On the basis of pulsed thermography, this article presents the development of a quantitative detection method for sub-surface defects. A two dimensional gapped smoothing algorithm (GSA) applied in this method to process the surface thermal distribution and no reference information is needed. A new thermal contrast is defined and a damage index based on the thermal contrast is proposed to estimate the probability of location of a defect, and differential curves of thermal contrast are applied for further quantification of defect size. One aluminium plate and one glass fibre reinforced composite laminate containing different depths of flat-bottom holes, delaminations and impurities are introduced to assess the performance of proposed method. Numerical and experimental results indicate that the influence of non-uniform heating is clearly suppressed. Compared with the differentiated absolute contrast approach, distinct enhancement of thermal contrast is achieved, and the location and size of sub-surface instances of damage can be quantitatively determined using the proposed method. The relation of damage index and flaw depth is investigated and a general expression is presented for the evaluation of flaw depth. The influence of the distribution density of the gapped grid on detection accuracy is further discussed. The proposed method, which is less dependent on the capture time, can be used for automatic detection and characterisation of defects. The reliability and applicability of the GSA applied to pulsed thermography are verified.