A graph Laplacian regularization for hyperspectral data unmixing

Ammanouil, Rita; Ferrari, A.; Richard, Cédric

doi:10.1109/icassp.2015.7178248

Cited by 33 publications

(27 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both simulated and real hyperspectral data are generated to evaluate the effectiveness of the proposed algorithms. In the synthetic data experiments, we compare our method with some representative methods: (1) SUnSAL [21]; (2) CLSUnSAL [22]; (3) DRSU-TV [4]; (4) GLUP-Lap [28]; (5) Graph-regularized unmixing method (without sparse constraint on abundances), abbreviated as GraphHU; and (6) Sparse graph-regularized unmixing method (set W 1 = W 2 as all ones matrices in problem (P2)), abbreviated as SGHU. For quantitative comparison, the signal-to-reconstruction error (SRE) is introduced to measure the quality of the estimated fractional abundances of endmembers.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Double Reweighted Sparse Regression and Graph Regularization for Hyperspectral Unmixing

et al. 2018

View full text Add to dashboard Cite

Hyperspectral unmixing, aiming to estimate the fractional abundances of pure spectral signatures in each mixed pixel, has attracted considerable attention in analyzing hyperspectral images. Plenty of sparse unmixing methods have been proposed in the literature that achieved promising performance. However, many of these methods overlook the latent geometrical structure of the hyperspectral data which limit their performance to some extent. To address this issue, a double reweighted sparse and graph regularized unmixing method is proposed in this paper. Specifically, a graph regularizer is employed to capture the correlation information between abundance vectors, which makes use of the property that similar pixels in a spectral neighborhood have higher probability to share similar abundances. In this way, the latent geometrical structure of the hyperspectral data can be transferred to the abundance space. In addition, a double weighted sparse regularizer is used to enhance the sparsity of endmembers and the fractional abundance maps, where one weight is introduced to promote the sparsity of endmembers as a hyperspectral image typically contains fewer endmembers compared to the overcomplete spectral library and the other weight is exploited to improve the sparsity of the abundance matrix. The weights of the double weighted sparse regularizer used for the next iteration are adaptively computed from the current abundance matrix. The experimental results on synthetic and real hyperspectral data demonstrate the superiority of our method compared with some state-of-the-art approaches.

show abstract

Section: Resultsmentioning

confidence: 99%

“…These methods ignore the underlying structure of the hyperspectral images. To take advantage of this property, some graph-based methods are proposed [28,29], which employ the graph topology and sparse group lasso regularization.…”

Section: Introductionmentioning

confidence: 99%

Double Reweighted Sparse Regression and Graph Regularization for Hyperspectral Unmixing

et al. 2018

View full text Add to dashboard Cite

show abstract

“…depth, intensity), the goal being to separate the large data sets into small regions that can be processed independently which reduces the computational cost. For this, the proposed method builds on the work [33] that evaluated a graph of similarity for the image pixels, then used it to perform a clustering step by considering the method described in [34]. A direct application of this approach can be done by considering the graph-nodes as the pixels' temporal responses, and their similarities to form the graph's edges.…”

Section: Clustering-based Image Restorationmentioning

confidence: 99%

High-resolution depth profiling using a range-gated CMOS SPAD quanta image sensor

Ren¹,

Connolly²,

Halimi³

et al. 2018

Opt. Express

View full text Add to dashboard Cite

A CMOS single-photon avalanche diode (SPAD) quanta image sensor is used to reconstruct depth and intensity profiles when operating in a range-gated mode used in conjunction with pulsed laser illumination. By designing the CMOS SPAD array to acquire photons within a pre-determined temporal gate, the need for timing circuitry was avoided and it was therefore possible to have an enhanced fill factor (61% in this case) and a frame rate (100,000 frames per second) that is more difficult to achieve in a SPAD array which uses time-correlated single-photon counting. When coupled with appropriate image reconstruction algorithms, millimeter resolution depth profiles were achieved by iterating through a sequence of temporal delay steps in synchronization with laser illumination pulses. For photon data with high signal-to-noise ratios, depth images with millimeter scale depth uncertainty can be estimated using a standard cross-correlation approach. To enhance the estimation of depth and intensity images in the sparse photon regime, we used a bespoke clustering-based image restoration strategy, taking into account the binomial statistics of the photon data and non-local spatial correlations within the scene. For sparse photon data with total exposure times of 75 ms or less, the bespoke algorithm can reconstruct depth images with millimeter scale depth uncertainty at a stand-off distance of approximately 2 meters. We demonstrate a new approach to single-photon depth and intensity profiling using different target scenes, taking full advantage of the high fill-factor, high frame rate and large array format of this range-gated CMOS SPAD array.

show abstract

“…Similarity search is a fundamental pre-processing method for many applications of hyperspectral remote sensing data sets (HRD), such as manifold learning-based dimensionality reduction (Gao et al 2015), graph-based classification (Bai, Xiang, and Pan 2013), spectral un-mixing (Ammanouil, Ferrari, and Richard 2015), etc. Exhaustive similarity search is a general problem-solving technique, but it will soon become the computation-bottleneck of the above applications for processing the large-scale data set.…”

Section: Introductionmentioning

confidence: 99%

A novel locality-sensitive hashing algorithm for similarity searches on large-scale hyperspectral data

Zhou

Liu

et al. 2016

Remote Sensing Letters

View full text Add to dashboard Cite

2016) A novel locality-sensitive hashing algorithm for similarity searches on large-scale hyperspectral data, Remote Sensing Letters, 7:10, 965-974, ABSTRACT Similarity search is a fundamental process in many hyperspectral remote sensing applications. In this article, we investigated the locality-sensitive hashing (LSH) algorithm for approximate similarity search on hyperspectral remote sensing data and proposed a new method that uses a regulated random hyperplane projection hash function family to index data and an adjacent graph-probing method for similarity queries. This method improves the performance of LSH over non-uniformly distributed hyperspectral data sets. Comparative experiments with three benchmark hyperspectral data sets showed that the proposed method is at least two times faster than the basic LSH and two other improved methods whilst achieving the same search accuracy. Moreover, results of the proposed method are proven equivalent to the accurate similarity search for real applications. ARTICLE HISTORY

show abstract

A graph Laplacian regularization for hyperspectral data unmixing

Cited by 33 publications

References 28 publications

Double Reweighted Sparse Regression and Graph Regularization for Hyperspectral Unmixing

Double Reweighted Sparse Regression and Graph Regularization for Hyperspectral Unmixing

High-resolution depth profiling using a range-gated CMOS SPAD quanta image sensor

A novel locality-sensitive hashing algorithm for similarity searches on large-scale hyperspectral data

Contact Info

Product

Resources

About