Research On Plane Defect Detection Method Of Optical Communication Devices

Optical communication technology plays an important role in information transmission technology,and the market demand for optical communication equipment is increasing rapidly.As a core component in optical communication equipment,the processing quality of optical communication devices has an important impact on the performance of the equipment,so it must be strictly tested before it leaves the factory.At present,the detection mainly adopts the method of manual detection.The detection result is subjectively influenced and the efficiency is low.To solve this problem,this paper develops a vision-based optical communication device defect detection software system.Firstly,the inspection items and indicators are analyzed,and the visual acquisition system and software system are designed.The visual acquisition system consists of a camera and a light source.The left and right cameras are used for photographing.The middle camera image is used for plane defect detection.The left and right cameras form a binocular stereo vision system for stereoscopic defect detection.The light source uses coaxial light.Combined lighting mode with low ring light.The software system mainly includes two modules: template creation and product detection.Secondly,according to the matching requirements of the intermediate image and the left and right images,different template matching methods are designed.For intermediate images,PD(photodiode)and TIA(transimpedance amplifier)use internal edge shape template matching due to more effective information on the inner edge.Capacitance is scaled due to the clear outer contour and large tolerance range.The outer contour shape edge template matches the way.For the left and right images,considering that the components above satisfy the homography,an anisotropic shape template matching method based on the homography transformation is adopted.The matching method adopted in this paper has high stability and meets the detection requirements.Then,according to the imaging characteristics of the components and the base on the product,the corresponding defect detection algorithm is designed.For the damage and contamination defects of the component,the difference template is used for detection,and the defect area whose variation exceeds the range is extracted by specifying the range of variation allowed for each pixel.For the base crushing defect,the color space feature similarity is used to detect,the suspected region is obtained under the coaxial light,the color space feature is extracted under the low ring light,and finally the chi-square distance is used to determine whether it is a defect.The defect detection method adopted in this paper has high accuracy.Finally,a large number of test experiments were carried out using the samples provided by the company.The experimental results show that the components can be successfully matched under different conditions.The false detection rates of component damage,contamination and base crush defect detection algorithms are 2.9% and 4.63%,respectively.The missed detection rate is 0%.The overall false detection rate of online testing was 4.24%,and the missed detection rate was 0%.Therefore,the matching method and detection algorithm adopted in this paper have high stability and accuracy.