Scalable One-Pass Self-Representation Learning for Hyperspectral Band Selection

Wei, Xiaohui; Zhu, Wen; Liao, Bo; Cai, Liang

doi:10.1109/tgrs.2019.2890848

Cited by 37 publications

(15 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To reduce the redundancy of the HSI, we construct sub-HSIs as the input samples by using correlation [32] between bands. Figure 9a is a visualization of the correlation coefficient matrix.…”

Section: Feature Extractionmentioning

confidence: 99%

“…RSSC kernels can be initialized directly in the first frame and then updated in subsequent frames without offline iterative training of large dataset to extract discriminative spatialspectral features of a HSI in real-time. The redundancy of HSI is reduced by dividing it into sub-HSIs using band correlation [32], and the weights of each sub-HSI to tracking are expressed by relative entropy. Specifically, RSSC kernels are first initialized from a set of sub-HSIs obtained from the initial frame.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Object Tracking in Hyperspectral-Oriented Video with Fast Spatial-Spectral Features

et al. 2021

View full text Add to dashboard Cite

This paper presents a correlation filter object tracker based on fast spatial-spectral features (FSSF) to realize robust, real-time object tracking in hyperspectral surveillance video. Traditional object tracking in surveillance video based only on appearance information often fails in the presence of background clutter, low resolution, and appearance changes. Hyperspectral imaging uses unique spectral properties as well as spatial information to improve tracking accuracy in such challenging environments. However, the high-dimensionality of hyperspectral images causes high computational costs and difficulties for discriminative feature extraction. In FSSF, the real-time spatial-spectral convolution (RSSC) kernel is updated in real time in the Fourier transform domain without offline training to quickly extract discriminative spatial-spectral features. The spatial-spectral features are integrated into correlation filters to complete the hyperspectral tracking. To validate the proposed scheme, we collected a hyperspectral surveillance video (HSSV) dataset consisting of 70 sequences in 25 bands. Extensive experiments confirm the advantages and the efficiency of the proposed FSSF for object tracking in hyperspectral video tracking in challenging conditions of background clutter, low resolution, and appearance changes.

show abstract

Section: Feature Extractionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Object Tracking in Hyperspectral-Oriented Video with Fast Spatial-Spectral Features

et al. 2021

View full text Add to dashboard Cite

show abstract

“…In fact, the spectral bands of HSI are highly redundant, so, we do not need to utilize full bands of HSIs. Dimensionality reduction is an effective preprocessing step to discard the redundant information [52][53][54]. Therefore, instead of using full bands we select part of the bands for spatial enhancement through the a parameter.…”

Section: Resblock2mentioning

confidence: 99%

Hyperspectral Image Super-Resolution with 1D–2D Attentional Convolutional Neural Network

