Aim 3:
Investigating the Impact of Various Shutter Speeds on the Positional Accuracy of Moving Targets

When comparing between similar-sized targets, no significant trend was determined between the shutter speed and the circular fit error of either the moving or the static targets.

There was a trend where moving targets had a larger average residual at faster shutter speed, however there was no way to determine what proportions of this was due to the synchronisation errors, to differences in target sizes, or to other contributing factors. Hence, no meaningful conclusion could be drawn from this.

An unexpected result that was discovered, however, was that in images with both static and rotating targets, the static targets had higher residuals at faster shutter speeds, as can be seen in the figure below. This is caused by the adjustment that needs to be performed by the software in order to obtain the smallest sum of residuals for all targets. This adjustment is necessarily greater at faster shutter speeds due to the synchronisation errors; there is a longer time difference between images. This is discussed in further depth in the previous section 4.2.1. Residuals.


Average Residuals of Static Targets.

This principle would extend to images within which the targets are moving unrelated to one another; for example, some targets moving from side to side, some rotating, and some moving up and down. However in this case the relative proportion of the residual due to this effect would be much more difficult to determine than it is for static targets.

Static Targets - Residuals

The static targets were expected to have similar average residuals in each set of images, as they would not have been affected by the synchronisation errors when stationary. However, as seen above, this was not the case.

The most likely explanation for this has to do with the fact that PhotoModeler cannot recognise that some targets (i.e. the static targets) should be considered “better” than others.

Since there is no way for PhotoModeler to take this into account, all targets are treated as having similar inter-image positional movements. Essentially, this means that the positions of all photographed targets were “rotated” to find the best overall fit. This idea is demonstrated in the diagram below.


Schematic illustration of target rotation in each image .
(Target 1 = Static, Target 2 = Moving)

The synchronisation errors between two cameras will result in slight differences in the target positions. When referenced in PhotoModeler, the “median” position is determined (red dotted line), which determines how far each image needs to be “rotated” to find the smallest sum of residuals for all targets. In order to do this, all targets are adjusted, not just the moving targets.

At faster shutter speeds, there is a larger time difference between the end of the image capture from the first camera and the start of the image capture from the last capture, hence these images need to be rotated more than those taken at slower shutter speeds.

mouseover