Defect Detection Based On Infrared Thermal Image Research

With the progress of the infrared thermal imager manufacturing technology and the rapid development of computer vision technology,based on the defects of infrared thermal image detection technology has become the material or equipment faults and defects early accurate quickly identify hot spots of research method,the in-depth study,will certainly to safe and reliable operation of the key major mechanical equipment has important significance and practical value.Many camera calibration methods,this paper studies the analysis of their advantages and disadvantages,combining with the existing experimental equipment,through the theory combining with experiment,on the basis of the identification accuracy meet the defect size and location,finally puts forward the suitable for the calibration model of infrared thermal imager.At the same time,aimed at the defects of infrared thermal image edge detection problem,further study of the edge extraction method,the variety of the existing combined with experiments of infrared thermal images,using a variety of edge detection algorithm simulation found that most of these algorithms cannot detect the material defect or detection efficiency is too low,is not conducive to the practical application,finally using Canny edge detection algorithm combined with the local cutting,can detect the defect edge is more accurate and efficient lowlands.In order to solve the problem of defect depth is estimated accurately,the BP neural network to carry on the design and testing,through many times to change the transfer function and training function and combined with the experimental data,within the allowed error range board the depth of the defect prediction is realized.Based on the in-depth research,analysis,and on the basis of summarizing the existing related algorithm,make full use of existing experimental equipment for experiments,completed the interior parameters of the infrared thermal imager calibration,the defects of infrared thermal image edge detection,defect depth prediction of sheet metal.Solved by using infrared thermal image technique to the defect of materials for plane position and the size of defect detection and quantitative detection problem of defect depth information,and can achieve good results,for the defect detection technology based on infrared thermal images useful contributions to the development and wide application.