“Geometric model for an independently tilted lens and sensor with application for omnifocus imaging,” Indranil Sinharoy, Prasanna Rangarajan, Marc P. Christensen, OSA Applied Optics Special Issue on Modern Optics, 2017.
Optical imaging systems in which the lens and sensor are free to rotate about independent pivots offer greater degrees of freedom for controlling and optimizing the process of image gathering. However, to benefit from the expanded possibilities, we need an imaging model that directly incorporates the essential parameters. In this work, we propose a model of imaging which can accurately predict the geometric properties of the image in such systems. Furthermore, we introduce a new method for synthesizing an omnifocus (all-in-focus) image from a sequence of images captured while rotating a lens. The crux of our approach lies in insights gained from the new model.
INDEX TERMS — Image formation theory, Computational imaging, Multiframe image processing, Imaging systems.
“Active computational imaging for circumventing resolution limits at macroscopic scales,” Prasanna Rangarajan, Indranil Sinharoy, Marc P. Christensen, Predrag Milojkovic, OSA Applied Optics Special Issue on Modern Optics, 2017.
Macroscopic imagers are subject to constraints imposed by the wave nature of light and the geometry of image formation. The former limits the resolving power while the latter results in a loss of absolute size and shape information. The suite of methods outlined in this work enables macroscopic imagers the unique ability to capture unresolved spatial detail while recovering topographic information. The common thread connecting these methods is the notion of imaging under patterned illumination. The notion is advanced further to develop computational imagers with resolving power that is decoupled from the constraints imposed by the collection optics and the image sensor. These imagers additionally feature support for multiscale reconstruction.
INDEX TERMS — Optical Superresolution, Computational imaging, Multiscale reconstruction.
We present a simple technique for synthesizing an infinite depth of field image from a sequence of photographs captured while rotating a symmetric lens about the center of the entrance pupil. We discuss the feasibility conditions and provide a Zemax simulation that verifies the method.
INDEX TERMS — Depth of Field, Omnifocus image, all-in-focus image synthesis.
“Optical super resolution using a lattice of light spots,” Prasanna Rangarajan, Indranil Sinharoy, Marc P. Christensen, Predrag Milojkovic, OSA Topical meeting on Computational Imaging, 2014.
The paper outlines a super resolution strategy that overcomes the severe anisotropy in the resolving power of a single-lens imager, by processing images acquired under a lattice of light spots.
INDEX TERMS — Superresolution, Computational Imaging.
“A critical review of the slanted-edge method for color SFR measurement,” Prasanna Rangarajan, Indranil Sinharoy, Marc P. Christensen, Predrag Milojkovic, OSA Topical meeting on Imaging Systems & Applications, 2012.
Critical examination of the slanted-edge method for color SFR measurement reveals inaccuracies in the estimated SFR, due to the use of demosaicing. The proposed method resolves these inaccuracies by eliminating the need for demosaicing during SFR measurement.
INDEX TERMS — Slanted Edge, Color Filter Array, Digital cameras.
“Pushing the limits of digital imaging using Structured Illumination,” Prasanna Rangarajan, Indranil Sinharoy, Panos Papamichalis, Marc P. Christensen , Proc. 13th IEEE International Conference on Computer Vision, ICCV 2011.
The present work describes an active stereo apparatus that can not only recover scene geometry but also resolve spatial detail beyond the camera optical cutoff. The apparatus is comprised of a camera and a projector whose center-of-perspective is located in the camera pupil plane. The scene is illuminated with warped sinusoidal patterns as opposed to periodic or coded patterns.The findings reported in this work can help design imaging systems that feature improved optical resolution and 3D acquisition capabilities.
INDEX TERMS — OSR, Optical Superresolution, Structured Illumination, Depth maps.
“Space-Variant Optical Super-Resolution using Sinusoidal Illumination,” Prasanna Rangarajan, Vikrant R. Bhakta, Indranil Sinharoy, Manjunath Somayaji and Marc P. Christensen, Computational Optical Sensing and Imaging (COSI),Toronto, Canada, July, 2011.
The present work extends the scope of Optical Super-Resolution to imaging systems with spatially-varying blur, by using sinusoidal illumination. It also establishes that knowledge of the space-variant blur is not a pre-requisite for super-resolution.
INDEX TERMS — Superresolution, Computational Imaging, OSR, Space varying blur.
“Model-Based Region-Of-Interest estimation for adaptive resource allocation in multi-aperture imaging systems,” Indranil Sinharoy, Scott C. Douglas, Dinesh Rajan, Marc P. Christensen, IEEE Int. Conf. Acoust., Speech, Signal Processing (ICASSP), Las Vegas, Nevada,pp. 1961-1964, Apr. 2008.
