Research and improvement of denoising method based on K-neighbors
Yong Guan · Journal of Computer Applications · 2009
An improved method for denoising the point clouds with noises and outliers acquired by a 3D scanner was presented.The method established the topology connection of the scattered points by searching the K-neighbors of each point.The Gaussian function was used as a kernel function to estimate the current point's effect on its neighbors,so the noises could be restricted and the outliers could be removed.The concept of density entropy and how to optimize the parameter of Gaussian function are the emphases.The method solves the problem of window-width's uncertainty in application.The results of emulation experiments show that the method can detect outliers easily,and it avoids mistaking points on the model as outliers.