Research And Exploration Of New Optical Cable Pitch Detection Method

With the development of artificial intelligence and machine vision technology,the disadvantages of traditional artificial detection are greatly reduced,and the demand of production modernization and high quality is further promoted,especially in the dangerous conditions of artificial.Optical cable is the basis of modern information transmission,among which optical cable is a practical cable product made by processing and manufacturing the finished optical fiber of ruoegan（1-2160）through processes such as molding,stranding,sheath extrusion and armor^[1].The pitch quality detection after stranding has always been a difficult problem in the industry,and the manual detection method has low efficiency and high cost.Compared with human,machine vision technology has the advantages of high efficiency,accuracy and low cost.Especially in the production line detection of repetitive work requirements,machine vision is not only accurate,but also fast detection speed,system stability.In view of the traditional length measurement method can not detect the pitch,this study proposes an on-line dynamic detection method based on machine vision for the pitch of optical cable,and proposes the application of depth learning to the pitch length measurement of optical cable.In this method,machine vision technology is used to dynamically collect the surface information of optical cable on-line.The new detection system uses Doppler principle to avoid the abnormal acquisition caused by the uneven linear speed,and sets reasonable lighting scheme and acquisition scheme^[2]In addition,this study proposes the application of depth learning method to achieve the purpose of pitch length detection,introduces the morphological adaptive modeling template matching method,and compares the depth learning method with it.The deep learning method optimizes the multi-scale application of yolov3 by using the yolov3 algorithm flexibly after the analysis of pitch visual information,so that the scale is more targeted;and according to the symmetrical features of visual information,the self-organizing network is used to get the final detection results,and then the application theory of deep learning measurement is proposed.Compared with the traditional image processing method and morphological adaptive modeling template matching method,the new method is better.Through theoretical analysis and experimental verification,the error between the measured pitch results and the standard pitch results is within the scope of industrial demand,which meets the needs of the project,and the operation is stable and reliable,providing a new way for detecting the pitch of optical cable.