Overview

Close-range photogrammetry is a well-established field, however limited research to date has been done into the effects of synchronisation delays and shutter speed on positional accuracy, which apply particularly in reference to moving targets.

The overarching topic for this thesis project is an investigation into close-range photogrammetry of moving objects. This topic has been broken down into the following four aims which have been addressed over the course of the year.

  1. To determine the time delay between “synchronised” camera shutters.
  2. To determine the effect of shutter synchronisation errors on point accuracy.
  3. To investigate the impact of various shutter speeds on the positional accuracy of moving targets.
  4. To investigate the impact of various target speeds on the positional accuracy of close range photogrammetry.

The first two aims were addressed using the data from Experiment 1. Then, taking into account the calculated synchronisation delay, Experiment 2 was shaped to address the second two aims. Both experiments are outlined below.

Experiment 1:
Shutter Synchronisation

Photographs were taken of a 60Hz refresh rate CRT (cathode-ray tube) screen to create a bright horizontal “refresh bar” across the screen in the photographs. The cameras were set up roughly perpendicular to the surface of the screen, as can be seen in the diagram to the left, and the cameras were connected to a single remote control.

The relative position of the refresh bar in each image was then measured against the static on-screen image of an arbitrary scale with lines numbered 0 to 30. This meant that the synchronisation errors between the six cameras could be measured to within ±1millisecond (Aim 1).

The data from this experiment was then also used to theoretically determine the positional errors introduced by synchronisation errors for objects travelling at various velocities (Aim 2).

Experiment 2:
Accuracy of Moving Targets

This experiment involved photographing a rotating beam with the six "synchronised" cameras, using a variety of different shutter speeds. The experiment setup, as can be seen in the image to the left, consisted of two stationary wooden posts covered in targets and a freely-rotating 1.8 metre beam with a variety of target sizes (18-38mm).

The beam was rotated at approximately 1 rotation per 3 seconds (120° per sec) in order to ensure that the speed of all targets were below PhotoModeler's recognition threshold. Seven sets of images were taken with shutter speeds reanging between 1/500s and 1/4000s, however only three sets were analysied due to time constraints.

Significant problems encountered in various sets included: blurred images due to an incorrectly set focus, un-recognisable targets, and overexposure.

The residuals and circular fit errors of each recognised target were collated and analysed to determine the positional accuracy effect of shutter speed (Aim 3) and target speed (Aim 4).

mouseover