Volume equivalent drop diameters were obtained from the captured
drop images by calculating the volume of the drop in each image frame
with the assumption of rotational symmetry of the drops. Using sequential
high-speed images of the same drop at a given measurement
station, the frame-averaged drop diameters were calculated by averaging
the volume equivalent diameters obtained from each of the frames.
The frame-averaged volume equivalent diameters were used in calculations
for the drop shape evolution analysis that is presented in Section 3.
To assess the impact of rotational symmetry assumption in the measurement
accuracy, the following two different comparisons were
made. For the first comparison, the frame-averaged drop diameter
values were compared with the average drop diameter for a given
needle, which was calculated by averaging the frame-averaged diameters
of all the drops captured at a given measurement station. This
comparison showed that frame-averaged drop diameter values were
very close to the average drop diameter values (i.e. standard deviations
less than 0.2 mm). For the second comparison, the average drop diameters
at different measurement stations for a particular needle were
compared. This comparison showed that the difference between the average
drop diameters of equilibrium-shaped drops and oscillating drops
was marginal (less than 3%). Based upon these comparisons, it was concluded
that measurement errors for the volume equivalent diameter
due to rotational symmetry assumption was insignificant for our
purposes.
To estimate the measurement error bounds, tests with spherical ball
lenses (refractive index of 1.46 and sphericity of ±2.5 μm) of 3 and
5 mm diameters were performed at different distances from the focal
plane and the test images were processed using the fixed thresholding
technique. For spherical lenses located at 3 cm towards the camera
from the focal plane, the measurement error in volume equivalent diameter
was less than 10%. For spherical lenses located at 5 cm towards
the light source from the focal plane, this error was less than 5%. Fig. 5
shows the captured images for both of these cases for the 5 mm and
3 mm spherical lenses. For improved measurement accuracy, an operator
visually selected the drop images that corresponded more to Fig. 5b
(than Fig. 5a) for analysis. Consequently, the measurement error of