Cui

et al. 2019

Remote Sensing

View full text Add to dashboard Cite

Hyperspectral image (HSI) super-resolution (SR) is of great application value and has attracted broad attention. The hyperspectral single image super-resolution (HSISR) task is correspondingly difficult in SR due to the unavailability of auxiliary high resolution images. To tackle this challenging task, different from the existing learning-based HSISR algorithms, in this paper we propose a novel framework, i.e., a 1D-2D attentional convolutional neural network, which employs a separation strategy to extract the spatial-spectral information and then fuse them gradually. More specifically, our network consists of two streams: a spatial one and a spectral one. The spectral one is mainly composed of the 1D convolution to encode a small change in the spectrum, while the 2D convolution, cooperating with the attention mechanism, is used in the spatial pathway to encode spatial information. Furthermore, a novel hierarchical side connection strategy is proposed for effectively fusing spectral and spatial information. Compared with the typical 3D convolutional neural network (CNN), the 1D-2D CNN is easier to train with less parameters. More importantly, our proposed framework can not only present a perfect solution for the HSISR problem, but also explore the potential in hyperspectral pansharpening. The experiments over widely used benchmarks on SISR and hyperspectral pansharpening demonstrate that the proposed method could outperform other state-of-the-art methods, both in visual quality and quantity measurements. a feasible scheme is to increase the pixel spatial size. However, because of fundamental physical limits in practice, it is difficult to improve the sensor capability. Consequently, compared with conventional RGB or multispectral cameras, the obtained hyperspectral image (HSI) is always with a relative low spatial resolution, which limits their practical applications. In recent years, HSI super-resolution (SR) has attracted more and more attention in the remote sensing community.HSI SR is a promising signal post-processing technique aiming at acquiring a high resolution (HR) image from its low resolution (LR) version to overcome the inherent resolution limitations [1]. Generally, we can roughly divide this technique into two categories, according to the availability of an HR auxiliary image, e.g., the single HSI super-resolution or pan sharpening methods with the HR panchromatic image. For example, the popular single HSI super-resolution methods are bilinear [2] and bicubic interpolation [3] based on interpolation, and [4,5] based on regularization. Pansharpening methods can be roughly divided into five categories: component substitution (CS) [6][7][8], which may cause spectral distortion; multiresolution analysis (MRA) [9][10][11][12], which can keep spectral consistency at the cost of much computation and great complexity of parameter setting; bayesian methods [13][14][15] and matrix factorization [16], which can achieve prior spatial and spectral performance at a very high computational cost; and hybrid metho...

show abstract

“…The virtual dimensionality (VD) of this scene was estimated by the Harsanyi-Farrand-Chang (HFC) method in [21][22][23] as 9. However, according to [24,25], n BS = 9 seemed insufficient, because when the automatic target generation process (ATGP) developed in [26] was used to find target pixels, only three panel pixels could be found among nine ATGP-found target pixels.…”

Section: Linear Spectral Unmixingmentioning

confidence: 99%

“…However, how to appropriately choose a threshold is challenging, because this threshold is related to how close the correlation is. Recently, with the prevalence of matrix computing, band selection has been transformed into a matrix-based optimization problem reflecting the representativeness of bands from different perspectives-one study [22] formulated band selection into a low-rank-based representation model to define the affinity matrix of bands for band selection via rank minimization, and reference [23] presented a scalable one-pass self-representation learning for hyperspectral band selection. The second issue is that a BP criterion may be good for one application but may not be good for another application.…”

Section: Introductionmentioning

confidence: 99%

Fusion of Various Band Selection Methods for Hyperspectral Imagery

Wang

Xie

et al. 2019

Remote Sensing

View full text Add to dashboard Cite

This paper presents an approach to band selection fusion (BSF) which fuses bands produced by a set of different band selection (BS) methods for a given number of bands to be selected, nBS. Since each BS method has its own merit in finding the desired bands, various BS methods produce different band subsets with the same nBS. In order to take advantage of these different band subsets, the proposed BSF is performed by first finding the union of all band subsets produced by a set of BS methods as a joint band subset (JBS). Due to the fact that a band selected by one BS method in JBS may be also selected by other BS methods, in this case each band in JBS is prioritized by the frequency of the band appearing in the band subsets to be fused. Such frequency is then used to calculate the priority probability of this particular band in the JBS. Because the JBS is obtained by taking the union of all band subsets, the number of bands in the JBS is at least equal to or greater than nBS. So, there may be more than nBS bands, in which case, BSF uses the frequency-calculated priority probabilities to select nBS bands from JBS. Two versions of BSF, called progressive BSF and simultaneous BSF, are developed for this purpose. Of particular interest is that BSF can prioritize bands without band de-correlation, which has been a major issue in many BS methods using band prioritization as a criterion to select bands.

show abstract

Scalable One-Pass Self-Representation Learning for Hyperspectral Band Selection

Cited by 37 publications

References 49 publications

Object Tracking in Hyperspectral-Oriented Video with Fast Spatial-Spectral Features

Object Tracking in Hyperspectral-Oriented Video with Fast Spatial-Spectral Features

Hyperspectral Image Super-Resolution with 1D–2D Attentional Convolutional Neural Network

Fusion of Various Band Selection Methods for Hyperspectral Imagery

Contact Info

Product

Resources

About