We review an optical method referred to as 3-D computer vision technique for nondestructive inspection of three-dimensional objects whose surfaces are specularly reflective. In the setup, a computer-generated cosinusoidal fringe pattern in the form of linear, parallel fringe lines of equal spacing is displayed on a TV monitor. The monitor is placed in front of the test object, whose specularly reflective surface behaves as a mirror. A virtual image (or mirror image) of the fringe lines is thus formed. For a planar surface, the fringe pattern of the image is undistorted. The fringe lines, however, are distorted according to the slope distribution if the surface is not flat. By digitizing the distorted fringe lines, employing a phase-shift technique, the fringe phase distribution is determined, hence enabling subsequent determination of the surface slope distribution. When applied to nondestructive flaw detection, two separate recordings of the virtual image of the fringe lines are made, one before and another after an incremental load is applied on the test object. The difference of the two phase-fringe distributions, or the phase change, represents the change in surface slope of the object due to the deformation. As a subsurface flaw also affects surface deformation, both surface and subsurface flaws are thus revealed from anomalies in the surface slope change. The method is simple, robust, and applicable in industrial environments.