由于顶点的曲率能够很好地反映3维网格的视觉特征,该文提出了一种利用曲率分析的3维网格质量评估方法。该方法首先估算各顶点的曲率,然后在每个顶点的邻域内构建一个曲率矩阵,并根据原始网格和失真网格对应曲率矩阵的奇异值差异评估顶点位置的失真,最后通过对这些局部失真的加权联合得到网格的整体失真。实验结果表明,相比于其他方法,该文提出的方法不但能够更准确地评估3维网格的质量,而且具有更好的鲁棒性。
In this paper, a novel metric is proposed to evaluate the 3-D mesh quality via analyzing curvature, since the curvature describes well the visual characteristics of a 3-D mesh. Firstly, the curvature at each vertex is estimated, then a curvature matrix is constructed in the neighbourhood of each vertex, and the local distortion at each vertex is measured in terms of the differences between the singular values of the curvature matrix in the original mesh and that of the corresponding matrix in the distorted mesh. Finally, the global distortion is obtained by weighted combination of the local distortions. Experimental results reveal that the proposed metric not only achieves superior performance in prediction accuracy over all the other competing metrics, but also has very good robustness and stability.