Professional Documents
Culture Documents
by swashford
Slide partially based on Stanford U. CS131
– Articulation
– Intra-category variations
normalized region
e.g. color e.g. color
• Goals:
• Key property:
– In the region around a corner, image gradient has two or
Small
Edge
Large
Small
Nothing
Small
Slide partially based on Stanford U. CS131 Slide partially from Selim Aksoy
METU Lecture 6 - 25 CENG 483
Harris Detector Formulation
• This can be rewritten:
Slide partially based on Stanford U. CS131 Slide partially from Selim Aksoy
METU Lecture 6 - 26 CENG 483
Harris Detector Formulation
• This measure of change can be further simplified as:
Gradient with
respect to x,
times gradient
Slide credit: Rick Szeliski
Image I Ix Iy IxIy
Slide credit: Rick Szeliski
• Reminder:
Slide partially based on Stanford U. CS131 Slide partially from Selim Aksoy
METU Lecture 6 -
Corner Response Function
λ2 “Edge”
θ<0
“Corner”
θ>0
Slide credit: Kristen Grauman
• Fast approximation
– Avoid computing the
eigenvalues
– α: constant
(0.04 to 0.06) “Flat” “Edge”
region θ<0
λ1
Slide partially based on Stanford U. CS131
3. Gaussian
filter g(σI)
g(Ix2) g(Iy2) g(IxIy)
4. Cornerness function – two strong eigenvalues
METU Lecture 6 - 35
Harris detector example
Slide partially based on Stanford U. CS131 Slide partially from Selim Aksoy
METU Lecture 6 -
Harris detector example
Corner responses R
Local maxima of R
Slide partially based on Stanford U. CS131 Slide partially from Selim Aksoy
METU Lecture 6 - 39
Harris detector example
• Translation invariance?
• Translation invariance
• Rotation invariance
• Scale invariance?