Infrared Thermography is an emerging technology for nondestructive testing. However, it is subject to blurring and degradation of the acquired signal, as the diffusive nature in the process. This make difficulties for qualitative and quantitative analysis, especially when deeper defects located within the substrate, as well as high thermal conductivity materials are inspected. In this article, we present the method of thermogram processing to eliminate the effect caused by inhomogeneous heating and surface emissivity variations. The method based on differential thermal contrast, recurring to thermo graphic signal reconstruction technique, the “whole” temperature evolution can be fitted instead of a few of frames, subsequently, time-derivatives processing is implemented for characteristic time definition, which play an important role in the modified thermal constrast method. The result is a noise-reduced image where the subsurface defects are highlighted. In this way, quantitative parameters, e.g. the area and the depth of the subsurface defects can be extracted easily with precision. In conclusion, experiments with three types of materials validate the theoretical prediction.