基于DCT的实值离散Gabor变换最优窗宽选择.PDF

第24卷 第10期 计 算机 技 术 与发 展 Vol.24 No.10 2014年10月 COMPUTERTECHNOLOGY AND DEVELOPMENT Oct. 2014 基于DCT 的实值离散Gabor变换最优窗宽选择 汪 琦,陶 亮 (安徽大学计算智能与信号处理教育部重点实验室,安徽 合肥 230039) 摘 要:文中针对Gabor变换中最优窗函数宽度选择的问题,提出了以提高Gabor表示的聚集性和时频分辨率为目的的基 于DCT核的实值离散Gabor变换最优窗函数宽度选择算法。 对香农熵的取值范围进行了研究,使其更适合度量时频分布 的聚集性,进而根据熵度量实现了与信号非平稳性相适应的最优窗函数宽度的选择。 实验结果表明,该算法对单分量及 多分量信号都能有效地选择最优窗函数宽度,同时能够获得聚集性好、时频分辨率高的Gabor表示,并且具有很好的抗噪 能力。 关键词:实值离散Gabor变换;窗函数;离散余弦变换;信号处理 中图分类号:TN911.72 文献标识码:A 文章编号:1673-629X(2014)10-0055-03 doi:10.3969/j.issn.1673-629X.2014.10.013 Optimal Window Width Selection for Real-valued Discrete Gabor Transform Based on DCT WANG Qi,TAO Liang (MOE Key Lab of Intelligent Computing Signal Processing,Anhui University,Hefei230039,China) Abstract:Aiming at theproblemofoptimalwindowwidth selectioninGabortransform,toimprovetheconcentrationandtime-frequency resolution of Gabor representation,an algorithm for the optimal window width selection of the DCT-kernel-based real-valued discrete Gabor transformsisproposed.The value range of Shannon entropy is studied to be more adequate for measuring concentration of time- frequency distribution.Moreover,based on the entropy,an optimal window width adapted to signal non-stationary can be searched.Ex- perimental resultsshowthattheproposedalgorithmgivesthebest Gaborrepresentationwithhigh concentrationandtime-frequencyreso- lution for mono-component signal or multi-component signal.Additionally,the proposed algorithm also performswell under low signal noise ratio. Key words:real-valued discrete Gabor transform;window function;DCT;signal processing 0 引 言 bor变换时