Spatio-temporal filtering of active thermography data for noise reduction and data compression

An algorithm was developed to process thermographic image sequences recorded after pulsed excitation in order to achieve noise reduction of data and at the same time a reduction of the necessary storage space. In contrast to existing methods like TSR, the algorithm does work on adjacent pixels both in space and time. The filter parameters are controlled by a physical model, based on the thermal broadening behavior of an instantaneous point source. Input parameters are the thermal diffusivity of the material and the maximum depth of observation. The results show efficient data smoothing, in particular at late times, where weak contrasts have to be evaluated. The data volume is compressed by 90%.