Congratulations to Shu for defending his dissertation!
Scatter Based Novel Imaging Systems
Abstract
Direct imaging refers to an image formation process where an object’s or scene’s optical properties can be spatially/temporally mapped using direct (isomorphic) interaction with ambient or active illumination. However, in many scenarios direct imaging is not feasible or possible, forcing us to utilize scattered photons that only indirectly interact with an object/scene. In this work, we explore two special cases of indirect imaging and propose novel scattering-based imaging approaches in visible and X-ray spectral bands. The first novel indirect imaging modality targets non-line-of-sight (NLOS) scenes. The existing time-resolved NLOS technologies predominately usually involve emission and detection of photons using ultra-fast laser and high-speed detectors. Our approach explores the limits of NLOS imaging using only ambient/passive light sources for illumination. Imaging is achieved through analysis of scattered photons on the visible wall. We refer this new technique as “Passive NLOS.” The second scatter-based imaging approach entails a novel X-ray diffraction (coherent scatter) tomography reconstruction algorithm. X-ray diffraction tomography gathers off-plane Rayleigh scattered X-ray photons and reconstructs a spatial map of material form factor. Existing reconstruction algorithms utilize either maximum likelihood (ML) estimation or maximum-a-posteriori (MAP) estimation with spatially smooth prior. In this study, we explore a new reconstruction algorithm that exploits the sparsity in the material form factors based on the physics of the underlying scattering mechanism. The proposed algorithm employs Expectation-Maximization approach incorporating Group TV spatial regularization and L1-norm spectral/momentum regularization.
Committee:
- Dr. Amit Ashok (Chair)
- Dr. Stanley Pau
- Dr. Leilei Peng
Next post
Congratulations to Kwan for defending his dissertation!