q***o 发帖数: 484 | 1 如果求
s1&s2
(&设为卷积运算)
可以用付氏变换先转换到频域,求积后在反付氏变换即可.
那么对应逆卷积的问题(即,s1&s2=ss,现已知s2,ss,求s1)
按道理是可以采用上述相同方法的,只不过把频域的求积
换成求商,现在的问题是,s2变换成频域后,无法保证不出现
0项,进行求商运算的时候就出现'除零'的情况.
请问在这种情况下,如何解决?
另外有没有别的求去卷积的方法?
谢谢. | c****u 发帖数: 584 | 2
Hehe I am struggling with this problems for weeks. I found FFT is the worst
method to do deconvolution in my case, since there is some Poission noise in
the signal, and I got horrible result after deconvolution. I am now try to use
the iterative method now, that's pain.
【在 q***o 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 如果求 : s1&s2 : (&设为卷积运算) : 可以用付氏变换先转换到频域,求积后在反付氏变换即可. : 那么对应逆卷积的问题(即,s1&s2=ss,现已知s2,ss,求s1) : 按道理是可以采用上述相同方法的,只不过把频域的求积 : 换成求商,现在的问题是,s2变换成频域后,无法保证不出现 : 0项,进行求商运算的时候就出现'除零'的情况. : 请问在这种情况下,如何解决? : 另外有没有别的求去卷积的方法?
| q***o 发帖数: 484 | 3
the worst
noise in
try to use
How did you do with 'iterative method'.
I am trying to use 'deconv' function in matlab, but still not work.
//t.
【在 c****u 的大作中提到】![](/moin_static193/solenoid/img/up.png) : : Hehe I am struggling with this problems for weeks. I found FFT is the worst : method to do deconvolution in my case, since there is some Poission noise in : the signal, and I got horrible result after deconvolution. I am now try to use : the iterative method now, that's pain.
|
|