Standard approaches for the co-registration of terrestrial laser scans (TLS) and close range digital images (CRDI) taken separately require often artificial targets. These approaches are reliable but not efficient for larger projects. Our approach applies scale invariant feature detection methods to make the co-registration process more flexible. But the reliability of feature matching based on the design of feature descriptors is sometimes questionable. The accuracy of the applied algorithm can be improved by introducing some additional geometric constraints. Our approach consists of a three-step procedure. In the first step scale invariant feature detection in the brightness image from the digital camera and the corresponding intensity image from the terrestrial laser scanner is carried out. In the next step, the initial matching values of the corresponding points are corrected by introducing additional constraints. Finally, from each set of match, the affine transformation parameters are calculated so that the 3D point cloud and brightness image can be registered together.