基于结构相似的三维模型视觉降质度量

Evaluation of visual perception degradation for 3D mesh based on structural similarity

摘要:

针对三维网格简化操作经常会引起一定的视觉降质和失真的问题,为了更准确评估这种视觉降质,利用三维网格显著特性信息作为网格顶点的属性,采用包围盒空间剖分的方法描述三维局部窗口,基于结构相似性和信息熵以定量的方式给出了三维网格模型的视觉感知降质的客观度量方法.理论分析和实验结果表明:此度量能非常有效地评估网格简化过程中引入的视觉降质,可用于同源不同分辨率的三维模型之间的相似性度量,并提供了多尺度特性的视觉降质分析.

Abstract:

Aimed at the problem that some visual perceptual degradation and distortion are usually generated during 3D mesh simplification process,we take the salient information as the property of the vertex and partition the bounding box of the mesh into some 3D local window.Then the mean perceptual structural degradation(MPSD) metric based on salient information and information theory is presented and used as a quantitative and objective metric to evaluate the visual perceptual degradation.Both analytical and experimental results show that the MPSD metric can effectively evaluate the visual perceptual structural degradation induced by mesh simplification operations and provide the functionality of multi-scale visual perceptual degradation.