Design And Development Of Magnetic Device Detection System Based On Machine Vision

With the development of modern society's information technology and industrialization,magnetic devices have been widely used in industrial production,medical rehabilitation,power electronics,defense science and technology,such as transformers,digital products,power supplies,household appliances,medical devices,etc.The quality inspection classification of magnetic devices is fast while maintaining reliability.At present,magnetic devices are basically automated in the production process,but most of them require manual detection during the quality inspection classification process,which has the problems of slow detection speed,easy to miss detection,and false detection.Therefore,this topic uses machine vision technology to start from the aspects of magnetic device product classification,automatic measurement of the backplane distance,internal parameter detection,and cloud storage,etc.,to study the automatic detection and planning methods of magnetic devices,to automate the detection process,improve detection efficiency and accuracy.Degree,has important engineering application value.First,study the magnetic device image acquisition and data set processing methods to preprocess the image.Analyze the advantages and disadvantages of median filter and mean filter when dealing with the noise of magnetic devices.An improved magnetic device image filtering algorithm is proposed to judge the noise and signal in the entire image of the magnetic device.Compared with the traditional median filtering algorithm,the proposed algorithm has a better effect.And the method of histogram equalization is used to process the image.The processed magnetic device image has high contrast and is easy to distinguish and the image distribution is more uniform.Secondly,the magnetic device edge detection algorithm and distance measurement method are studied.The edge information extracted by the Canny edge detection algorithm is discontinuous.The gradient value in the gray image is used to further improve the edge information and accurately extract the magnetic device edge contour.Experiments verify that the improved method can effectively detect the edges of magnetic devices.The measurement principle and method of geometric quantity are used to achieve accurate distance measurement.The results of manual measurement and machine vision measurement are compared and analyzed to verify the effectiveness of geometric measurement principle.Then,study the convolutional neural network model to classify the magnetic devices,use the self-built data set to test the designed neural network model,compare the changes in the loss rate and correct rate of the network model during the training process,and use the test data set After actual verification,the recognition rate is97.46%,and the average image of each magnetic device is only 0.1ms.The designed convolutional neural network model performs well in accuracy and real-time.Finally,through the design of the overall scheme of the magnetic device detection system in the enterprise,the selection of hardware and the formulation of the data transmission protocol,the measurement of the inductance,loss factor and quality factor of the magnetic device is realized,so that the machine vision and industrial instruments Quality inspection classification of magnetic devices.