In this paper, we propose a complete 2.5D and 3D human face acquisition framework based on a stereo sensor coupled with a structured lighting source. We aim to develop an accurate and low-cost solution dedicated to the 3D model-based face recognition techniques (FRT). In our approach, we first calibrate the stereo sensor in order to extract its optical characteristics and geometrical parameters (the offline phase). Second, epipolar geometry coupled with a projection of special structured light on a face (the online capture phase), improves the resolution of the stereo matching problem, by transforming it into a one-dimensional search problem and a sub-pixel features matching. Next, we apply our adapted and optimized dynamic programming algorithm to pairs of features which are already located in each scanline. Finally, 3D information is found by computing the intersection of optical rays coming from the pair of matched features. The final face model is produced by a pipeline of four steps: (a) Spline-based interpolation, (b) Partial models’ alignment then integration, (c) Mesh generation, and (d) Texture mapping. Furthermore, this paper presents an approach which evaluates the reconstruction techniques. We consider a scan from a laser scanner as “ground truth”, then we compute spatial deviation between it and the homologue reconstructed model, based on the well-known Iterative Closest Point matching algorithm.