SIMPC: Learning Self-Induced Mirror-Point Consistency for Unsupervised Point Cloud Denoising

In point clouds, noise directly perturbs point coordinates that encode both spatial location and geometry, making one-to-one correspondence construction more challenging than in images. Existing methods impose statistical mappings across noisy variants via noise or optimal transport, but suffer from correspondence ambiguity. In this work, we propose Self-Induced Mirror-Point Consistency (SIMPC) to learn deterministic correspondences between points and the underlying surface in an unsupervised manner. For each noisy point, SIMPC generates a mirror-point on the opposite side of the underlying surface, guided by geometric priors during the denoising process. By encouraging consistency between the denoising targets of the original point and its mirror counterpart, SIMPC effectively localizes the position of underlying surface. Extensive experiments on synthetic and real-world datasets demonstrate that SIMPC significantly outperforms state-of-the-art unsupervised methods and surpasses several strong supervised counterparts.

Many technologies, such as autonomous driving, robotics, and 3D scanning, use point clouds: collections of points that describe the shape of real-world objects and scenes. However, these points are often noisy, meaning that they do not lie exactly on the true surface. Removing this noise is important, but it is difficult to train denoising models because clean, perfectly corrected 3D data are expensive or sometimes impossible to obtain. In this work, we introduce SIMPC, a method that learns to clean noisy point clouds without requiring clean training examples. The key idea is to create a ``mirror'' point for each noisy point, placed on the opposite side of the estimated surface. If the original point and its mirror point are both cleaned correctly, they should move toward the same underlying surface location. By encouraging this agreement, SIMPC learns a more reliable direction for moving noisy points back to the true surface. Experiments on synthetic and real-world 3D data show that SIMPC produces cleaner point clouds than existing unsupervised methods, making it useful for practical 3D perception and reconstruction tasks.