An Extended Spectral Angle Map for Hyperspectral and Multispectral Imaging

O'Sullivan, Jack; Hoy, Paul R.; Rutt, H.N.

doi:10.1364/cleo_si.2011.cmg2

Cited by 5 publications

(4 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to evaluate the performance of the proposed approach on the simulated data, three commonly employed metrics have been adopted: MPSNR (mean peak signal-to-noiseratio), MSSIM (mean structural similarity index), and MSA (mean spectral angle). These metrics calculate respectively the average of the PSNR (peak signal-to-noise-ratio), the SSIM (structural similarity index) [43] and the SA (spectral angle) [44] [45] in the spectral domain.…”

Section: B Accuracy Metricsmentioning

confidence: 99%

A Single Model CNN for Hyperspectral Image Denoising

Maffei

Haut

Paoletti

et al. 2020

IEEE Trans. Geosci. Remote Sensing

120

View full text Add to dashboard Cite

Denoising is a common pre-processing step prior to the analysis and interpretation of hyperspectral images (HSIs). However, the vast majority of methods typically adopted for HSI denoising exploit architectures originally developed for grayscale or RGB images, exhibiting limitations when processing highdimensional HSI data cubes. In particular, traditional methods do not take into account the high spectral correlation between adjacent bands in HSIs, which leads to unsatisfactory denoising performance as the rich spectral information present in HSIs is not fully exploited. To overcome this limitation, this paper considers deep learning models -such as convolutional neural networks (CNNs)-to perform spectral-spatial HSI denoising. The proposed model, called HSI single denosising CNN (HSI-SDeCNN), efficiently takes into consideration both the spatial and spectral information contained in HSIs. Experimental results on both synthetic and real data demonstrate that the proposed HSI-SDeCNN outperforms other state-of-the-art HSI denoising methods. Source code: https://github.com/mhaut/HSI-SDeCNN

show abstract

Section: B Accuracy Metricsmentioning

confidence: 99%

A Single Model CNN for Hyperspectral Image Denoising

Maffei

Haut

Paoletti

et al. 2020

IEEE Trans. Geosci. Remote Sensing

120

View full text Add to dashboard Cite

show abstract

“…The processing techniques specific to multi-and hyperspectral imaging are derived from methods developed for satellite imagery, among which are spectral angle mapping [19] and spectral correlation mapping [20]. There are classification methods for assessing the similarity of spectra to a reference used, for example, to identify spectra of areas, for example, forest or agricultural, or in the case of works of art to identify pigments.…”

Section: -2 -Multispectral Image Processingmentioning

confidence: 99%

“…Appendix C -Binning to increase the DR and SNR One advantage of having a hyper high-resolution image of 240,000,000 pixels with the multispectral camera is that it allows us to increase the dynamic range of the image by using the binning method [19]. Binning is usually performed on the sensor by combining the outputs of neighbouring photosites, or by mathematical algorithms.…”

Section: Graphical Abstractmentioning

confidence: 99%

Mona Lisa’s spolvero revealed

Cotte¹,

Simonot

2020

Journal of Cultural Heritage

View full text Add to dashboard Cite

“…It treats spectra as vectors in a highdimensional space. The angle between two spectral images can be calculated as in [62]. Smaller angles indicate higher spectral similarity.…”

Section: ) Denoising With Unsatisfactory Reference Featuresmentioning

confidence: 99%

Hyperspectral Image Denoising With Dual Deep CNN

Shan

Liu

et al. 2019

IEEE Access

View full text Add to dashboard Cite

A new hyperspectral image denoising algorithm, called the dual deep convolutional neural network (DD-CNN), is proposed in this paper. In contrast to internal denoising methods that utilize only the features from the target noisy image, the DD-CNN extensively explores the similarities between the target noisy image and the clean reference image from other bands. As external data, the reference images are selected based on the structural similarity index metric (SSIM). The DD-CNN is composed of two CNNs: one is responsible for extracting the features of the target image, and the other is responsible for extracting features from the reference image. A new activation function is proposed that activates the two types of features in the DD-CNN. Based on the dual structure and the new activation function, the external features extracted from the reference images are thoroughly integrated into the internal features of the target noise image. We experimented on different datasets with different noise levels; we also tested special cases for reference images with extra or undesirable features. The DD-CNN algorithm can effectively utilize the similarity between the external image and the target image. When the noise level is high, the advantages of the DD-CNN are obvious.INDEX TERMS Hyperspectral image denoising, deep dual neural network, feature learning, activation function.

show abstract

An Extended Spectral Angle Map for Hyperspectral and Multispectral Imaging

Abstract: Abstract:We present an extension to the widely used spectral angle metric, calculating an azimuthal angle around a reference vector. We demonstrate that it provides additional information, thus improving the classification ability of the spectral angle.

Cited by 5 publications

References 2 publications

A Single Model CNN for Hyperspectral Image Denoising

A Single Model CNN for Hyperspectral Image Denoising

Mona Lisa’s spolvero revealed

Hyperspectral Image Denoising With Dual Deep CNN

Contact Info

Product

Resources

About