Unsupervised Hyperspectral Image Band Selection Based on Deep Subspace Clustering

Zeng, Meiqin; Cai, Yaoming; Cai, Zhihua; Liu, Xiaobo; Hu, Peng; Ku, Junhua

doi:10.1109/lgrs.2019.2912170

Cited by 48 publications

(22 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Finally, the evaluation of the model is conducted by checking its performance when unseen samples are processed, measuring its ability to generalize the information learned during the training and adjustment phase. In this work, depending on the knowledge about the data employed during the third stage, we consider three main kinds of algorithms: 1) Unsupervised learning [112], [113], in which the classes are not known in advance; namely: only the characteristics of the data are known, but not the existing classes, and, therefore, it is the algorithm itself that has to recognize the patterns and group them according to their characteristics. This means that the algorithm creates the classes.…”

Section: B Machine Learning To Process Rs Imagesmentioning

confidence: 99%

Analysis of Remotely Sensed Images Through Social Media

Redondo

Haut

Paoletti

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

During the past decades, the volume of big data available in remote sensing (RS) applications has grown significantly. In addition, a number of applications related to monitoring human activity are being developed based on this kind of data. This has considerably increased the demand for RS processing methods. In this sense, the scientific community is facing the challenge of how to maximize the potential of the data that are produced in a fast and efficient way. In particular, the provision of processing algorithms that can be developed in an easy way is a fundamental problem for the RS community, due to the large volume of data offered by different portals and agencies, and the need for algorithms developed on different platforms and using a variety of programming frameworks. To address these challenges, this paper takes advantage of social media tools to bring images and algorithms closer to users. In particular, a new system based on the Telegram messaging application Bot (called @ThuleRSbot) has been developed to provide a wide variety of RS image processing methods to users under the same interface. The system has been developed using the Python language and the Telegram Bot application program interface (API). The most remarkable characteristics of the system are: 1) a completely open architecture that facilitates the incorporation of new algorithms without effort, and 2) an easy way to automatically gather satellite images from the Sentinel Hub platform. Code available on: https://github.com/mhaut/ThuleRSbot.

show abstract

Section: B Machine Learning To Process Rs Imagesmentioning

confidence: 99%

Analysis of Remotely Sensed Images Through Social Media

Redondo

Haut

Paoletti

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Mixture, connectivity and sparsity are the three major directions involved in the regularization term design [16]. The sparse representation coefficient matrix is computed with the regularization terms based on L0 and L1.…”

Section: Related Workmentioning

confidence: 99%

Hyperspectral Image Processing in Internet of Things model using Clustering Algorithm

Bindhu¹,

Ranganathan²

2021

JISMAC

View full text Add to dashboard Cite

With the advent of technology, several domains have b on Internet of Things (IoT). The hyper spectral sensors present in earth observation system sends hyper spectral images (HSIs) to the cloud for further processing. Artificial intelligence (AI) models are used to analyse data in edge servers, resulting in a faster response time and reduced cost. Hyperspectral images and other high-dimensional image data may be analysed by using a core AI model called subspace clustering. The existing subspace clustering algorithms are easily affected by noise since they are constructed based on a single model. The representation coefficient matrix connectivity and sparsity is hardly balanced. In this paper, connectivity and sparsity factors are considered while proposing the subspace clustering algorithm with post-process strategy. A non-dominated sorting algorithm is used for that selection of close neighbours that are defined as neighbours with high coefficient and common neighbours. Further, pruning of useless, incorrect or reserved connections based on the coefficients between the close and sample neighbours are performed. Lastly, inter and intra subspace connections are reserved by the post-process strategy. In the field of IoT and image recognition, the conventional techniques are compared with the proposed post-processing strategies to verify its effectiveness and universality. The clustering accuracy may be improved in the IoT environment while processing the noise data using the proposed strategy as observed in the experimental results.

show abstract

“…Unlike feature extraction, which transforms spectral information through linear or nonlinear operations, BS not only preserves the physical property contained in HSIs [5] but reduces the dimensionality of HSI. In recent years, plenty of BS methods have been proposed [8], [2], [11], [12] and can basically be divided into three categories: rankingbased [13], [2], clustering-based [14], [7], and searchingbased [15], [5] methods. The ranking-based method aims to rank each spectral band according to the estimation of its significance or information [16], [17].…”

Section: Introductionmentioning

confidence: 99%

“…Recently, many works have devoted to utilizing the self-representation model for BS of HSI, e.g., [27], [7], [24]. Most of these methods are relying on subspace clustering [7], [12], [11]. Despite the impressive performance, the traditional self-representation models often fail to exploit the structural relationships of spectral bands.…”

Section: Introductionmentioning

confidence: 99%

Efficient Graph Convolutional Self-Representation for Band Selection of Hyperspectral Image

Cai

Zhang

Liu

et al. 2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Hyperspectral image (HSI) band selection (BS) is an important task for HSI dimensionality reduction, whose goal is to select an informative band subset containing less redundancy. However, traditional BS methods basically work in the Euclidean domain and thus often neglect to consider the structural information of spectral bands. In this paper, to make full use of the structural information, a novel BS method termed as Efficient Graph Convolutional Self-Representation (EGCSR) is proposed by incorporating graph convolution into the selfrepresentation model. Since the proposed method is typically modeled in the non-Euclidean domain, it tends to result in a more robust self-representation coefficient matrix. We provide a closed-form solution to the EGCSR model, which leads to high computational efficiency. We further propose two strategies to determine the informative band subset from the coefficient matrix. The first is a ranking-based strategy (EGCSR-R), which ranks every band by calculating the cumulative contribution, and the second is a clustering-based strategy (EGCSR-C), which treats band selection as a band clustering task based on using subspace segmentation. Extensive experimental results on three real HSI datasets show that the proposed EGCSR model is dramatically superior to many existing band selection methods and with high computational efficiency.

show abstract

Unsupervised Hyperspectral Image Band Selection Based on Deep Subspace Clustering

Cited by 48 publications

References 16 publications

Analysis of Remotely Sensed Images Through Social Media

Analysis of Remotely Sensed Images Through Social Media

Hyperspectral Image Processing in Internet of Things model using Clustering Algorithm

Efficient Graph Convolutional Self-Representation for Band Selection of Hyperspectral Image

Contact Info

Product

Resources

About