摘要
为克服时频关系为线性的基函数的不足 ,提出了一种新的时频信号表示—— Dopplerlet变换。对其物理机制的分析表明 :该变换实质上是对信号的时频能量分布进行非线性分割 ,而以线性分割为特征的 Fourier变换、短时 Fourier变换 (包括 Gabor变换 )、小波变换、chirplet变换则均是Dopplerlet变换在其参数取特定值时的特例。采用自适应匹配投影分解法搜索出一组与信号分量最佳匹配的 Doppler-let基函数 ,据此可用尽可能少的波形重构原信号。文中还提出了时间和频率分辨率均达到理论值极限的“伪时频分布”的概念。
A new time frequency signal representation called Dopplerlet transform is presented to overcome the deficiency of basis functions with linear time frequency relations. Physical analysis and theoretical predictions indicate that the proposed transform essentially performs a nonlinear segmentation of the time frequency energy distribution, as opposed to many other transforms characterized by linear segmentation such as the Fourier transform, short time Fourier transform (including the Gabor transform), wavelet transform and chirplet transform which are all special cases of the Dopplerlet transform with specific parameters. Matching pursuits are used to decompose the signal into Dopplerlet basis functions which best match the signal components, so the signal can be reconstructed using as few waveforms as possible. A new concept, the “pseudo time frequency distribution” is proposed which resolutes both time and frequency domains to their theoretical limits.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2000年第3期55-58,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金!69775 0 0 9
国防科技重点实验室基金!97JS3 4.7.1JW0 10 5
关键词
时频分析
Dopplerlet变换
基函数
信号分析
time frequency analysis
Dopplerlet transform
matching pursuits
pseudo time frequency distribution
