Está en la página 1de 10

Point Detection

Dan MacLean
What is it?
z Consists simply of detecting sharp
transitions in intensity (discontinuities) in
an image via a mask.
z Very straightforward in principle.
Equations!
z R = w1 z + w z + … + w z
1 2 2 9 9

z = 9 wz

i =1
i i

z W is the weight of that mask element


z Z is the gray level of the pixel in that mask
element
z Result of the mask is reflected through the
center location of it.
How to actually detect points
z |R| >= T
z T is a non-negative threshold
z An isolated point will be different from
neighboring pixels, therefore easily
detectable.
3x3 Mask with Coefficients
More on Point Detection
z Same as one implementation of
Laplacian mask shown in Ch. 3.
z Coefficients sum to 0, so mask will yield
0 in areas of constant gray level.
Problems
z Specialized approach
z Only looks for single pixel discontinuities
z Problems with this mask occur when
discontinuities appear in clusters greater
than 1 pixel.
z Larger mask?

También podría gustarte