The computational image systems evaluation toolbox (CISET) : a software environment for rapid prototyping and evaluating computational imaging hardware and algorithms
- The Computational Image Systems Evaluation Toolbox (CISET) is a software simulation that models the transformation from three-dimensional spectral scene radiance, through multicomponent spherical lenses, to spectral irradiance, sensor responses, and computational imaging algorithms. This toolbox is designed to help the engineer and scientist understand the impact of many important factors that determine the performance of computational imaging systems such as light field cameras and depth estimation systems. The ability to methodically control scene and imaging parameters, which are difficult to control in the real world, helps us understand and design the combination of hardware and software efficiently. The optics tool, which can be used to characterize lenses, models a series of spherical surfaces, each with its own wavelength-dependent index of refraction. CISET accounts for the aperture, which helps characterize depth of field effects and models the effects of lens diffraction using two different methods. Moreover, CISET is computationally efficient so that the designer can interactively adjust parameters and visualize the consequences. Computational imaging systems engineers looking to design and evaluate new systems can easily change the scene geometry, surface reflectance, scene lighting, lens, sensor, and image processing parameters. This controlled environment allows for designers to quickly evaluate engineering trade-offs, isolate image systems parameters, and compare computational imaging results to ground truth information. We used CISET to simulate and explore the compliance range of several famous computational imaging systems, such as a flash/no-flash detail transfer system, a flash/no-flash illuminant estimation system, and depth reconstruction from a stereo image pair. We have also built a full pipeline spectral simulation for a light field camera, which measures light intensity and direction due to a microlens array placed in between the main lens and the imaging sensor. In this case, we utilized CISET to explore different engineering trade-offs between spatial and angular capture resolution of the light field, and evaluated the effects of diffraction due to the microlenses on the performance of the system. Furthermore, we used CISET to prototype a novel 2-flash depth estimation algorithm. CISET also includes optics tools for rendering and analyzing the point spread function (PSF). By storing light field information of a PSF in a plenoptic PSF (pPSF), sensor distances, aperture sizes, and apodization can be changed without recomputation of the PSF. The transform of a lens can also be summarized as a volume of linear transforms (VoLT). VoLT achieves similar benefits as the pPSF, allowing for PSFs with different apertures, sensor distances, and apodization to be computed quickly and efficiently.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Lin, Andy Lai
|Stanford University, Department of Electrical Engineering.
|Wandell, Brian A
|Wandell, Brian A
|Farrell, Joyce E
|Farrell, Joyce E
|Statement of responsibility
|Andy Lai Lin.
|Submitted to the Department of Electrical Engineering.
|Thesis (Ph.D.)--Stanford University, 2015.
- © 2015 by Andy Lai Lin
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...