Infrared Thermography has been successfully used as an experimental, non-destructive, real-time and non-contact technique both to perform non-destructive evaluations and to study the fatigue behaviour of materials. However, the temperature is a very sensitive parameter to the environment conditions such as the thermal heat exchanges. It follows that, heavy and time expensive algorithms have to be setup to accurately filter out the overall ‘disturbing’ heat sources. Otherwise, an in-depth analysis of thermal signal allows the assessment of different indexes related to physical processes of fatigue damage and failure. Theoretical and experimental framework becomes complicated in case the material is a composite due to the layups of lamina or due to the viscous properties of the bulk of the matrix, or due to the pattern described by the yarns or fibers making difficult any quantitative and qualitative analysis. In fact, anisotropy and heterogeneity of composites influences unavoidably the mechanical response of the material to external excitation and the failure mechanisms. In all the cases, the study of thermal heat sources related to dissipative phenomena becomes complicated. Thermal signal analysis provides a localised analysis for assessing qualitatively and quantitatively the state of degradation of material in terms of stiffness or in term of damage detection, by extracting temperature components related to the appearance of plastic zones or cracks or in general to dissipative heat sources. The focus of the present research is to provide an innovative method and algorithm for processing the signal from innovative composites obtained by Automated Fiber Placement process in order to assess the fatigue behaviour and damaged regions qualitatively and quantitatively.