Dealing with Noisy Data on Point Cloud Models

Most of the studies working on point cloud data focused on complete and clean data (even though some of them took missing values into account), while in practice we often have to deal with incomplete and unclean data, just as there might be missing values and noise in data. We study noise handling, and we put our focus on processing a noisy point cloud of a visual object or a 3D model. We propose an approach where we first identify data points that might be noise and then lower the impact of the noisy values. To identify noise, we use supervised learning on data whose features are density and distance. To lower the impact of the noisy values, we use triangular surfaces and projection. The experimental results show the effectiveness of the proposed approach. Our contributions are as follows: First, we show how machine learning can help computer graphics. Second, we propose to use distance and density as features in learning for noise identification. Third, we propose to use triangular surfaces and projection to save execution time in noise reduction. Fourth, the proposed approach could be used to improve 3D scanning.