Image Processing Reference
In-Depth Information
(a) Circle
(b) Soccer ball edges
(c) Noisy soccer ball edges
(d) Accumulator for (a)
(e) Accumulator for (b)
(f) Accumulator for (c)
(g) Circle from (d)
(h) Circle from (e)
(i) Circle from (f)
Figure 5.11
Applying the Hough transform for circles
maximum number of points; it is possible to include other constraints to control the circle
selection process, such as gradient direction for objects with known illumination profile. In
the case of the human eye, the (circular) iris is usually darker than its white surroundings.
Figure 5.12 also shows some of the difficulties with the HT, namely that it is essentially
an implementation of template matching, and does not use some of the richer stock of
information available in an image. For example, we might know constraints on size ; the
largest size and iris would be in an image like Figure 5.12 . Also, we know some of the
topology : the eye region contains two ellipsoidal structures with a circle in the middle.
Search WWH ::




Custom Search