A Computational Introduction to Digital Image Processing, 2nd Edition

Neighborhood Processing 95

So to obtain, for example, the image in Figure 5.6(c), you could use the following Python

commands:

Python

In: c = io.imread(’cameraman.png’)

In: cf = ndi.convolve(c,ones((9,9))/81,mode=’constant’)

In: io.imshow(cfs)

Alternately, you could use the built in uniform

_

filter f unction:

Python

In:

cf = ndi.uniform

_

filter(c,[9,9],mode=’constant’)

Python differs from MATLAB and Octave here in that the default behavior of spatial

convolution at the edges of the image is to reflect the image in all its edges. This will

ameliorate the dark borders seen in Figure 5.6. So with leaving the

mode parameter in its

default setting, the commands

Python ...