Enhanced geometric constraint-based phase unwrapping algorithm in binocular stereo vision fringe projection system

Abstract

Phase unwrapping plays an important and central role in phase-based digital fringe projection profilometry. The unwrapping quality directly influences the three-dimensional measurement accuracy. Recently, an effective geometric constraint-based phase unwrapping algorithm has been proposed to obtain the continuous absolute phase map and the unwrapped phase accuracy was found to be high. However, in this technique the virtual depth plane at z = zmin is often created empirically, which increases the manual measurement error. For this reason, this paper proposes a method for accurately constructing the virtual plane and further applies it to phase unwrapping of objects with a larger depth range. In this method, a binocular stereo vision system is used as the measurement set-up for the virtual depth plane construction and a series of virtual depth planes at z = zimin (i ≥ 2) is automatically built using a computational framework. Then, the phase is unwrapped for each region according to the continuity of the unwrapped phase and a complete absolute phase map is obtained by merging the unwrapped phases in all regions for 3D reconstruction. In this process, the virtual depth planes are created automatically and quantitatively by the measurement system. No human intervention is required and it greatly reduces the manual measurement error. Experiments show that the artificial virtual planes can be built accurately and the phase is unwrapped correctly and readily.