Robust Normal Estimation on Unorganized Point Clouds

Provided by: Binary Information Press
Topic: Cloud
Format: PDF
Normal estimation on unorganized point clouds is one of fundamental problems in point-based graphics. While accurate and high-quality point-based rendering heavily depends on the normal property, many recent surface reconstruction algorithms also require normal information to obtain faithful reconstructions. This paper aims to present a robust method to estimate normal on unorganized point clouds. The normal vectors at points are evaluated by a new robust estimator-Maximum Kernel Density Estimator (MKDE) based on the nonparametric kernel density estimation technique. Therefore, the algorithm is highly robust with respect to noise and outliers.

Find By Topic