Quantitative Evaluation Of Defects In Metal Based On Infrared Thermal Image

Aiming at the metal surface and subsurface defects detection, quantitative evaluation ofinfrared thermal imaging detection is carried out to detect the hole defects of aluminum plate.The temperature variation of the cooling process of surface defects with different sizeand different depth is studied by finite element analysis. The results show that: the surfacetemperature of defect is higher than the normal position; when in same depth, the larger thesize of the defects, the faster the heat diffusion, the wider the diffusion range, the temperatureis inversely proportional to the defect size. In the condition of the same size, the larger thedepth of the defects, the slower the heat diffusion, the temperature is proportional to the defectdepth.The experimental scheme is designed based on the analysis of simulation. The specimenis heated on single surface in active infrared detection technology, The temperature in coolingprocess are collected by infrared camera. The results show that: for the defects of same size,the temperature (or gray level) is proportional to the depth; for the defects of same depth, thetemperature (or gray level) is inversely proportional to the size, which is consistent with thesimulation; In the deeper subsurface region C (hole depth7mm) and D (hole depth9mm)defects are easy to be detected because the temperature difference between defect and normalregion is large; In the shallow surface region A (hole depth3mm) and B (hole depth5mm)defects are hard to be detected because the temperature difference between defect and normalregion is small.In order to solve thermal image aliasing problems between surface(or subsurface) defectand normal region, a method to enhance the image characteristics of genetic algorithm andsequential image based on weighted superposition is proposed, the gray level of defect andnormal region were increased by8.5%and31.0%, so that the defect characteristics can begreatly enhanced; Especially to the subsurface defects (minimum defects with2.5mm of thediameter, hole depth3mm), the characteristics is enhanced more significant.The enhanced image are processed by the contrast enhancement, median filtering andthreshold segmentation, defects are processed by edge detection and edge extraction, defectlocation and shape can be qualitative detected; the area of the defect are estimated through thepixel area calculation in defects area, for the defects in A (hole depth3mm, the diameter are2.5mm、4.2mm、6mm and8mm) region, the average error is about9.64%; for the defects in B(hole depth5mm, the diameter are2.5mm、4.2mm、6mm and8mm) region, the average erroris about6.13%; for the defects in C (hole depth7mm, the diameter are2.5mm、4.2mm、6mmand8mm) region, the average error is about7.44%; for the defects in D (hole depth9mm, the diameter are2.5mm、4.2mm、6mm and8mm) region, the average error is about8.88%.