Figure 14.7 The action of blur kernels is easy lo understand—each pixel in the new image is the kernel-weighted average of a group of pixels in the old image. In the sequence above, progressively more weight is given to outer neighbors over inner neighbors. Image of the lunar crater Clavius by Steve Lee.
Gaussian, x 5 Boxcar
The gentle blur kernel: ~ x 16
endows the central pixel with the same weight as the sum of the other eight pixels in the neighborhood. The smoothing effect is gentle enough that it should be called a "softening" kernel. The filter pass-through rating is 0.5.
Was this article helpful?