Abstract: A model based on total curvature and variational level set methods is proposed for surface denoising.It is the natural analogue of total variation based on ROF model for image denoising in geometry processing.The model based on the implicit variational level set methods can handle topological changes automatically.By evolving the gradient decent equation，the optimal solution is obtained.The half-point differencing method is used to get more accurate results.Experimental results show that the model removes noise efficiently and preserves salient features of the surface.

Key words: surface denoising, total curvature, variational level set, half-point differencing

摘要： 针对曲面去噪问题，提出了一种基于整体曲率变分水平集方法的曲面去噪模型。该模型是应用于图像去噪的ROF模型在几何形状处理中的自然拓展。它基于隐式水平集变分方法，能够自动处理曲面拓扑结构的变化。对该能量模型采用变分水平集方法求其梯度最速下降方程，通过演化该方程，最终得到模型最优解。为使计算结果更加准确，采用了半点差分格式离散。实验结果表明，该模型具有良好的去噪性能，同时能有效地保持曲面中的特征信息。

关键词: 曲面去噪, 整体曲率, 变分水平集, 半点差分