[blueocean]

----------------------------------------------------------------------------------------------------------
本主题是专题讨论"时频分析及其在地震勘探中的应用"的一部分，欢迎访问该专题讨论的主贴：
http://forum.petro-c...?showtopic=4585
----------------------------------------------------------------------------------------------------------


In 1996, Stockwell published one paper about s-transform. S-transform can be considered as pseudo-continous wavelet transform. The normal CWT is transformed to (time, scale) domain, while S-transform is transformed to (time, frequency) domain. On the other hand, s-transform is very similar to Short Window Fourier Transform. The difference is the frequency dependent window length. The kernel basis of s-transform is scaled and frequency dependent Morlet wavelet.

Here is the Stockwell's webiste:
http://www.cora.nwra.com/~stockwel/

There're a bunch of information about s-transform.

The first s-tranform paper is:
http://www.cora.nwra...cs/loc_ieee.pdf

The matlab codes are:

http://www.cora.nwra...rm/m_files/st.m

http://www.cora.nwra...m_files/stest.m

Contribution by Robert G. Stockwell

[blueocean]

附加文件

•  loc_ieee.pdf (951.7K)
  下载次数: 35

[juliaxj]

谢谢楼主！！
真的非常感谢，我正需要这方面的资料呢。

[juliaxj]

:(
这个程序我已经有了，有C的程序就好了

[jacklove]

http://www.cora.nwra.com/~stockwel/
作者主页上有的

[juliaxj]

那个网页我进去过了的，C程序没法下载。
不过，没关系了

[yht721521]

请问有S反变换的MATLAB程序吗？谢谢

[kjq0321]

好东西，一定要顶~

[余依宁]

好东西，顶

[余依宁]

弱弱的问一句，楼主给的st 的程序的链接里
进行[st_matrix_chirp,st_times,st_frequencies] = st(chirp)得到的是chirp时间-频率-振幅图还是时间——频率——能量图？程序没看懂，求指导

[余依宁]

引用框（blueocean ＠ 03-20-2006 - 07:50）

----------------------------------------------------------------------------------------------------------
本主题是专题讨论"时频分析及其在地震勘探中的应用"的一部分，欢迎访问该专题讨论的主贴：
http://forum.petro-c...?showtopic=4585
----------------------------------------------------------------------------------------------------------


<span style='font-size:14pt;line-height:100%'>In 1996, Stockwell published one paper about s-transform. S-transform can be considered as pseudo-continous wavelet transform. The normal CWT is transformed to (time, scale) domain, while S-transform is transformed to (time, frequency) domain. On the other hand, s-transform is very similar to Short Window Fourier Transform. The difference is the frequency dependent window length. The kernel basis of s-transform is scaled and frequency dependent Morlet wavelet. </span>

<span style='font-size:14pt;line-height:100%'>Here is the Stockwell's webiste:
http://www.cora.nwra.com/~stockwel/

There're a bunch of information about s-transform.

The first s-tranform paper is:
http://www.cora.nwra...cs/loc_ieee.pdf

The matlab codes are:

http://www.cora.nwra...rm/m_files/st.m

http://www.cora.nwra...m_files/stest.m</span>

<span style='font-size:14pt;line-height:100%'>Contribution by Robert G. Stockwell</span>

请问楼主在上面的s变换中1.f = (minfreq + [0:spe_nelements-1]*freqsamplingrate)/(samplingrate*length(timeseries))是为什么？为什么要除以(samplingrate*length(timeseries))呢？
2 .ind = [0:n-1]';
r = polyfit(ind,timeseries,2);
fit = polyval(r,ind) ;
timeseries = timeseries - fit;
为什么要进行上面处理而不直接对timeseries进行处理
求指导