and the qualification judgment depend entirely on the experience and level of the inspectors,resulting in the inconsistency of the actual judgment results.In this paper,a glasses-based surface flaw detection device based on machine vision was developed.With the theoretical support of optical characteristics analysis and optical path simulation research of glasses surface flaw,the machine vision system including visual inspection and automatic feeding is designed to lay the foundation for the automatic detection of the glasses.The main research contents of the thesis are as follows:1.Study on the optical imaging method of the glasses surface flaws.Based on the basis of simulation theory,the projection imaging optical simulation model is established.Based on the raster icon and scatterer icon,the imaging effects of different illumination schemes on surface flaws are compared and analyzed,and the point source projection illumination scheme is determined and optimized.By adding a collimating lens to the front of the point source,The density of received light in the flawed area on the receiving surface is 3.7%,and in the normal area is 63%,the best imaging effect of the planar flaws can be achieved.which provides a theoretical basis and reference for the later experiments.2.Design and of glasses image acquisition device.According to the results of optical simulation results,the device suitable for the detection of glasses surface flaws was designed.The components of the device such as industrial camera,lens and point light source are determined according to the process requirements and basic tests.And correspondingly the visual inspection box was designed.3.Research on image processing methods for glasses facial flaws.The binary image is generated by region growth for background segmentation,and hole filling is used to remove internal holes due to region growth.Then,the mask image is dot-multiplied with the glasses image to obtain the image with background removed.Mean filtering,edge detection,threshold segmentation,morphology and feature screening are used to extract the glasses surface flaws.The algorithm recognition success rate is up to 95.91%,and the average detection time of the glasses is 794 ms.The recognition rate and real-time performance of the test meet the online detection requirements.4.Research on automatic detection system for glasses surface rickets.A glasses detection system based on machine vision inspection technology was designed independently,including the loading system,rotation system and detection system of the glasses.The glasses loading system can load different sizes and different thickness glasses,and the feeding success rate is up to 100%,which satisfies the accurate feeding requirements.According to the requirements of glasses clamping,the clamping device was designed to clamp different sizes and different thickness glasses without damaging the edge,and the clamping success rate is up to 100%.Detect the glasses detection system on-line,the detection success rate is 95.48%,which satisfies the detection of resin glasses.