Problem 2 [16] Consider convolution based filtering given by (,) = (,)(,)(,). Assume that the image is and that the filter kernel is defined for an neighborhood. (a) [4] State what characterizes all smoothing kernels. Give an example of a smoothing kernel. (b) [2] State one simple way to handle boundary conditions. (c) [6] Explain what is meant by kernel separability and how it affects the computational cost associated with convolution based filtering. Give an example using a 3x3 kernel.

(d) [4] Explain how unsharp masking (aka sharpening) can be accomplished by smoothing. State the formula.

Consider convolution based filtering given by I(u,v)=(m,n)h(m,n)I(um,vn). Assume that the image is NxN and that the filter kernel is defined for an MxM neighborhood. (a) [4] State what characterizes all smoothing kernels. Give an example of a smoothing kernel. (b) [2] State one simple way to handle boundary conditions. (c) [6] Explain what is meant by "kernel separability" and how it affects the computational cost associated with convolution based filtering. Give an example using a 33 kernel. (d) [4] Explain how unsharp masking (aka sharpening) can be accomplished by smoothing. State the formula