Active thermography has been emerging as a reliable non-destructive testing method due to its whole field, non-contact and non-invasive evaluation capability. It uses the acquired history of surface infrared emission over the object and subsequently obtains subsurface details. But faint contrast detail accompanied with spurious data generated in experimentation leads to false interpretations with raw data analysis. A suitable processing, which enhances the signal to noise ratio and un-correlate noise by separating the noise from other features facilitates enhanced detail extraction. Either pulse compression or feature separation methods accomplish this task and improve the potential defect detection performance. In this paper introduces various feature separation methods and investigates their enhanced defect detection performance than existing conventional processing methods in Infrared imaging for non- stationary thermal wav e imaging methods. a correction has been made to the theory of thermal waves generated with Frequency modulated thermal wave imaging (FMTWI) and in the next stage various processing methods have been tested for captured thermographic data generated during FMTWI with a goal of exploring fine subsurface details. Finally features of various processing methods have been compared from defect visualization perspective.