Linearity and convolution
Linearity ….
• An operator H (or system) is linear if two properties
hold (f 1 and f 2 are some functions, a is a constant):
• Additivity :
H ( f 1 + f 2 ) = H ( f 1 ) + H ( f 2 ) (distributive law)
• Multiplicative scaling (Homogeneity):
H ( a • f 1) = a • H ( f 1 ) (constant scales) a times H applied to f1
Linearity ….
if we could sum up a bunch of things to make an image, then if
we apply a linear operator to that whole image, it's going to be
the same as the sum of applying that linear operator to each of the
pieces.
An impulse function...
• In the discrete world, an impulse is just a value of 1
at a single location.
An impulse response
• If we have an unknown system and we "put in" an impulse, the
response is called the impulse response.
• So if the black box is linear you can describe H by h(x)
Filtering an impulse signal
Impulse response looks like in an image.
Result of filtering the impulse signal (image) F with the arbitrary kernel H?
Filtering an impulse signal
Filtering an impulse signal
Filtering an impulse signal
Filtering an impulse signal
Filtering an impulse signal
Assuming center
coordinate is "reference point".
Correlation vs Convolution
We had this problem, of when we put in through a correlation, we got back out
sort of this flip thing.
Convolution
Shift invariant:
• Operator behaves the same everywhere, i.e. the value of the
output depends only on the pattern in the image neighborhood,
rather than the position.
Properties of convolution convolution or correlation are built on
multiplication and
addition, these are linear operators.
Computational Complexity
Separability
• In some cases, filter is separable, meaning you can get the
square kernel H by convolving a single column vector by some
row vector:
r
c H
Separability
r
c H
G = H * F = (C * R) * F = C * (R * F)
• So we do two convolutions but each is W*N*N. So this is useful
if W is big enough such that 2 • W • N2 << W2 • N2
• Used to be very important. Still, if W=31, save a factor of 15.
Boundary issues
• What is the size of the output?
• We have three choices:
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
Periodic signal
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
• reflect across edge
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
• reflect across edge
Called symetric
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
• reflect across edge
Called symetric
Boundary issues
What about near the edge?
• methods:
• clip filter (black)
• wrap around
• copy edge
• reflect across edge
Called symetric
Boundary issues
• methods (IPT MATLAB):
• clip filter (black): imfilter(f, g, 0)
• wrap around: imfilter(f, g, 'circular')
• copy edge: imfilter(f, g, 'replicate')
• reflect across edge: imfilter(f, g, 'symmetric')
Practice with linear filters
Practice with linear filters
Practice with linear filters
Practice with linear filters
Filtering examples: sharpening
Different kinds of noise
• Gaussian averaging was a reasonable thing
to do if the noise was independent at each pixel and
centered about zero such as if created by a Gaussian
or normal noise process.
• But there are other kinds of noise.
Different kinds of noise
Additive Gaussian noise Salt and pepper noise
Recall: Assumptions for removing noise
• The way to approach this is to remember our other assumption
about images: that the underlying image changes smoothly around a
pixel. So if you were to inject some arbitrary value in there the
question is how to find the original pixel.
• Remember that when we are the blurring, we replace the pixel value
by the local average. That's fine when the noise is modest and tends
to add to zero over a neighborhood. But if there are a few totally
random values thrown in, we need another approach.
• As many of you know the way to deal with such ugly perturbation is
to use what's called a median filter.
Median filter
No new pixel
values introduced.
Removes spikes:
good for impulse,
salt & pepper noise.
Linear?
Salt and Median
pepper filtered
noise
Plots of a row of the image
Median filter
Median filter is edge preserving
Normalized Correlation
1D (nx)correlation
1D (nx)correlation
Signal
Filter
1D (nx)correlation
Signal
Filter
Normalized
cross-correlation
Matlab cross-correlation doc
Matlab cross-correlation doc
Matlab cross-correlation doc
Template matching Filters as templates
A toy example
Template (mask)
Scene
K. Grauman
Template matching
Template (mask)
Detected template
Template matching
Detected template Correlation map
Where's Waldo?
Scene
Template
K. Grauman
Where's Waldo?
Template
Detected Template
Where's Waldo?
Detected template Correlation map
Template matching
What if the template is not identical to some sub¬
image in the scene?
Template
Scene K. Grauman
Template matching
Match can be meaningful, if scale, orientation, and
general appearance is right.
Detected template
