Abstract:Aiming at the problem that the current point cloud denoising algorithm is easy to ignore the edge features, in order to retain the significant structural features of the point cloud and improve the accuracy of point cloud denoising, a denoising algorithm based on low-dimensional manifolds is proposed. Assumes that the point cloud is distributed on a low-dimensional manifold in a high-dimensional space, and uses the self-similarity of the surface between the point clouds to establish a Turplacian model to approximate the manifold structure, combined with the regularization constraints of the graph to achieve the accuracy of the point cloud Denoise, and finally evaluate the algorithm quantitatively by calculating the mean square error. Experimental results show that the point cloud denoising algorithm proposed has smaller errors and can better retain the visually significant structural features.