mexh wavelet about pytorch-tvmisc HOT 3 CLOSED

Hi,

the Mexican hat wavelet is a continuous wavelet, so you cannot use it in a discrete wavelet transform directly.
If you need just one convolution, you could discretise the wavelet on some grid and use that in the convolution. Likely, you will need rather large kernel widths (and have more boundary issues / need for padding).
Do you have a particular use case in mind?

Best regards

Thomas

from pytorch-tvmisc.

thanks for your response,
i want to use mexhat for cwssim implementation. i want to convolve a matrix with mexhat kernel.

from pytorch-tvmisc.

I'm afraid this is out of scope for now.

from pytorch-tvmisc.