Not entirely sure what's happening here, but I would expect sum(img_blurred) == sum(img). Maybe consider adding a unitary parameter to fftconv().
Note that simply setting normalize=True in the kernel-creation call to gauss() may be part of the solution but it blows up all of the tests.
Not entirely sure what's happening here, but I would expect
sum(img_blurred) == sum(img). Maybe consider adding aunitaryparameter tofftconv().Note that simply setting
normalize=Truein the kernel-creation call togauss()may be part of the solution but it blows up all of the tests.