Algorithms for efficient and quantitative non-destructive testing by pulsed thermography

This dissertation is devoted to pulsed thermography (PT). The main research topic is reconstruction of defect shape from thermographic data. The data compression problem is also considered. The reconstruction of the defect shape from pulsed thermography data is an illposed inverse problem. The existing algorithms are applicable only for ideal defect cases. In this work a new defect shape reconstruction algorithm is proposed. The developed algorithm combines the principle of the thermal pulse reflection in the onedimensional domain with the numerical simulation of the PT experiment. The algorithm can be applied for wide classes of defects. The experimental reconstruction of the corrosion profile in a steel sample by the developed algorithm is presented. In this work the problem of data compression in PT is discussed. Lossy and lossless methods are proposed. Both methods are based on pre-processing the data and application of existing compression algorithms, an image compression algorithm (JPEG2000) for the lossy method and a lossless compression algorithm (ZIP) for the lossless method. In each case the pre-processing transforms the data based on the specific information content of PT that, in consequence, increases the compression ratio. A comparison with other compression algorithms is presented