We introduced recently phase measurements usually performed in interferometry to the domain of image processing and intelligent vision [IEEE Trans. Instrum. Meas. 49, 867 (2000)]. Our purpose is to sense with a high accuracy the position, orientation, and displacement of two-dimensional (2D) surfaces observed by a static vision system. We report on significant improvements of the method. Experimental measurements reveal a peak-valley noise of approximately 10(-2) CCD pixel, corresponding approximately to a 10(-3) period of the phase reference pattern. Then the observation of 10 microm scaled features enables an accuracy of a few nm in the position sensing of the phase reference pattern for the extended 2D measurement range.