<s>
In	O
image	B-Algorithm
processing	I-Algorithm
,	O
line	B-Algorithm
detection	I-Algorithm
is	O
an	O
algorithm	O
that	O
takes	O
a	O
collection	O
of	O
n	O
edge	B-Algorithm
points	I-Algorithm
and	O
finds	O
all	O
the	O
lines	O
on	O
which	O
these	O
edge	B-Algorithm
points	I-Algorithm
lie	O
.	O
</s>
<s>
The	O
most	O
popular	O
line	O
detectors	O
are	O
the	O
Hough	B-Algorithm
transform	I-Algorithm
and	O
convolution-based	O
techniques	O
.	O
</s>
<s>
The	O
Hough	B-Algorithm
transform	I-Algorithm
can	O
be	O
used	O
to	O
detect	O
lines	O
and	O
the	O
output	O
is	O
a	O
parametric	O
description	O
of	O
the	O
lines	O
in	O
an	O
image	O
,	O
for	O
example	O
ρ	O
=	O
r	O
cos(θ )	O
+	O
c	O
sin(θ )	O
.	O
</s>
<s>
Therefore	O
,	O
a	O
line	O
in	O
the	O
image	O
corresponds	O
to	O
a	O
point	O
in	O
the	O
Hough	B-Algorithm
space	I-Algorithm
.	O
</s>
<s>
The	O
Hough	B-Algorithm
space	I-Algorithm
for	O
lines	O
has	O
therefore	O
these	O
two	O
dimensions	O
θ	O
and	O
ρ	O
,	O
and	O
a	O
line	O
is	O
represented	O
by	O
a	O
single	O
point	O
corresponding	O
to	O
a	O
unique	O
set	O
of	O
these	O
parameters	O
.	O
</s>
<s>
The	O
Hough	B-Algorithm
transform	I-Algorithm
can	O
then	O
be	O
implemented	O
by	O
choosing	O
a	O
set	O
of	O
values	O
of	O
ρ	O
and	O
θ	O
to	O
use	O
.	O
</s>
<s>
In	O
a	O
convolution-based	O
technique	O
,	O
the	O
line	O
detector	O
operator	O
consists	O
of	O
a	O
convolution	B-Algorithm
masks	O
tuned	O
to	O
detect	O
the	O
presence	O
of	O
lines	O
of	O
a	O
particular	O
width	O
n	O
and	O
a	O
θ	O
orientation	O
.	O
</s>
<s>
Here	O
are	O
the	O
four	O
convolution	B-Algorithm
masks	O
to	O
detect	O
horizontal	O
,	O
vertical	O
,	O
oblique	O
( +45	O
degrees	O
)	O
,	O
and	O
oblique	O
( −45	O
degrees	O
)	O
lines	O
in	O
an	O
image	O
.	O
</s>
