Geometric constraint algorithms can solve phase ambiguity for fringe projection profilometry (FPP). However, they either require multiple cameras or suffer from a small measurement depth range. To overcome these limitations, this Letter proposes an algorithm combining orthogonal fringe projection and geometric constraints. A novel, to the best of our knowledge, scheme is developed to assess the reliabilities of the potential homologous points, which works with depth segmentation to determine the final HPs. With full consideration of lens distortions, the algorithm reconstructs two 3D results from every set of patterns. Experimental results verify that it can effectively and robustly measure discontinuous objects with complex motion over a large depth range.