Research On Image Enhancement Method In Pulsed Eddy Current Thermal Imaging Technology

Pulse eddy current thermal thermography detection technology has grown rapidly into an important branch of non-destructive testing with its advantages such as high efficiency and fastness,no contact and conveniently testing.It has also been extensively studied by researchers at home and abroad and developed many Basic image processing technology.From the initial manual observation to determine the defect,to the subsequent introduction of a series of defect recognition algorithms based on the time and space domains,although the defect identification methods can be completed at present,the detection models constructed by these detection algorithms often cannot be related with pulse eddy current thermal imaging.The detection principle of the technology is related to the imaging mechanism,so it is not reliable.Therefore,building a more reasonable infrared image enhancement algorithm model and improving the accuracy of defect detection are of great significance for defect detection using pulsed eddy current thermal imaging technology.In this paper,through the exploration and analysis of pulse eddy current thermal imaging detection technology,defect feature analysis,algorithm model construction and experimental data verification are performed both in the time and space domains respectively.It is proved that the method proposed in this paper can build an algorithm model based on the detection principle.And effectively enhance the infrared image data,improve the accuracy of defect detection.The main contents of this paper are as follows:(1)Theoretical research and analysis.Firstly,this article analyzes the principle of pulsed eddy current thermal imaging detection technology and the imaging processing principle of infrared images,and then uses the laboratory detection platform as an example to explain the composition of the thermal imaging detection system;then the data collected by the laboratory platform based on the analysis of the data features in the spatial domain and the data features in the time domain.(2)Research on image enhancement methods based on spatial domain features.The research of infrared image enhancement algorithm based on weighted estimation algorithm is proposed firstly.In order to eliminate the impact of high frequency noise in the image,the step size is set when the template points are selected,similar to the sampling method so that the influence of high-frequency noise is avoided;at the same time,the temperature difference between the template and the target point is used as the basis for calculating the weighting coefficient,which eliminates the interference of the heating coil on the infrared image and improves the contrast between the defect area and the background area;The infrared image enhancement method of the improved GAN algorithm performs PCA and uniform sampling processing on the image data to eliminate noise in the thermal time data.According to the process of generating thermal data in the infrared image,more training data is obtained through neighborhood data fusion data.Besides,a new loss function is designed to improve the generalization performance of the GAN,and the defect contours in the optimized image are significantly enhanced.(3)Research on image enhancement methods based on time domain features.The physical information contained in the thermal image time series is related to the distribution of eddy currents inside the specimen,and a unique path is reversed from the infrared image sequence to the eddy current distribution in the specimen,and the eddy current distribution inside the specimen is derived by theoretical calculation.Iterative algorithm can be used to effectively and intuitively observe the distribution of eddy current in the test piece,so as to find the position and contour of the defect.