High-resolution hyperspectral imaging via matrix factorization

By Rei Kawakami, Yasuyuki Matsushita, John Wright, Moshe Ben-Ezra, Yu-Wing Tai, Katsushi Ikeuchi. Computer Vision and Pattern Recognition, 2011

335 citations recorded by the source corpus.

Hyperspectral imaging is a promising tool for applications in geosensing, cultural heritage and beyond. However, compared to current RGB cameras, existing hyperspectral cameras are severely limited in spatial resolution. In this paper, we introduce a simple new technique for reconstructing a very high-resolution hyperspectral image from two readily obtained measurements: A lower-resolution hyper-spectral image and a high-resolution RGB image. Our approach is divided into two stages: We first apply an unmixing algorithm to the hyperspectral input, to estimate a basis representing reflectance spectra. We then use this representation in conjunction with the RGB input to produce the desired result. Our approach to unmixing is motivated by the spatial sparsity of the hyperspectral input, and casts the unmixing problem as the search for a factorization of the input into a basis and a set of maximally sparse coefficients. Experiments show that this simple approach performs reasonably well on both simulations and real data examples.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/0ecc4c66aced60ee633cfdb89486caebf8ec9713

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Hyperspectral Image Super-Resolution via Non-Negative Structured Sparse Representation (2016, methodology): “In [20] a sparse matrix factorization technique was proposed to decompose the LR hyperspectral image into a dictionary of basis vectors and a set of sparse coefficients.”
A Convex Formulation for Hyperspectral Image Superresolution via Subspace-Based Regularization (2014, background): “[10] fused hyperspectral images with images from RGB cameras,”
Light field video capture using a learning-based hybrid imaging system (2017, positive): “However, they are either used to increase the spatial resolution (Bhat et al. 2007; Boominathan et al. 2014; Sawhney et al. 2001; Wang et al. 2016b,a), deblur the image (Ben-Ezra and Nayar 2003), or create hyperspectral images (Cao et al. 2011; Kawakami et al. 2011).”
Hyperspectral Image Super-Resolution via Non-local Sparse Tensor Factorization (2017, methodology): “However, hyperspectral images (HSIs) usually have abundant spectral information, but limited spatial resolution due to hardware restrictions [13].”
Decomposable Nonlocal Tensor Dictionary Learning for Multispectral Image Denoising (2014, background): “As compared with the traditional image system, MSI helps to deliver more faithful representation for real scenes, and has been shown to greatly enhance the performance of various computer vision tasks, such as inpainting [8], superresolution [12] and tracking [21].”
Sparse Recovery of Hyperspectral Signal from Natural RGB Images (2016, methodology): “[18] suggested a matrix factorization method in order to obtain high resolution hyperspectral data from input that constitutes both a low resolution hyperspectral image and a high resolution RGB image.”