Research On Fault Location Of Infrared Thermal Image Detection Of Circuit Board

With the continuous progress of society and the continuous development of science and technology,various complex circuit boards are widely used in high-speed rail,maglev train and national defense equipment.These circuit boards are very precise,and different components are expensive,high maintenance cost,the use of traditional maintenance methods can no longer meet the actual needs,so it is often necessary for inspectors to use professional equipment to maintain them regularly.The infrared thermal image detection technology is applied to the detection of circuit board card,and the position of abnormal heat source is obtained by using the method of image processing.The fast location of fault components can be realized in the on-line detection of circuit board card.In this paper,the infrared image and visible light image of the circuit board are processed and the fault location is carried out in the background of the national science and technology support plan "research on the integrated system of high-speed maglev traffic engineering" sub-project "analysis of the fault location of the maglev train power grid controller board.The main contents and results are as follows:(1)Image edge detection and registration.Because there are many components in the circuit board,and the features are more complex,in order to achieve fast and effective image registration,In this paper,Canny operator is used for image preprocessing,and the gray image with clear outline characteristics of components is obtained.Combined with the method of Harris operator feature point registration,the corresponding feature points in gray image are selected,and the infrared and visible light images of circuit board with equal size and relatively overlapping position are obtained.(2)Get the edge feature of the visible image and the heat source feature of the infrared image after registration.The low-frequency subband and high-frequency subband of the circuit board image are separated by the edge preserving multi-scale image smoothing algorithm.The low frequency subband contains a large number of basic characteristic information of the circuit board,which is used as the base layer to obtain the fusion image by weighted average method.The high frequency subband contains many detailed features of the circuit board,selects the maximum value strategy,obtains the high frequency subband fusion part from the detail feature by the method of obtaining the active level graph,and proposes a deep learning model based on the convolution neural network to extract the feature map as the initial weight to solve the active level graph and realize the fusion of the detail layer.Then the detail layer image and the base layer image are weighted and fused to obtain the new image with the detailed background feature and the heat source feature in the circuit board,and it is used as the main basis for fault identification.(3)To realize the detection and positioning of circuit board fault components.Using the image difference method,the standard circuit board image obtained by bilayer subband fusion and the image to be tested are differentially processed,compared with the standard circuit board card diagram established in advance,the fault component number is obtained,and the fault component of the circuit board is detected and located.Using the image difference method,the standard circuit board image obtained by bilayer subband fusion and the image to be tested are differentially processed,compared with the standard circuit board card diagram established in advance,the fault component number is obtained,and the fault component of the circuit board is detected and located.The combination of circuit board image processing and traditional detection can simultaneously detect a wide range of circuit board components,which can accurately and quickly realize the location of fault components,can effectively ensure the safety of testing personnel,effectively reduce the workload in the face of a large number of complex components,and has good practical value in the field of engineering.