This paper presents a method based on geometry for three-dimensional (3D) face reconstruction without the need for additional images, hardware components, or objects. In our proposed method, we consider part of the nose as the feature region because its shape remains almost constant during the measurement. The geometry of this region was used to provide cues for phase unwrapping. We first spatially unwrap the phase and determine the integer multiple of 2π to be added by comparing the recovered result of the feature region and its actual shape. Then, the face can be reconstructed with the acquired absolute phase. Experimental results demonstrated that our method is capable of reconstructing a dynamic face with high measurement speed, and only three phase-shifted fringes are required per frame.