【摘要】 模型的细节层次化表示是图形学中的重要研究方向,可以加速虚拟漫游中的成像速度而不降低成像质量.目前的简化技术大都是基于模型的几何特征进行的,难以很好地处理带有纹理图像的模型.该文提出一种结合纹理特征的模型简化误差度量算子,能高效度量简化过程中与纹理图像相关的视觉信息的变化,并与关于几何特征的度量进行结合,使简化模型能更好地保持重要的视觉信息,提高简化模型的成像质量.新的度量算子能与高效的简化方法相结合,并支持保持重要语义信息的用户主导简化方法.更多还原

【Abstract】 The level of detail (LOD) construction for polygonal models is an important topic in computer graphics. It can be used to accelerate the rendering process in walking through a virtual environment without impairing the image quality. Many of the existing model simplification methods are based on the geometric features, so it’s not easy to handle the models with texture images to improve the rendering effect. This paper investigates the important information in texture image and the cause of texture distortion in simplification. Two factors are presented; the visual information is measured by color variation, and the possibility of texture distortion is estimated by the unevenness of texture intensity (the area of a triangle’s corresponding texture divided by its area in geometric space). These measurements are then combined with geometric metrics to form a new metric to guide simplification. The metric can be used with efficient simplification methods to simplify complex models, and the visual features will be preserved well in the simplified models.更多还原