CHARACTERIZATION OF HAWAII SUPERPAVE ASPHALT CONCRETE USING IMAGE ANALYSIS
H.G. Brandes(P.I.), C.T. Nagata
Funded By: Hawaii Department of Transportation and the Federal Highways Administration
Department of Civil Engineering
Content
Application of Image Analysis Techniques
Figure 1 – Hawaii’s 1st Superpave
Road Project
Introduction
Background
The Hawaii Department of Transportation
(HI-DOT) is in the process of implementing the new Superpave
paving system.A major objective of Superpave is to improve long-term pavement
performance by tightly controlling the component materials.The internal
structure of asphalt pavements play a significant role in the ability of the
pavement to resist rutting, fatigue cracking, thermal cracking, and
low-temperature cracking. This study is intended to develop a fast, non-destructive,
environmentally friendly process that uses image analysis of digitized images
to quantify geometric feature such as aggregate particle size, elongation,
angularity, gradation, and percentage voids.In fact, image analysis has the
potential to characterize and investigate the effects of a large range of
particles and void space geometric properties and spatial arrangement measures
which are not usually considered even in the new Superpave mix design system.
The
15 Superpave Gyratory Compactor (SGC) specimens used in this study were
obtained during the mix design process for Hawaii’s first Superpave road
project(Figure 1).Figure 2 shows the 15 core samples
after they were cut lengthwise for imaging.The SGC specimens are 6 inches in
diameter and 4.5 inches high.Figure 3 shows a
close-up of one such specimen.
|
Table1.
Properties of Asphalt and Aggregates Used in Mixture Design |
||
|
Aggregate
Gradation |
|
|
Sieve(mm) |
Percent passing
Kapa'a Quarry |
Percent Passing
Makakilo Quarry |
|
50 |
100 |
100 |
|
37.5 |
100 |
100 |
|
25 |
100 |
100 |
|
19 |
99.9 |
99.7 |
|
12.5 |
89.1 |
76 |
|
9.5 |
70.7 |
60.2 |
|
4.75 |
42.3 |
42 |
|
2.36 |
27.9 |
27.7 |
|
1.18 |
17.9 |
18 |
|
0.6 |
12.1 |
12.2 |
|
0.3 |
8.8 |
8.8 |
|
0.15 |
6.8 |
6.8 |
|
0.075 |
5.3 |
5.2 |
Application of Image Analysis Techniques
Images were acquired by scanning the
asphalt core sample using a high resolution scanner set at 1200 dots per inch
(dpi).Figure 5 shows a digitized image of a
vertical section of compacted asphalt.The asphalt images are enhanced using
level and un-sharp masking filters and then reduced to 300 dpi using Adobe Photoshop.The
images are then processed with Matlab and its Image Processing
Toolbox.
At
present, the image analysis program being written is able to determine void
ratio, asphalt content, gradation, elongation, and percent air voids.Currently,
the majority of the work is focusing on image-based aggregate gradation curves
for comparison with true gradations.This includes investigating the effects of
gray scale threshold levels, 3-D particle frequency modeling from
two-dimensional image data, and other aspects.Figure 6
and 7 show aggregate gradations from image
measurements and mechanical sieve analysis for the Kapa’a and Makakilo quarry
aggregates.
We
are confident that the algorithms being developed will be able to accurately
measure aggregate particles down to a 2 mm diameter.Due to limitations
associated with image acquisition, resolution, and specimen size, it will be
difficult to match results at sizes less than about 2mm.Aggregate particle
segmentation is another limiting aspect of image analysis that we are
investigating.Figure 8 shows aggregate segmentation
for one of the digitized samples.Note that the current technique, based on
neighborhood relationships, is not quite successful yet.This study has just
begun and will continue for the next two years.There are many other exciting
aspects that we would like to explore.Eventually, we hope to be able to
investigate correlations between image-based properties that are not routinely
determined during mix design and field performance.