Here is my code:
# 'a' is a 3D array which is the RGB data a = np.array([[[1, 2, 3], [4, 5, 6]], [[4, 5, 6], [1, 2, 3]]]) # I want to do some calculate separately on R, G and B np.apply_along_axis(lambda r, g, b: r * 0.5 + g * 0.25 + b * 0.25, axis=-1, arr=a) # my target output is: [[1.75, 4.75], [4.75, 1.75]]
But the above way will give me error, missing 2 required positional arguments.
I have tried to do like this:
np.apply_along_axis(lambda x: x[0] * 0.5 + x[1] * 0.25 + x[2] * 0.25, axis=-1, arr=a)
It works but every time when I need to do computation on the array element, I need to type the index, it is quite redundant. Is there any way that I can pass the array axis as multi iuputs to the lambda when using np.apply_along_axis?
Advertisement
Answer
apply_along_axis
is slow and unnecessary:
In [279]: a = np.array([[[1, 2, 3], [4, 5, 6]], [[4, 5, 6], [1, 2, 3]]]) In [280]: r,g,b = a[...,0],a[...,1],a[...,2] ...: r * 0.5 + g * 0.25 + b * 0.25 Out[280]: array([[1.75, 4.75], [4.75, 1.75]])
or
In [281]: a.dot([0.5, 0.25, 0.25]) Out[281]: array([[1.75, 4.75], [4.75, 1.75]])
or
In [282]: np.sum(a*[0.5, 0.25, 0.25], axis=2) Out[282]: array([[1.75, 4.75], [4.75, 1.75]])