基于变分水平集方法的体素模型修复  

Voxel model repair using variational level set method

作　　者：周跃[1] 郑江滨[1] 李秀秀[1] 

机构地区：[1]西北工业大学陕西省语音与图像信息处理重点实验室,西安710129

出　　处：《计算机应用研究》2012年第5期1967-1969,1973,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(60970069);西北工业大学研究生创业种子基金资助项目(Z2011136)

摘　　要：针对体素着色方法所重建出的模型存在大量空洞与缺损的问题,提出了一种模型修复方法。首先采用改进的变分水平集方法恢复出封闭完整的体素模型曲面;再利用原始体素模型的颜色信息,对新生成的模型曲面进行重着色并滤除杂色,最终完成体素模型的修复。其中,改进的变分水平集方法在水平集能量函数中引入了符号距离函数自动规整项,避免了重初始化操作;添加了新的进化加速项,防止了隐式曲面穿透模型而造成模型侵蚀。实验表明,修复好的体素模型较原模型相比具有完整的外型和平滑的色彩,且曲面恢复效率更高。Aiming at the above problem of the existence of holes and gaps in the voxel model reconstructed via voxel coloring,this paper proposed a model repairing method.This method recovered a closing and complete surface of the voxel model using an improved variational level set method.Then it utilized the color of the original voxel model to recolor the newly generated model.After being filtered,built the ultimate voxel model finally.Within the surface recovering step,the improved variational level set methods introduced a distance regularizing term to the level set energy function,by which the re-initialization could be avoided.And it added a new accelerating term to the energy function to prevent the surface from penetrating the original model.Experiment results show that the repaired voxel model has an intact appearance with smoother color,and the surface recovering efficiency rises considerably via the improved variational level set method.

关 键 词：变分水平集方法 曲面恢复 空洞填充 体素着色 重初始化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]