Research On Industrial Online Detection Technology Based On Image Processing

The concept of "Industry 4.0" is gradually gaining popularity,and most manufacturing companies have begun the implementation phase of the project.From concept to practice:"Industry 4.0" involves three major directions:smart factory,intelligent production and smart logistics.All of them require technical support such as product identification and tracking,industrial robots,system software and big data.In terms of product identification and tracking,bar code identification technology has become a major means of product identification and tracking in modern manufacturing manufacturing control management due to its low cost,high reliability,and ease of use.In actual use,factors such as noise,light,and motion blur often interfere with the quality of the bar code image obtained by the camera,thereby affecting the accuracy of the bar code recognition.Although the current technology for identifying patterns of EAN8,EAN13,code9,code128,and QR codes is mature and widely used,the research on industrial bar codes is still rare.Under the current rapid development of the electronics industry,the electronics industry is moving toward miniaturization and high integration.Printed circuit boards are widely used in various fields in the electronics industry as a highly-informed collection of various electronic components.Due to the high density and high concentration of printed circuit boards,traditional manual visual inspections have brought great challenges.How to achieve bar code identification or pcb automatic detection in industrial production lines with precise,high-efficiency and low-cost has very important research significance.The content of this article mainly includes two parts:The first is the study of bar code identification algorithms for industrial production lines:Several common bar codes are introduced,and the structure,coding principle and decoding method of codel28 bar codes used for identification and tracking in the industry are analyzed.The noise bar code and fuzzy bar code processing algorithms commonly used in industrial production inspections are introduced.The weighted median filtering algorithm based on noise detection is applied to the denoising of noise bar codes;direction and scale of motion blurring PSF are estimated using directional differential and differential autocorrelation algorithms,and Wiener filtering algorithm is used for image restoration.According to the characteristics of the bar code image(the bar code has small gaps between the middle and the discontinuity of the granularity),the grayscale morphology erosion algorithm is used to erode the gaps in the bar code strips to make them continuous lines,which facilitates bar code recognition in late period.Finally,the bar code is identified using the one-dimensional barcode decoding algorithm provided by zbar.Experiments show that the algorithm can effectively improve the accuracy of the barcode recognition algorithm.The second is the research on pcb board defect detection technology in the production line:In order to reduce the impact of various noises produced in the production,the pcb board is first denoised;then the affine transform is used to achieve image calibration;Finally,according to the characteristics of the pcb components,different detection methods are proposed for different components.For capacitive elements with regular geometric shapes,positioning and matching methods based on circular arc elements are used;for irregularly shaped elements such as pins and diodes,an improved SSDA matching method is used.Finally,experimental verification was performed.