The main direction of 3D face reconstruction lies in the accurate reconstruction of the surface shape， and insufficient attention has been paid to the effect of the side face reconstruction with missing part of the face information. To solve this problem， a single-image 3D face reconstruction method was proposed based on face information completion. Firstly， the frontal face image was used to generate a three-dimensional face model， which was used to obtain a side face image， then this set of corresponding pictures was used to train the network， so that a single side face image was reconstructed finally with three-dimensional face model with complete face information. Experimental results prove that this proposed method is stable and reliable， and the generated 3D face model has the advantages of high precision and complete texture.