Using intelligent resource allocation based on the information content in the imaging systems’ field-of-view for the successful design of a flat-profile multiplexed optical imaging system requires the use of adaptive techniques. This paper describes a model-based technique for determining regions of interest in aerial images using the 2D normalized power spectral density within Gilles’ saliency map estimator. The proposed technique exploits the spatial spectral shape of such natural imagery in a computationally-simple approach that is robust to additive noise. Application of the method to candidate aerial images shows its ability to identify consistent regions of interest for such data.
INDEX TERMS — Optical imaging, image region analysis, information theory, multisensor systems, spectral analysis.
“Region-of-interest estimation for adaptive resource allocation in multi-aperture imaging systems,” I. Sinharoy, S.C. Douglas, IEEE Int. Conf. Acoust., Speech, Signal Processing (ICASSP), Honolulu, HI, vol. 2, pp. 597-600, Apr. 2007.
Successful design of a flat-profile multiplexed optical imaging system requires the use of adaptive techniques to make intelligent resource allocation based on the information content in the imaging system’s field-of-view. This paper explores techniques for rending regions of interest in aerial images using local entropy as a descriptor. A novel method for identifying regions-of-interest in images is developed using the 2D normalized power spectral density within Gilles’ saliency map estimator. Application of the method to candidate aerial images shows its ability to identify consistent regions of interest for such data for varying block sizes and under additive noise.
INDEX TERMS — computational imaging systems, information theory, saliency, image reconstruction.
Despite the enormous success of iris recognition in close-range and well-regulated spaces for biometric authentication, it has hitherto failed to gain wide-scale adoption in less controlled, public environments. The problem arises from a limitation in imaging called the depth of field (DOF): the limited range of distances beyond which subjects appear blurry in the image. The loss of spatial details in the iris image outside the small DOF limits the iris image capture to a small volume–the capture volume. Existing techniques to extend the capture volume are usually expensive, computationally intensive, or afflicted by noise. Is there a way to combine the classical Scheimpflug principle with the modern computational imaging techniques to extend the capture volume? The solution we found is, surprisingly, simple; yet, it provides several key advantages over existing approaches. Our method, called Angular Focus Stacking (AFS), consists of capturing a set of images while rotating the lens, followed by registration, and blending of the in-focus regions from the images in the stack. The theoretical underpinnings of AFS arose from a pair of new and general imaging models we developed for Scheimpflug imaging that directly incorporates the pupil parameters. The model revealed that we could register the images in the stack analytically if we pivot the lens at the center of its entrance pupil, rendering the registration process exact. Additionally, we found that a specific lens design further reduces the complexity of image registration making AFS suitable for real-time performance. We have demonstrated up to an order of magnitude improvement in the axial capture volume over conventional image capture without sacrificing optical resolution and signal-to-noise ratio. The total time required for capturing the set of images for AFS is less than the time needed for a single-exposure, conventional image for the same DOF and brightness level. The net reduction in capture time can significantly relax the constraints on subject movement during iris acquisition, making it less restrictive.
Scaling down traditional optical imaging systems to enhance their form factor presents fundamental challenges in terms of loss of resolution and a decrease in the optical SNR due to the reduced light gathering ability of such scaled imagers. Computational imaging systems can address these issues through joint optimization of their optics and signal processing subsystems. One class of computational imagers is the thin, flat-profile multiplexed imaging system, which uses a combination of several scaled individual imaging units to capture a number of low-resolution images that are then digitally processed to reconstruct a high resolution version of the observed scene.
The performance of multiplexed imagers may be enhanced through the use of adaptive techniques wherein imager resource utilization is maximized through intelligent resource allocation based on the information content in the scene. The body of work laid out in this thesis describes techniques to find regions of interest in a scene and serves to enhance the efficiency of resource allocation in adaptive multiplexed imaging systems. The power spectral density (PSD) is used to derive local entropy maps of input scenes towards identification of regions of interest. Empirical evidence supporting the superiority of PSD-based saliency maps over their histogram-based counterparts in terms of relative local saliency representation of various regions within a scene is provided. Statistical analysis of noise in a scientific-grade digital camera shows that the noise power in images increased with pixel intensity indicating Poisson noise characteristics. A numerically fast and efficient method for computing model-based local saliency maps is presented, and its performance is evaluated in terms of the number of adders, multipliers and table lookups.
Non-peer reviewed papers, technical articles, pre-prints.
“Geometric model of image formation in Scheimpflug cameras,” Indranil Sinharoy, Prasanna Rangarajan, and Marc P. Christensen, PeerJ Preprints 4:e1887v1.
“Array ray tracing in Zemax OpticStudio from Python using the DDE extension,” Indranil Sinharoy, Knowledgebase article for Zemax.
“Talking to ZEMAX from Python using PyZDDE,” Indranil Sinharoy, Knowledgebase article for Zemax.
“Image Denoising Using Modified Neighshrink Algorithm,” Indranil Sinharoy, Tech. Report, Southern Methodist University 2012.