Directed Point Clouds Denoising Algorithm Based on Self-learning

Traditional statistical scan cleaning methods usually make assumptions about the scanned surfaces or noise model, which requires users to manually adjust the settings. The learning-based method needs a data set for training, and the denoising effect of objects outside the data set is general. A self-learning directed point cloud denoising algorithm has been proposed. By introducing the self-learning method without pre training, this method makes denoising and gridding promote each other, and achieves good denoising effect. Our method does not require pretraining or preset parameters and has a good denoising effect on various noises.