
194 Digital Geometry in Image Processing
(a) (b) (c)
FIGURE 6.4: (a) An object (set-square) (b) distance transform using {112},
(c) Euclidean distance transform us ing 8SED algorithm [53].
[53], which is base d on local propagation of distance values and uses a scheme
similar to chamfering. Danielsson reported this algorithm in the early eight-
ies. Subsequently, other researchers [133, 170] reported many variations and
improvements of this scheme.
In the c omputation of EDT, initially all the object points are assigned to
a large distance value and all the background points are set to zeroes. How-
ever, in the representation of Euclidean distance at ...