Infrared Thermal Image Recognition Method For Typical Circuit Board Faults

In the electronics industry,with the development of manufacturing processes,the integration of electronic components on circuit boards has become higher and higher,and circuits have become more and more complex.When it fails,it will inevitably require a lot of time and energy to use the traditional contact diagnosis method.Therefore,non-contact fault diagnosis methods have become an urgent need in the industry.As an emerging non-intrusive fault diagnosis technology for circuit boards,the diagnosis method based on image recognition is not only favored by maintenance personnel,but also attracted more and more attention from related research and technology developers.Based on the in-depth study of the principle of infrared thermal image recognition,this paper proposes a method to identify typical faults of the circuit board based on the infrared thermal imager.According to the current research status of circuit board inspection methods at home and abroad,comprehensive domestic and foreign research on image processing algorithms,including image denoising,image registration and other algorithms,analyzes the current mainstream wavelet threshold denoising algorithm,SIFT algorithm,HS optical flow method,temperature change curve method and other deficiencies,provide a theoretical basis for optimizing image processing methods.The infrared image of the circuit board is taken by an infrared thermal imager for experimental research.Three pairs of infrared images of the circuit board are selected as the experimental data,and methods are compared and improved in each link of image processing.In the process of denoising and enhancing infrared images,due to the poor effect of wavelet threshold algorithm in denoising infrared images,an improved wavelet threshold algorithm is proposed.Aiming at the defects of threshold and threshold function,a threshold with adaptive characteristics is proposed.As well as the threshold function,the test results show that the improved wavelet threshold algorithm can better apply to infrared images.In the process of image registration of the denoised image,due to the lack of accuracy and speed of the current SIFT algorithm,an improved SIFT algorithm is proposed to focus on the feature point extraction,feature point description,and feature point matching in the SIFT algorithm.Optimized,the extraction area of feature points was changed to a circular area;the descriptor of feature points was changed to a circular descriptor to reduce the dimension of the descriptor;the concept of hierarchical threshold was introduced in the feature point matching.In addition,the threshold value increases successively;and the improved algorithm is tested,and the results show that the optimization of the SIFT algorithm effectively improves the accuracy of image matching and improves the running speed of the algorithm.In the fault component location and fault type identification of the matched image,because the single optical flow method cannot judge the fault type,the single temperature curve method is cumbersome,and the method of combining the optical flow method with the temperature curve is proposed.The test results of this method show that the method can accurately identify the faulty component,and can effectively judge the fault type by drawing the temperature change curve of the faulty component.A series of experimental tests in the whole process show that based on the infrared thermal imaging diagnosis theory and method proposed in this paper,faulty components can be located quickly and accurately,and fault types can be diagnosed at the same time,which can be applied to the fault diagnosis of actual circuit boards.