Proceedings of the Institution of
Civil Engineers
Transport 157
November 2004 Issue TR4
Pages 252–254
Paper 2312795
Paper published
Proc. Instn Civ. Engrs Transport,
2003, 156, Feb., 17–28
Discussion: Segmentation algorithms for road marking digital
image analysis
M. P. N. Burrow, H. T. Evdorides and M. S. Snaith
S. Xavier, Highways Agency, Traffic Signs Team, Dorking Office, UK
1. Congratulations to the authors! This is a very useful
exercise as it uses the mathematical technique of
algorithms to study the body of road markings against the
contrast of bituminous road surface. It is able to identify to
what extent the material present is worn out. This is very
useful for high speed inspection purposes as a walk about
inspection with the involvement of traffic management is
not only resource intensive but also disrupts traffic flow
and causes delay to road users.
2. Practising civil engineers approach this aspect of the
inspection of road markings in a different way as set out in
BS 3262 which is now superseded by BS EN 1824.
1,2
The
former used a wire grid to place over the road marking. In
that method a panel of inspectors formed a view of how
much of the road marking remained within each grid
square in a formal road trial and summed it over the test
area and applied a simple calculation to determine the wear
index. In the BS EN 1824 there is a ‘faster’ method that
permits a coloured photograph to be taken on site and
analysed back in the office. Instead of this somewhat
cumbersome operation, an alternative permitted in the BS
EN 1824 is to measure the property known as minimum
luminance coefficient in diffuse illumination ( Q
d
) using a
portable instrument. This does not, however, tell us
anything about how much of the line is worn out but
nevertheless measures reflection under daylight or street
lighting. This helps to form a view about threshold values
for pre-view time of the road users. Incidentally, there is a
useful European report and software for this which
unfortunately does not take into consideration the presence
of road studs.
3
I believe the BS ENs are reliable as there are
years of tried and tested engineering methods behind them
but they are costly and disruptive to traffic as the authors
themselves have found.
3. In this context, the authors proposal is very attractive if
photographic or video images can be taken from a high
speed monitor and they are speedily analysed for useful
results for decision making to replace a badly eroded line.
Some pioneering work in this area was carried out at TRL
between 1992 and 1995. Both Mr B. W. Ferne and I refer to
it in our earlier discussion contribution.
4
Moving on from
there, there are two monitors used by industry for a
different measurement at high speed. They are known as
the Ecodyn and Laserlux. Both measure the property of
retroreflection (R
L
) which is important for visibility at
night. The latter system also has an option to take video
images and a GPS facility.
4. Having taken snap shots at high speed, which may be
considered as equivalent to taking walk about inspection
photographs to BS EN 1824, the next step is analyse them
in minute form for engineering decision making processes.
5. The authors have taken the trouble to study seven
photographs. Simple visual inspection is described but I
wonder what would be the corresponding assessment using
the BS 3262 grid method and BS EN 1824 method albeit
the photographs may be in black and white. It may be Q
d
measurements were not taken on site at that time or
samples may not have been taken for closer examination in
the laboratory.
6. The Highways Agency maintenance standard TD 26 pre-
publication draft
5
gives guidance on measurement on wear
index or Q
d
by inspectors under traffic management
conditions permitting. The use of high speed monitors for
the measurement of R
L
is also permitted subject to
correlation with static measurement instruments.
7. Perhaps the authors could comment on whether the
algorithmic methodology could be easily devised to form a
view of how much of the road markings were remaining
per square metre in each of the seven photographs and if
so what the results are and how they compare with BS
3262 and BS EN 1824 methods. Do they have a view on
how long they need to develop their methodology before
road making industry partners who are already operating
Edcodyn
6
and Laserlux
7
may begin to consider them as
useful bolt-on devices to their systems?
REFERENCES
1. BS EN 3262-2: Hot-applied thermoplastic road marking
materials – Specification for road performance. British
Standards Institution, London, 1999.
2. BS EN 1824: Road marking materials – Road trials. British
Standards Institution, London, 1999.
3. European Commission 1999, Directorate General Transport,
Co-operation in the Field of Scientific and Technical
Research, COST 331 – Requirements for Horizontal Road
Markings – Final Report of the Action.
Transport 157 Issue TR4 252 Discussion
Downloaded by [] on [22/05/20]. Copyright © ICE Publishing, all rights reserved.