The damaged faces of statues are prone to getting redamaged during the restoration process, and the historical image data of statues are insufficient. To solve these problems, we propose a digital restoration method based on a deep learning network, which can use a single statue face image to restore the damaged face of the statue. First, the deep learning network is used to process the single image data of the undamaged statue to generate a point cloud. Second, the damaged statue point cloud is obtained through a laser scanner. Finally, the two data are registered and fused to generate a complete statue model to realize the digital restoration of the statue face. In this study, the artificially damaged point cloud of undamaged facial statues is used to simulate the damaged facial statues, and the proposed method is used for restoration. The experimental results show that the proposed method can effectively realize the digital restoration of facial damaged statues, and the generated statue model has correct facial details and high restoration accuracy.