A digital particle tracking algorithm designed to monitor small (10(-4)-10(
-2) m) movements of gravel clasts from digital video imagery is presented.
Using the image brightness values as a set of weights, the algorithm identi
fies the sub-pixel co-ordinates of black circular transfer markers applied
to contrasting tracer clasts. Once identified, marker centroids are tracked
from frame to frame at a sampling frequency of 25 Hz to give high resoluti
on time-traces df particle trajectories. Theoretical tests on the algorithm
using simulated data with additive Gaussian errors suggest centroid detect
ion to sub-pixel accuracies of 0.008 pixels (1 sigma). Experimental tests w
hich incorporate the effects of camera lens aberrations, CCD defects and su
b-optimal illumination degrade the precision by an order of magnitude, alth
ough centroid detection to 0.1 pixels at the 99% confidence interval is pos
sible. Finally, the algorithm was applied to monitor particle movements in
a gravel framework under shear. Preliminary results suggest significant exp
ansion of the coarse framework just prior to entrainment, a phenomenon whic
h may be important in controlling the pattern of fine sediment ingress. (C)
2000 Elsevier Science Ltd. All rights reserved.