The aim of this work is the investigation of cyclically loaded specimens. Due to cyclic loading fatigue, cracks initiate on the surface of the specimens. To get better understanding of crack behaviour a new routine for examining these cracks with digital imaging processing was developed. The routine was implemented into MATLAB. This program is able to perform automatic crack detection and calculation of crack parameters, like crack length, crack width, distance between neighboured cracks and opening angle of the crack. Within the tests evaluated in this work, a correlation between crack parameters and fatigue life of the specimens was found. Furthermore another routine was developed, that allows the calculation of parameters for material models. These material models are used in FE-simulation software for modelling of plastic material behaviour. Those methods have been developed to reduce user based influence and to ensure objectivity. When looking at TMF (Thermo-Mechanical Fatigue) test data, failure occurs at very low stress amplitudes and high load cycles. These very low stress amplitudes cannot be implemented in lifetime calculations of mechanical engineering components, as these components will be loaded at higher stress levels. This method allows the determination of realistic stress data for lifetime prediction. Micrographs of tested TMF loaded specimens were taken by light image microscopy. The developed routine for image processing is carried out on these micrographs. Visible cracks on the surface are identified and crack parameters are calculated. These parameters can be related to the lifetime of tested specimens, which allows better understanding of crack behaviour with increasing number of cycles. For full implementation of TMF test data into FE-software, temperature dependent parameters are needed. Plastic material data of recorded hysteresis loops from tests at distinct temperatures were taken. The data were fitted to obtain material parameters for implementation of the material model. The material model allows the simulation of the specimen with various loads and temperatures. The presented methods have been applied on two different test series with different materials. Due to satisfying results these routines are now implemented into the standard evaluation process of TMF-data.