| Summary: | Plastic optical fibres (POF) can be made sensitive to various parameters. Therefore, a successful implementation of tomographic imaging based on POF sensors will open the way to develop imagers for various parameters utilising this inexpensive sensor material. The work reported in this thesis is aiming to implement real time tomographic imaging based on the POF sensor. As the system uses light (photons) guided along the sensor to capture the information and deliver a signal at the periphery of the sensor, this technique is referred to as Photonic Guided Path Tomography (PGPT). As an initial stage of the work, we perform systematic testing of the sensitivity of POF transmission to bending. The results confirm that grooving can enhance this sensitivity of the POF sensor. They also provide an empirical proof of the anticipated changes in transmission loss by positive and negative bending for grooved POF (opening and closing of grooves respectively). We show that small positive bending increases the power loss, while for negative bending the power loss experiences a minimum, corresponding to the best achievable waveguiding recovery because of the closing of grooves. Beyond certain deformation, the sensitivity of the POF sensor is dominated by losses in the un-grooved regions. A bending test for the mechanical integrity of grooved POFs shows that repetitive bending to a small radius results in a quick deterioration depending on the depth of the groove. However for bending radii more than 20mm and groove depths up to 0.3mm, the lifetime of the POF sensor is still acceptable for many targeted sensor application.While it is clear that the grooved POF is more sensitive to small bending compared to the un-grooved POF, the latter offers uncompromised mechanical integrity and is more desirable when the highest possible sensitivity is not a priority. Therefore, in the final stage of the work, both grooved and un-grooved POF are considered as candidates for developing the tomographic imaging modality. Further detail is considered in view of the targeted deliverable, a real PGPT system for footstep imaging, based on POF sensors. This has been successfully achieved by designing, integrating and testing two different PGPT systems: one based on grooved POF with sensor head size of 0.9m x 0.9m incorporating a multiplexed photodetector and another based on un-grooved POF with sensor head size of 2m x 1m, incorporating independent POF sensors. Both are capable of performing the real time imaging task as well as storing the numerical data for alternative processing. The image reconstruction is by applying a median-filtered Landweber method to solve the inverse tomography problem. The frame rate achieved is 2Hz and 1Hz with spatial resolution estimated as 10cm and 3cm for the first and second system respectively
|
|---|
