Hyperspectral Image Analysis in Environmental Monitoring: Setup of a New Tunable Filter Platform

Moroni, Monica; Lupo, Emanuela; Marra, Emanuela; Cenedese, A.

doi:10.1016/j.proenv.2013.06.098

Cited by 44 publications

(22 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Food fraud detection [85,109] Object identification and feather extraction of Earth's surface in remote sensing [2] Viability prediction of seeds [78] Monitoring and tracking changes in environmental conditions [75] Target recognition and tracking [79] Identification of illegal/fake drugs [49] Evidence analysis in forensic investigation [47] Quality control of thin-film [34] Exploration of oil and gas [22] Creating vegetation index of agricultural data [2,3,106] A hyperspectral camera can detect fake food where impurities are mixed with food items that have a similar color and texture. In the visible spectrum, such types of fake food items are not identifiable, while a hyperspectral camera catches them very easily due to the impurities showing different chemical properties.…”

Section: Exploration Of Hyperspectral Imagesmentioning

confidence: 99%

Current research opportunities of image processing and computer vision

Gupta

2019

csci

View full text Add to dashboard Cite

Image processing and computer vision is an important and essential area in today’s scenario. Several problems can be solved through computer vision techniques. There are a large number of challenges and opportunities which require skills in the field of computer vision to address them. Computer vision applications cover each band of the electromagnetic spectrum and there are numerous applications in every band. This article is targeted to the research students, scholars and researchers who are interested to solve the problems in the field of image processing and computer vision. It addresses the opportunities and current trends of computer vision applications in all emerging domains. The research needs are identified through available literature survey and classified in the corresponding domains. The possible exemplary images are collected from the different repositories available for research and shown in this paper. The opportunities mentioned in this paper are explained through the images so that a naive researcher can understand it well before proceeding to solve the corresponding problems. The databases mentioned in this article could be useful for researchers who are interested in further solving the problem. The motivation of the article is to expose the current opportunities in the field of image processing and computer vision along with corresponding repositories. Interested researchers who are working in the field can choose a problem through this article and can get the experimental images through the cited references for working further.

show abstract

Section: Exploration Of Hyperspectral Imagesmentioning

confidence: 99%

Current research opportunities of image processing and computer vision

Gupta

2019

csci

View full text Add to dashboard Cite

show abstract

“…H YPERSPECRAL image (HSI) is generally regarded as a data cube, which provides abundant spectral information, spatial information, and radiation information. In view of this advantage, HSIs are widely applied to agriculture [1], [2], environmental monitoring [3], [4], target detection [5], [6], and more. However, there exists a trade-off between the spatial resolution and spectral resolution physically.…”

Section: Introductionmentioning

confidence: 99%

Bidirectional 3D Quasi-Recurrent Neural Network for Hyperspectral Image Super-Resolution

Liang

You

2021

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

View full text Add to dashboard Cite

Hyperspectral imaging is unable to acquire images with high resolution in both spatial and spectral dimensions yet due to physical hardware limitations. It can only produce low spatial resolution images in most cases and thus hyperspectral image (HSI) spatial super-resolution is important. Recently, deep learning based methods for HSI spatial super-resolution have been actively exploited. However, existing methods do not focus on structural spatial-spectral correlation and global correlation along spectra, which cannot fully exploit useful information for super-resolution. Also, some of the methods are straight forward extension of RGB super-resolution methods, which have fixed number of spectral channels and cannot be generally applied to hyperspectral images whose number of channels varies. Furthermore, unlike RGB images, existing HSI datasets are small and limit the performance of learning based methods. In this paper, we design a bidirectional 3D quasi-recurrent neural network for HSI super-resolution with arbitrary number of bands. Specifically, we introduce a core unit that contains a 3D convolutional module and a bidirectional quasi-recurrent pooling module to effectively extract structural spatial-spectral correlation and global correlation along spectra, respectively. By combining domain knowledge of HSI with a novel pre-training strategy, our method can be well generalized to remote sensing HSI datasets with limited number of training data. Extensive evaluations and comparisons on HSI super-resolution demonstrate improvements over state-of-the-art methods, in terms of both restoration accuracy and visual quality.

show abstract

“…The wealthy information in hyperspectral images (HSIs) [1] guarantees the superiority in material recognition and object detection. Thus, hyperspectral data has been applied to a wide range of fields, such as agriculture [2], environmental monitoring [3], food safety [4], mineralogy [5], and surveillance [6]. Traditional machine learning methods employed in HSI classification are based on the fact that different materials exhibit different spectral reflective curves and process each pixel independently.…”

Section: Introductionmentioning

confidence: 99%

Adaptive Spatial-Spectral Feature Learning for Hyperspectral Image Classification

Zhu

Liu³

et al. 2019

IEEE Access

View full text Add to dashboard Cite

The combination of spectral and spatial information provides an effective way to improve the hyperspectral image (HSI) classification. However, the local spatial contexture changes with different neighborhood regions over the HSI image plane, and methods with fixed weights to integrate spatial information for all neighborhood regions could result in inaccurate spatial features, leading to adverse effects on classification performance. To address this issue, a novel adaptive spatial-spectral feature learning network (ASSFL) has been proposed to reflect spatial contexture changes and learn robust adaptive features in this paper. In the implementation of the proposed method, a convolution neural network (CNN) is first applied to learn weight features for each pixel within a hyperspectral patch and adaptive weights can be obtained based on a softmax normalization. Then, the shallow joint adaptive features can be acquired according to these weights. After that, a stacked auto-encoder (SAE) is proposed to further extract deeper hierarchical features for the final classification. The experimental results on four benchmark HSI data sets demonstrate that the proposed method can achieve competitive classification results compared with other existing classifiers. INDEX TERMS Adaptive spatial-spectral features, convolutional neural network (CNN), hyperspectral image (HSI) classification, stacked auto-encoder (SAE).

show abstract

Hyperspectral Image Analysis in Environmental Monitoring: Setup of a New Tunable Filter Platform

Cited by 44 publications

References 10 publications

Current research opportunities of image processing and computer vision

Current research opportunities of image processing and computer vision

Bidirectional 3D Quasi-Recurrent Neural Network for Hyperspectral Image Super-Resolution

Adaptive Spatial-Spectral Feature Learning for Hyperspectral Image Classification

Contact Info

Product

Resources

About