Light-Field Imaging in Laboratory Studies of Planet Formation

Understanding the steps involved in the formation of planetary systems is one of the key challenges currently faced within Astronomy. Modelling planet formation requires empirical data from collisions involved in the earliest stages of the process. As this data is not directly observable, laboratory-based studies are performed in which dusty, icy aggregates collide in conditions analogous to those of a protoplanetary disk. Technology capable of capturing footage of these collisions and extracting the three-dimensional motions of the aggregates is crucial to accurately analysing them in a practical time frame. This thesis aims to determine whether light-field based technology can be used to automate the 3D tracking of aggregates colliding in conditions appropriate for the earliest stages of planet formation. To begin, pre-existing footage of dust collisions was analysed using traditional methods to provide a scientific grounding for the technology development to follow. Results of the science analysis suggested that planet formation models should consider the result of a temperature-dependent minimum fragmentation velocity for dust aggregates on their distribution and maximum possible size within a protoplanetary disk. Key technological challenges of existing analysis methods were also identified as: the technical specification of the camera, the experimental lighting setup, and the data extraction and analysis methods. Light-field technologies were then developed over three main iterations. Each version built on the principle of a light-field camera, using an array of micro-lenses to capture multiple views of a scene on a single sensor. These views could then be processed to establish the 3D location of observed objects. Firstly, light-field calibration methods were developed alongside software to automate detection and tracking processes. These were used to extract the 3D motion of bb-pellets from which collision parameters were calculated. This showed the feasibility of using light-field techniques for tracking of millimetre-sized objects as well as the 500 times reduction in analysis time possible by automating the process. For the second iteration, the light-field camera design was adapted for integration with the Open University's aggregate collision experiment. This aimed to assess the suitability of the technology for imaging and analysing ice-based collision experiments. Collisions of water ice aggregates were recorded using the camera and analysed by adapting the previously developed software. Using the existing mounting setup of the experiment, the camera had to be positioned more than 27 cm from the colliding aggregates. This was identified as a limitation to how accurately an aggregate could be located in 3D space. The collision experiment's view port was replaced with a custom re-entrant design to allow the camera to be mounted closer to the colliding aggregates. A third iteration light-field camera was designed to make use of this new view port. This led to a reduction in the variation in location measurements estimated by each stereo-pairing of the light-field camera. Millimetre-sized ice aggregates were successfully tracked with an uncertainty of ± 0.12 mm when measuring the relative location of an aggregate from one frame to the next. Parameters were also extracted to define the ice-aggregate collisions including the relative velocities of the aggregates and coefficients of restitution. The work of this thesis outlines the first use of light-field technologies in experiments recreating the collisions involved in forming planets. It highlights the potential of the technique for future experiments designed with automated aggregate detection in mind and where size and weight constraints exist. By automating the analysis process, large numbers of collisions can be processed giving insight into the possible range of collision parameters at a level not previously achievable. Providing this empirical data will be key to ensuring the accuracy of future computational models of the planet formation process.