Pavement Macrotexture Determination Using
Multi-View Smartphone Images
Xiangxi Tian, Yong Xu, Fulu Wei, Oguz Gungor, Zhixin Li, Ce Wang, Shuo Li, and Jie Shan
Abstract
Pavement macrotexture contributes greatly to road sur-
face friction, which in turn plays a vital role in reducing
road accidents. Conventional methods for macrotexture
measurement are either expensive, time-consuming, or of
poor repeatability. Based on multi-view smartphone images
collected in situ, this paper develops and evaluates an af-
fordable and convenient alternative approach for pavement
macrotexture measurement. Photogrammetric computer
vision techniques are first applied to create high resolution
point clouds of the pavement. Analytics are then developed
to determine the macrotexture metric: mean profile depth by
using the image-based point clouds. Experiments are car-
ried out with 790 images over 25 spots on three state routes
and six spots at an Indiana Department of Transportation
test site. We demonstrate multi-view smartphone images can
yield results comparable to the ones from the conventional
laser texture scanner. It is expected that the developed ap-
proach can be adopted for large scale operational uses.
Introduction
The quality properties of road pavements have direct and
significant impacts on road safety. According to
Indiana
Crash Facts 2017
(Sapp, Thelin, and Nunn 2017), about 200
000 vehicle crashes occurred in Indiana every year from 2013
to 2017, of which about 34 000 accidents resulted in injuries
and 755 accidents caused fatalities. Twenty percent of these
accidents were further attributed to insufficient surface fric-
tion on road curves.
It is therefore essential to provide adequate friction and
drainage to reduce the possibility of acci
These two properties are mainly determi
(surface) textures. Pavement texture, wh
the deviation of a pavement surface from a true planar surface
(Li, Wang, and Li 2016), directly affects various parameters
resulting from tire-road interactions such as friction, tire
noise, skid resistance, tire wear, rolling resistance, splash and
spray, traffic vibration, etc. (Ejsmont
et al.
2016; Das, Rosauer,
and Bald 2015; Yaacob
et al.
2014). It was suggested at the
1987 Permanent International Association of Road Congresses
(
PIARC
), depending on the amplitude and wavelength of a
feature, to divide the pavement surface characteristics (the
geometric profile of a road in the vertical plane) in four cat-
egories: roughness (unevenness), megatexture, macrotexture,
and microtexture (Dong, Prozzi, and Ni 2019; Bitelli
et al.
2012; Dunford 2013).
Roughness refers to the unevenness, potholes, and large
cracks on road surfaces that are larger than a tire footprint
(Dong, Prozzi, and Ni 2019; Bitelli
et al.
2012; Dunford 2013).
Megatexture is associated with deviations in wavelengths
from 50 mm to 500 mm and vertical amplitudes ranging from
0.1 mm to 50 mm (Dong, Prozzi, and Ni 2019; Dunford 2013).
Texture of this size is mainly caused by poor construction
practices or surface deterioration. This level of texture causes
vibrations in tire walls, resulting in vehicle noise and some
external noise. Macrotexture and microtexture refer to the
relatively small pavement surface irregularities that primar-
ily affect friction and skid resistance. Macrotexture refers to
the changes in wavelengths ranging from 0.5 mm to 50 mm
horizontally and variations ranging from 0.1 mm to 20 mm
vertically (Dong, Prozzi, and Ni 2019). However, microtexture,
which corresponds to wavelengths less than 0.5 mm horizon-
tally and vertical amplitudes up to 0.2 mm, is related to the
roughness of the individual stone elements used in the surface
layer and to the natural mineral aggregate (Bitelli
et al.
2012).
Although both microtexture and macrotexture contribute
to pavement friction, there is currently no practical procedure
for direct measurement of the microtexture profile in traffic
(Dong, Prozzi, and Ni 2019; Henry 2000). The
PIARC
Model for
the International Friction Index avoids the need for measuring
microtexture if macrotexture measures are available. A mea-
surement at any slip speed, together with the macrotexture pa-
rameter, determines the friction as a function of the slip speed
ical parameter to describe pavement
ean profile depth (
MPD
) or the mean
PD
is linearly related to
MTD
and is usu-
ally converted to
MTD
when comparing different macrotexture
calculation methods (Fisco and Sezen 2014; Henry 2000).
The conventional methods for determining road macrotex-
ture include the sand patch method, the outflow method, and
laser profiling. The sand patch method is operator-dependent,
and the test results have poor repeatability (Sengoz, To-
pal, and Tanyel 2012). Other problems with the sand patch
method include that on surfaces with very deep textures,
it is very easy to overestimate the texture depth (Fisco and
Sezen 2014), and accurate sand patch testing cannot be done
when the road surface is sticky or wet (Praticò and Vaiana
2015). As is the case with the sand patch method, the outflow
method also is labor-intensive and time-consuming, and the
reliability of the results depends largely on the operator. The
circular laser-based device has been deployed for routine
macrotexture measurement since 2002 (Abe
et al.
2001), and
the more portable handheld laser meter, such as the Ames
Xiangxi Tian, Zhixin Li, Ce Wang, and Jie Shan are with
Lyles School of Civil Engineering, Purdue University, West
Lafayette, IN 47907 (
).
Yong Xu is with the School of Geographical Sciences,
Guangzhou University, Guangzhou, Guangdong 510006, China.
Fulu Wei is with the Key Laboratory of Road and Traffic
Engineering, Ministry of Education, Tongji University,
Shanghai 201804, China.
Oguz Gungor is with the Department of Geomatics Engineering,
Karadeniz Technical University, Trabzon 61080, Turkey.
Shuo Li is with the Division of Research and Development,
Indiana Department of Transportation, West Lafayette, IN 47906.
Photogrammetric Engineering & Remote Sensing
Vol. 86, No. 10, October 2020, pp. 643–651.
0099-1112/20/643–651
© 2020 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.86.10.643
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
October 2020
643
