Abstract: Triangular mesh denoising is one of the classical problems in computer graphics.Various methods of mesh denoising have emerged in recent years.This paper focuses on the introduction of state-of-the-art methods of feature-preserving mesh denoising,and summarizes the geometric representation of the triangular meshes and a series of feature structures.The existing denoising techniques have been divided into three types:optimization-,filtering-,and data-driven-based methods,and they are analyzed and discussed in detail according to different denoising models and utilized mesh attributes.Four common evaluation criteria are described to measure the denoising quality.Meanwhile,the advantages and disadvantages of mesh denoising algorithms are presented through comparisons from the five aspects of capability:sharp feature preservation,volume preservation,outlier removal,vertex drift phenomenon,facet flipping.Finally,possible directions of future work are pointed out to solve the common problems of these approaches.