We find a practical solution to improve remote 3D measurement accuracy by investigating surface effects on a real-virtual object alignment task in a stereoscopic augmented reality environment. A psychophysical method of adjustment was used for aligning a computer-generated graphic virtual pointer with the designated targets on the image of a real hemisphere surface. The known conflict between binocular fusion and object occlusion cues was confirmed to play a major role as an extra depth cue. Surface texture density had significant impact on real-virtual object alignment task. A model was proposed for the process of detecting the interactions of real-virtual objects and guidelines were also given for augmented reality interface design accordingly.