One criticism which can be leveled at thermal wave images is that their resolution is often less than that of the very best ultrasonic images of similar targets. This reduction of the resolution arises from the transverse diffusion of heat in the thermal waves reflected from the subsurface defects in the sample. In this paper we describe a technique for removing the blurring of pulsed thermal wave images of planar defects through the reconstruction of the shape of the scatterer by use of inverse scattering techniques. Although the method at present is restricted to planar defects, this special class of defects includes delaminations and disbonds in layered materials, defects which are of great interest to a variety of industries. Therefore, the availability of a reconstruction algorithm provides a solution to an important problem in nondestructive evaluation. The algorithm produced we have developed is quite simple and very effective when it is applied to thermal wave images of such defects. The idea behind the design of the model is the manipulation of the equations of thermal wave scattering theory in such a way that the scattered wave at the surface of the sample ends up being expressed as a convolution of a “heat spread” function, with a function which describes the shape of the scatterer. The Fourier transform of the surface temperature contrast (the contrast in the image is essentially just a representation of the scattered wave) can then be expressed as a simple product of the Fourier transform of the heat spread function and the Fourier transform of the shape function of the scatterer. In principle, application of the algorithm consists of performing a two-dimensional spatial Fast Fourier Transform (FFT) on the experimental image of the (unknown) scatterer, dividing the resulting Fourier transform by the transform of the (known) theoretical heat spread function, and finally doing an inverse FFT to obtain the shape of the scatterer.