Range scanners based on camera-projector configurations have been attracting the attention of many researchers and computer vision practitioners due to its ability of performing 3D scanning with high accuracy and frame rates. This class of scanners is more effective under dark capturing environments by projecting bright colored patterns on objects. As a consequence, the projected pattern is prone to interfere with the reflectance and texture characteristics of the surfaces of the objects. This paper introduces a novel technique for recovering texture information from objects obfuscated by projected patterns. To that end, prior to 3D scanning our method acquires a small amount of pairs of images of the object with and without the projected pattern. These pairs of images are then used to build a dictionary that establishes a relationship between the actual textures and the projected patterns. Following that, 3D scanning is performed normally using a camera-projector setup, and the video frames captured by the camera are matched against the dictionary to filter candidates that resemble the original textures. Regularization is applied on the potential set of candidates to cull out poorly detected one, resulting in smooth texture-aware images.