Active thermography with inductive heating is a method for testing ferromagnetic parts contact-free for near surface defects like cracks or oxide films. Consequently this method can be classified as a non-destructive material testing method and because of the fast evaluation it opens up the possibility of a continuous quality control that is completely documentable. The thesis covers a feasibility study for a fully automated test bench, which inspects cast iron parts with complex geometries for defects from all sides. While the test specimen is heated inductively by one to two degrees celsius, near defects a different temperature-profile occurs, as compared to faultless parts. Several experiments served to establish appropriate components and an adequate set-up for the test bench. The test-process uses the phase image as the theoretical foundation, which is computed from image-sequences of a measurement by Fourier transform. Both gradient- and edgedetection algorithms as well as algorithms for comparing multiple images of parts with a reference image were implemented and tested for their ability to automatically identify defects. With the comparison of the acquired temperature data and an analytical defectmodel it is possible to estimate the depth of a defect. The estimated depth of defects were verified with the help of polished and etched microsections.