I adjusted the filter transformation (2-D and 3-D) to work with non-square data.
I also made a test sample (Haar transform 4x8 matrix) with PyWavelets, since I'm not sure if Wavelab supports 2-D transform on non-square signals.
Should there also be testcases for other Wavelet classes?
Coverage increased (+0.04%) to 96.0% when pulling 8b687ec0283376918ae8363aec4f379d6954908e on antholzer:rectangular into 000d2eb6f84bdb14dfdbe32394d58502d6e6fb3d on JuliaDSP:master.
Thanks for that. I thinks the tests are ok for now. Tests could be added later comparing random signals to the inverse transform of transformed signal.
I adjusted the filter transformation (2-D and 3-D) to work with non-square data.
I also made a test sample (Haar transform 4x8 matrix) with PyWavelets, since I'm not sure if Wavelab supports 2-D transform on non-square signals. Should there also be testcases for other Wavelet classes?