Thursday, March 22, 2012

Reconstruction of hidden 3D shapes using diffuse reflections and Flutter Shutter Video Camera for Compressive Sensing

Three additional papers gathered my interest this week when it came to compressive imaging (the first two are related). I just have a word for these: Wow.

We analyze multi-bounce propagation of light in an unknown hidden volume and demonstrate that the reflected light contains sufficient information to recover the 3D structure of the hidden scene. We formulate the forward and inverse theory of secondary and tertiary scattering reflection using ideas from energy front propagation and tomography. We show that using careful choice of approximations, such as Fresnel approximation, greatly simplifies this problem and the inversion can be achieved via a backpropagation process. We provide a theoretical analysis of the invertibility, uniqueness and choices of space-time-angle dimensions using synthetic examples. We show that a 2D streak camera can be used to discover and reconstruct hidden geometry. Using a 1D high speed time of flight camera, we show that our method can be used recover 3D shapes of objects "around the corner".

The recovery of objects obscured by scattering is an important goal in imaging and has been approached by exploiting, for example, coherence properties, ballistic photons or penetrating wavelengths. Common methods use scattered light transmitted through an occluding material, although these fail if the occluder is opaque. Light is scattered not only by transmission through objects, but also by multiple reflection from diffuse surfaces in a scene. This reflected light contains information about the scene that becomes mixed by the diffuse reflections before reaching the image sensor. This mixing is difficult to decode using traditional cameras. Here we report the combination of a time-of-flight technique and computational reconstruction algorithms to untangle image information mixed by diffuse reflection. We demonstrate a three-dimensional range camera able to look around a corner using diffusely reflected light that achieves sub-millimetre depth precision and centimetre lateral precision over 40 cm×40 cm×40 cm of hidden space.
Thanks Sylvain for the heads up. Let me note that the team used CoSAMP and SPGL1 and that CoSAMP seems to be working better. I wonder where that comes from ? maybe because CoSAMP is solving a different problem than an L1 minimization. I also note that this is the second wow of the week, and that in both instances, the solvers used are great but improvement could probably be gotten using some of the newer and faster ones.

Video cameras are invariably bandwidth limited and this results in a trade-off between spatial and temporal resolution. Advances in sensor manufacturing technology have tremendously increased the available spatial resolution of modern cameras while simultaneously lowering the costs of these sensors. In stark contrast, hardware improvements in temporal resolution have been modest. One solution to enhance temporal resolution is to use high bandwidth imaging devices such as high speed sensors and camera arrays. Unfortunately, these solutions are expensive. An alternate solution is motivated by recent advances in computational imaging and compressive sensing. Camera designs based on these principles, typically, modulate the incoming video using spatio-temporal light modulators and capture the modulated video at a lower bandwidth. Reconstruction algorithms, motivated by compressive sensing, are subsequently used to recover the high bandwidth video at high fidelity. Though promising, these methods have been limited since they require complex and expensive light modulators that make the techniques difficult to realize in practice. In this paper, we show that a simple coded exposure modulation is sufficient to reconstruct high speed videos. We propose the Flutter Shutter Video Camera (FSVC) in which each exposure of the sensor is temporally coded using an independent pseudo-random sequence. Such exposure coding is easily achieved in modern sensors and is already a feature of several machine vision cameras. We also develop two algorithms for reconstructing the high speed video; the first based on minimizing the total variation of the spatiotemporal slices of the video and the second based on a data driven dictionary based approximation. We perform evaluation on simulated videos and real data to illustrate the robustness of our system.

Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.

No comments: