A new approach for estimating the depth of both diffuse and specular surlaces is presented. The method uses the 1-D local spatial phase of a 2-D image of the object acquired using structured-light illumination. The induced phase modulation of the structured-light pattern is used as a depth cue. For diffuse surfaces, the local phase is used directly to estimate the depth. For specular surfaces, fiduciary level crossings of the phase are used. The use of level crossings has been found to enhance the performance for specular targets. The spatial phase is estimated using both an analytic signal representation and a linear prediction model. The analytic signal representation requires less computation and is shown to give performance comparable to the linear-prediction method. The new approach provides high-resolution depth maps and avoids the difficulty of image registration.