Three-dimensional (3D) reconstruction based on a coded structured light method is a widely used technique to recover objects surfaces. This technique is based on projecting a pattern onto object to reconstruct and the illuminated scene is imaged by calibrated camera. The pattern creates an artificial features on the surfaces of objects and can be a correspondence points. The 3D information is reconstructed by triangulating the correspondence between projected pattern and decoded pattern in imaged scene. In this paper, 3D reconstruction method based on a coded structured light is proposed. The pattern is encoded in binary pattern and it has uniqueness in the search range to find correspondence points. The pattern is projected using an infrared laser and a diffraction optical element (DOE) for the invisible areas. Then, the illuminated scene is imaged by a sensitive camera. As a result, the disparity of correspondences between matched points draw a 3D surface information.