Carbon fiber reinforced plastic (CFRP) is recently used in a lot of fields, and efficient non-destructive inspection (NDI) techniques for the CFRPs are also required with increasing the use of CFRPs. Infrared thermography is one of a convenient NDI method which features early and non-contact inspection. However, a disadvantage of the thermographic method is that its defect detectability is not enough for the practical applications. One of techniques to improving the defect detectability is using phase images obtained by applying Fourier transformation to the temperature data. In this study, we tried to apply various image filters to further improve the defect detectability in the phase images by reducing shot noise appears in the temperature images. Analytical and experimental studies revealed that applying median or moving average filter reduce the noise in temperature-time data, and the reduction of the temperature noise also lead to improve the defect detectability in the phase images.
