Research And Development Of Machine Vision Inspection System For Code Spraying Quality Of Small Circuit Breaker

The low-voltage circuit breaker is one of the main electrical switches in the lowvoltage distribution network.It plays the role of a control circuit and a protection circuit by turning on and off the load circuit.Low voltage circuit breakers need to be coded on the circuit breaker shell during the production process.Quality problems that occur during the coding process include poor pattern definition,incomplete spraying,and deviation in direction.Therefore,defect detection of circuit breaker coding is a guarantee.Important link in product quality.The traditional manual detection method has high labor intensity,strong judgment subjectivity,and high rate of missed detection.There are great limitations.It is urgent to develop an automated detection system to achieve efficient and reliable detection of inkjet defects.Machine vision detection technology is an interdisciplinary field involving computer science,artificial intelligence and image processing.This technology is widely used in industrial automation detection.Its working principle is to first collect images of detection targets through industrial cameras,and then After preprocessing operations such as graying,filtering and denoising of the image,the feature extraction algorithm is used to extract the corner,blob,or edge features of the image,and finally the features are found to find defects in the image.Compared with manual inspection methods,machine vision inspection technology has many advantages,such as high detection accuracy,fast detection speed,stability and reliability,low cost,and easy maintenance.It is especially suitable for high-intensity repeated detection occasions such as code defect detection.In view of this,this paper proposes a coding vision defect detection scheme based on machine vision,and designs and builds a set of machine vision inspection system for automatic identification and detection of coding defects of low-voltage circuit breakers.The main research work of this thesis includes: on the basis of in-depth investigation of the current status and development trends of machine vision technology at home and abroad,according to the composition and design requirements of machine vision systems,a hardware vision-based inkjet defect detection hardware system is built;the image processing process is studied Common noise and preprocessing methods,and then compare and analyze various feature detection algorithms and their recognition effects;according to the characteristics of the coding defect image,this paper proposes a coding defect image detection solution based on the improved SIFT algorithm,which uses Harris detection algorithm It is used in the corner extraction of SIFT algorithm to remove redundant feature points to reduce the amount of calculation and shorten the calculation time.After the image is preprocessed by graying,threshold processing and Gaussian filtering,the improved SIFT is used.The feature detection algorithm performs feature detection on the image,and then uses the FLANN algorithm for feature matching.Finally,according to the matching result,the image defect position is calibrated and the detection result is output.Through performance tests,it is verified that the system has factors such as the image rotation angle and light intensity.Strong robustness against defective images Identifying a high detection efficiency and detection accuracy.