However, point cloud registration methods struggle to achieve precise global pose estimation, whereas previous pose-free NeRFs overlook geometric consistency in global reconstruction.

However, establishing these correspondences has been challenging due to the noisy, irregular, non-uniform, and textureless nature of 3D point clouds. Feature matching has long been the mainstream of data association without pose priors.

Our method achieves advanced partialto-partial non-rigid point cloud registration results onthe4DMatch/4DLoMatch benchmark under both no-learned and supervised settings.

Unlike AEE (defined in Eq. 6), the reconstruction task