Research On Circuit Board Fault Detection Technology Based On Infrared Image

For its safe,efficient and other advantages circuit board fault detection technology based on infrared image is applied to circuit board fault detection technology.Comparing and analyzing the infrared image of circuit board under normal and fault working conditions to identify the faults on the circuit board.In this thesis,infrared image preprocessing,infrared image registration,fault target recognition and other directions are studied in the field of circuit board fault detection technology.Due to the limitations of the infrared image capture conditions,the infrared images has more noise than others,and its features are not obvious.First of all,the infrared image should be preprocessed.The image noise can be filtered out more effectively and the image eigenvalue can be extracted by improving the wavelet threshold de-noising function.For better identify the faulty components in the infrared images,there is a registration work between the standard board image and the test board image which have spatial location difference.A new fast registration algorithm based on ring descriptor was proposed in this study.Introducing the rapid detection feature extraction and creating a ring descriptor to increase speed.The improved algorithm of RANSAC was used to eliminate the mismatch points.The experimental results of three groups of image registration simulation show that compared with the traditional SIFT and SURF algorithms,the registration accuracy and operation time have improved.Three methods were used to identify and analyze the determination of faulty components.The difference diagram of the board fault obtained by using the differential infrared image method can identify the faulty components with obvious temperature changes;To determine the faulty components to compare the standard and the fault element temperature curve by using the image set temperature curve method;The improved optical flow method of weighted function was used to obtain the target optical flow field,and the motion vector of the fault target was calculated to identify and locate the faulty components.The identification and localization of the faulty components were realized by combining three discriminant methods.Circuit board fault detection technology had been improved due to improvements in research,and it is more accurately to identify the faulty components.The data and experimental results showed that the methods in the thesis had effectively improved and optimized the batch detection technology of circuit board faults in aerospace companies,and reduced the false detection rate of the fault detection at the same time.