Face Recognition on Point Cloud with cGAN-TOP for Denoising

Junyu Liu, Jianfeng Ren, Hongliang Sun, and Xudong Jiang

Proposed, implemented and evaluated an end-to-end 3D facial point cloud de-noising and recognition framework. Proposed cGAN on Three Orthogonal Plane to effectively de-noise 3D facial point clouds and synergistically combine it with an adapted advanced recognition model. The experiment results our method improves recognition accuracy under all noise settings compared to state-of-the-art methods, with a maximum improvement of 14.81%.