摘要
通过真实特征参数重建人体是虚拟现实领域的重要问题。为了实现高精度人体重建并降低计算复杂度,构建并行门控循环单元(Gated recurrent unit,GRU)网络建立人体特征参数和远场数据之间的关系,使用Fourier方法通过远场数据重建人体形状。从低维度数据向高维度数据映射会导致模型泛化能力减弱,通过并行GRU网络实现高维远场数据快速计算,有效建立低维人体特征参数和高维远场数据之间的关系。实验结果表明,该方法可以实现三维人体重建,在真实数据的数据集上具有较低的平均误差。
Reconstructing the human body using real characteristic parameters is an important issue in the field of virtual reality.In order to achieve high-precision human body reconstruction and reduce computational complexity,a parallel gated recurrent unit(GRU)network is constructed to establish the relationship between human body characteristic parameters and far-field data,and then Fourier method is used to reconstruct human body shape from far-field data.Mapping from low-dimensional data to high-dimensional data can degrade the model’s generalization ability.By using a parallel GRU network to quickly calculate high-dimensional far-field data,the relationship between low-dimensional human body characteristic parameters and high-dimensional far-field data can be effectively established.The experimental results show that this method can effectively achieve 3D human body reconstruction with low average error on real datasets.
作者
刘岩
尹伟石
LIU Yan;YIN Weishi(School of Mathematics and Statistics,Changchun University of Science and Technology,Changchun 130000,China)
出处
《黑龙江大学自然科学学报》
2025年第2期154-161,共8页
Journal of Natural Science of Heilongjiang University
基金
吉林省教育厅科学技术研究项目(JJKH20240893KJ)。
关键词
三维人体重建
并行GRU网络
远场数据
人体特征参数
3D human body reconstruction
parallel GRU network
far-field data
characteristic parameters
