Inductive thermography is an emerging nondestructive testing and evaluation (NDT&E) technique and has been applied for a wide range of conductive materials. This research sets out to propose the thermal pattern separation and analysis of inductive thermography. It specify: 1) The fundamental understanding of inductive thermography and related basic NDT&E methods; 2) The principle of pattern separation algorithms and how they link to inductive thermography NDT&E; 3) The spatial-transient thermal pattern separation algorithm to automatically extract valuable spatial and time patterns to identify defects; and 4) The algorithm of spatial-transient-stage tensor mathematical model of inductive thermography system and multi-dimentional pattern decomposition algorithm for characterizing and tracking the variation of material properties. In addition, both mathematical and physical models are discussed and linked. Real testing samples such as an artificial slot, thermal fatigue natural crack and gear tooth with contact fatigue are applied to validate the proposed methods.