GRAPH-BASED POINT CLOUD COLOR DENOISING WITH 3-DIMENSIONAL PATCH-BASED SIMILARITY

Point clouds are utilized in many 3-D applications such as cross-reality (XR) and realistic 3-D displays. They are a set of points having 3-D coordinates and associated color signals. These color signals are often perturbed by noise induced by the measurement errors of scanning devices. In this paper, we propose a point cloud denoising method for color signals. Since many conventional methods for point cloud color denoising are based on a low-pass filter in the graph spectral domain, denoising accuracy is affected by the choice of the graph construction method. We propose a graph construction method using 3-D patch-based similarity, in which the similarity is calculated with small 3-D patches around the connected points. This is in contrast with conventional graph construction methods for denoising which are based on point properties, such as coordinate and color signals. Second, we propose a method to select a low-pass filter with preferable frequency response automatically with an estimated noise level. Our experimental results with 3-D human point clouds showed that the proposed method, called 3-D patch-based similarity (3DPBS), achieved the best denoising accuracy compared with some state-of-the-art methods.