一种改进的无损可逆时频联合特征提取算法  

An Lmproved Algorithm of Reversible Time Frequency Joint Feature Extraction Without Destruction

作　　者：杨征[1] 

机构地区：[1]泸州职业技术学院信息工程系,四川泸州646005

出　　处：《科技通报》2014年第6期228-230,共3页Bulletin of Science and Technology

基　　金：四川省科技厅高效智能照明管理系统(2008JY0164)

摘　　要：充分研究反演积分量对S变换控制的无损可逆性,提出基于宽度不可变高斯窗函数的无损可逆时频联合特征提取算法,对高频复杂电磁特征提取实现和精确挖掘。对目标电磁信号进行时频分析,在S变换中加入宽度可变反演积分函数窗,进行特征分流控制,根据S变换无损可逆性,结合时频分析,提取无损可逆联合分布特征。仿真实验在不同信噪比下进行特征提取,实验结果表明不同信噪比下提取的时频联合分布特征具有相对不可变性,特征分类准确性比传统方法高,健壮性强,在目标识别等领域研究价值和应用前景。The non-destruction and reversibility of inversion integral for S transform control. An improved reversible joint fea-ture extraction algorithm without destruction based on variable width for inversion integral function window was proposed. The feature extraction and accurate mining was realized for the high frequency electromagnetic signal. The time frequency analysis carried out, the variable width window was implemented in the S transform, and the extracted feature could reflect the attribution of the target. Simulation experiment was implemented in different SNR. Result shows that the time frequency jointed distribution feature which extracted by the new method has higher classification precision rate with less variability.

关 键 词：反演积分量 电磁波信号 特征提取 时频分析 

分 类 号：V249[航空宇航科学与技术—飞行器设计]