摘要
头相关传输函数(HRTF)描述了人耳空间听觉特征,但其巨大的数据量是影响空间信息实时重现的主要障碍。目前的压缩方法主要是基于主成分分析(PCA)在方向维度的压缩。分析了方向维度压缩信号的稀疏性,提出在方向维度压缩后,再利用压缩感知进行时间维度的压缩方法。在重构误差仅增加0.2%的前提下,数据压缩率较PCA方法可提升11%左右。
Head - related transfer function （HRTF） is used to represent the characteristics of human spatial hearing. However, its huge amount of data is the main impediment to real time reconstruction of spatial information. The principal compo- nent analysis （PCA） is used to compresse HRTF in direction dimension. In this paper, the sparsity of the compressed signal in direction dimension is analyzed, and a novel compression method is proposed by utilizing compressive sensing theory to recompression in time dimension for the HRTF, which has been compressed in direction dimension. With reconstruction error only increased by 0.2%, the proposed algorithem can promote the compression ratio by 11% compareing with the PCA method.
出处
《电声技术》
2015年第2期41-45,50,共6页
Audio Engineering
基金
金华市科技计划重点项目(2013-1-007)
关键词
头相关传输函数
数据降维
主成分分析
压缩感知
HRTF
dimensionality reduction
principal component analysis
compressive sensing
