A two-step processing procedure of pulse thermography experimental data is proposed to synthesize thermal D-scans of composite materials and structures. The first step consists in the identification of the damaged areas by applying the thermographic signal reconstruction (TSR) technique and the red-green-blue (RGB) projection of a combination of polynomial coefficients. The second step takes full advantage of the correlation between the colors of the resulting TSR-RGB map and the defect depths: a limited number of thermograms (one for each color shade) is analyzed to estimate the emergence times of the contrasts locally created by the defects, which enables to assess their depths and provide a depth-scale to the thermal image, at low cost. Once validated for the quantitative imaging of calibrated defects of known depths, the procedure is applied to actual real-life damage, whether initially introduced by manufacturing processes or induced by external solicitations such as impact, fire exposition or quasi-static mechanical loading. Simple planar plates and more complex structures (stiffened panel, honeycomb sandwich and riveted assembly) are considered.