search

KeKsBoTer / torch-dwt

1D,2D,3D Discrete Wavelet Tansform (DWT) implemented in Pytorch
7 stars 2 forks source link

A problem about the change of the width, height and depth #1

Open tq-quan opened 1 year ago

tq-quan commented 1 year ago

The width, height and depth have change if we perform 3d dwt on a tensor, whose size is (1, 1, 100, 100, 100). After the forward dwt operation, the width, height and depth are 51, 51, 51, respecively. Not what we expect : 50, 50, 50, look forward to your reply, thx.

RY-97 commented 4 months ago

The width, height and depth have change if we perform 3d dwt on a tensor, whose size is (1, 1, 100, 100, 100). After the forward dwt operation, the width, height and depth are 51, 51, 51, respecively. Not what we expect : 50, 50, 50, look forward to your reply, thx.

I was doing a similar job recently and encountered the same problem. I would like to ask for your help. I want to input a (b, c, w, h, d) 3D image data. The output after discrete wavelet transform is still consistent in shape, but does not require inverse wavelet transform. Do you have any solution?

39c5bb-miku commented 1 month ago

The width, height and depth have change if we perform 3d dwt on a tensor, whose size is (1, 1, 100, 100, 100). After the forward dwt operation, the width, height and depth are 51, 51, 51, respecively. Not what we expect : 50, 50, 50, look forward to your reply, thx.

You can try other wavelets, such as the haar wavelet, which will become 50, 50, 50