一种基于模糊C均值和均值滤波的点云去噪算法

Point Clouds De-noise Algorithm Based on Fuzzy C-means and Mean Filter

因点云数据中存在噪声,通常对不同特征的点云数据采用相同的处理方法,虽然能删除噪声但也会因删除尖锐特征造成过光顺。提出了一种基于模糊C均值(FCM)聚类算法且均值滤波的点云去噪算法。该算法使用模糊C均值聚类算法删除大尺度噪声后,再将均值滤波应用到点云光顺中,对数据点中的小尺度噪声进行光顺。实验结果表明,该算法去噪效果明显,在去噪光顺过程中较好地保持了边界特征,也避免了过光顺问题的产生。

Because noise exists in point cloud algorithm, the different treating methods are adopted in the algorithms. Though this method can be used to denoise, over-smoothing coursed by general de-nosing algorithm is cut off too. This paper presents a three -dimension point clouds de-noising algorithm combined with fuzzy-c-means（FCM） , mean filtedng. The large-scale noise is cut off by the improved fuzzy-c-means and the small-scale noise is smoothed by the improved mean filter. Experimental results show that this approach can be used to reserve boundary feature and effectively de-noise. Meanwhile, the problem of over-smoothing of point cloud is effectively avoided.