Subject Area
Electrical, Electronics Engineering
Abstract
There are numerous problems in the physical sciences wherein the scattering of light fundamentally limits our ability to perceive and reason about objects within the visual field. Examples include navigation in foggy weather, detecting hidden contraband/explosives, detecting diseased regions in agricultural produce, and detecting tumors within tissue. In all these scenarios, the scrambling of light paths due to scattering reduces the ability to locate embedded objects and identify their spatial or spectral features.
Existing approaches address the challenge by augmenting optical sensing modalities with complementary techniques that do not rely on optical phenomenology, such as Magnetic Resonance Imaging (MRI) or X-ray Imaging. However, these methods face certain limitations, such as the need for exogenous contrast agents or the use of ionizing radiation, which present challenges when applied to fields like food inspection or chemical detection. Additionally, in biomedical imaging, the increasing adoption of neural implant devices complicates the recovery of anatomical imagery using MRI methods. This prompted us to seek a solution that strictly relies on optical phenomenology to detect inhomogeneities and identify the spatial (or spectral features) over a wide field of regard. To achieve this objective, this work adopted a biologically inspired solution that integrates sensory modalities individually optimized for a specific task, dubbed multi-sensory integration.
The multi-sensory approach combines a wide-field coherent sensing method with a Frequency Modulated Continuous Wave (FMCW) lidar for this purpose. The wide field coherent sensing approach leverages the variations in the spectral evolution of speckle patterns to generate contrast between the inhomogeneous regions and background. These speckle patterns were produced by tuning the emission frequency of the laser and were recorded by a spatially dense Complementary Metal Oxide Semiconductor (CMOS) focal plane array (FPA). This work examined the statistical properties of the speckle pattern differences to devise a processing scheme that recovered high-contrast images from the speckle images. The insights from this analysis facilitated the adoption of bio-inspired neuromorphic sensors—which offer significant advantages in dynamic range and data throughput—for wide-field probing of scattering media. These techniques operated at thicknesses in tissue-like scattering media, significantly improving over the operation range of state-of-the-art CMOS detector-based diffuse imaging techniques.
The multisensory approach utilizes the wide-field topographic maps obtained from the neuromorphic camera to guide a point-scanning FMCW lidar system for retrieving higher-resolution images. Through this method, topographic maps of objects embedded inside an thick scattering medium , over a field-of-view of , were recovered with resolution. This capability—realized using strictly optical means—bypasses the need for prior anatomical information from MRI or precomputed atlases, enabling broader adoption in fields such as food processing and non-invasive chemical monitoring.
Finally, this work examined a scenario with added complexity arising from the absence of a wide-field sensing modality. A foveated sensing strategy that utilizes a physics-informed time-integrated contrast metric to intelligently allocate the sampling resources of an FMCW lidar was devised to address this scenario. Empirical results indicate an order of magnitude reduction in the overall measurements compared to traditional sampling techniques.
Degree Date
Summer 2024
Document Type
Dissertation
Degree Name
Ph.D.
Department
Electrical and Computer Engineering
Advisor
Prasanna Rangarajan
Number of Pages
165
Format
".pdf"
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
Recommended Citation
Madabhushi Balaji, Muralidhar, "Multisensory System for Probing Scattering Media" (2024). Electrical Engineering Theses and Dissertations. 81.
https://scholar.smu.edu/engineering_electrical_etds/81
