Thursday, March 13, 2014

Hyperspectral Imaging: Hardware and Data Reconstruction

Several papers and preprints on hyperspectral imaging from the hardware all the way to signal reconstruction.

First a presentation entitled: Sensing Sparse Spectral Signals by Using Compressive Sensing Techniques with Liquid Crystal Devices by Yitzhak August and Adrian Stern

Sparse signals are signals that have small amount of component, i.e., the signal can be presented by using small number of vectors. The interest in sparse signal has grown in the past decade. One of the theories in the filed called "compressed sensing" or "compressive sensing". Compressed Sensing) [1-3], (CS) is a fast emerging field in the area of digital signal sensing and processing. This theory provides a sensing framework for sampling sparse or compressible signals in a more efficient way that is usually done with Shannon-Nyquist sampling scheme. With CS, a compressed version of the signal is obtained already in the acquisition stage, thus reduce the need for digital compressing process. Since CS requires fewer measurements it can be applied to reduce the number of sensors or to reduce the acquisition time.
One natural implementation of the CS theory is in the field of spectroscopy or spectral imaging (e.g., hyperspectral imaging). One of the main limitations of classical hyperspectral imaging methods is the relatively slow scanning process. Other limitations arise from the fact that huge amounts of data needs to be processed and transmitted. Compressive sensing inspired methods can help handling these limitations [4-8]. In this work we present a new method for HS image acquisition using CS separable encoding both in the spatial and spectral domains our technique employing a liquid crystal phase retarder. The use of an LC retarder with known spectral response (spectrally calibrated cell), in conjunction with CS techniques, provides numerous benefits compared with conventional spectroscopic and spatially based CS methods. Specifically, the proposed method may facilitate: (a) reduction of acquisition time, (b) reduction of the number and the size of detectors, (c) reduction of system size and complexity, (d) reduction of system costs, (e) reduction of noise, and (f) reduction of power loss.

Abstract: This letter presents a new snapshot approach to hyper-spectral imaging via dual optical coding and compressive computational reconstruction. We demonstrate that two high speed spatial light modulators, located conjugate to the image and spectral plane, respectively, can code the hyper-spectral datacube into a single sensor image such that the high-resolution signal can be recovered in post-processing. We show various applications by designing different optical modulation functions, including programmable spatially-varying color filtering, multiplexed hyper-spectral imaging, and high-resolution compressive hyper-spectral imaging.

This paper addresses the problem of blind and fully constrained unmixing of hyperspectral images. Unmixing is performed without the use of any dictionary, and assumes that the number of constituent materials in the scene and their spectral signatures are unknown. The estimated abundances satisfy the desired sum-to-one and nonnegativity constraints. Two models with increasing complexity are developed to achieve this challenging task, depending on how noise interacts with hyperspectral data. The first one leads to a convex optimization problem, and is solved with the Alternating Direction Method of Multipliers. The second one accounts for signal-dependent noise, and is addressed with a Reweighted Least Squares algorithm. Experiments on synthetic and real data demonstrate the effectiveness of our approach.

Compressive Sensing (CS) is receiving increasing attention as a way to lower storage and compression requirements for on-board acquisition of remote-sensing images. In the case of multi- and hyperspectral images, however, exploiting the spectral correlation poses severe computational problems. Yet, exploiting such a correlation would provide significantly better performance in terms of reconstruction quality. In this paper, we build on a recently proposed 2D CS scheme based on blind source separation to develop a computationally simple, yet accurate, prediction-based scheme for acquisition and iterative reconstruction of hyperspectral images in a CS setting. Preliminary experiments carried out on different hyperspectral images show that our approach yields a dramatic reduction of computational time while ensuring reconstruction performance similar to those of much more complicated 3D reconstruction schemes.

Join the CompressiveSensing subreddit or the Google+ Community and post there !
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: