Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071812
Aihua Li, Ryan Will, N. Glenn, S. Benner, L. Spaete
Soil Organic Carbon (SOC) is a key soil property and is important for understanding carbon storage and soil-vegetation dynamics. Hyperspectral imagery (imaging spectroscopy) providing detailed spectral signatures of vegetation and soil make it possible to continuously map SOC content over a watershed scale. In this paper, the Next Generation Airborne Visible / Infrared Imaging Spectrometer (AVTPJSng) was used with an unmixing algorithm, the Multiple Endmember Spectral Mixture Analysis, to differentiate fractional cover of healthy vegetation, stressed vegetation and soil at the Reynolds Creek Critical Zone Observatory (PC-CZO). The fractional cover information was used to remove noisy spectra and the resulting residual spectra were used to predict SOC by Partial Least Squares Regression (PLSP). The results showed that the root mean standard error and mean bias of the predicted SOC (%) are 0.75 and 2.4, respectively. We found the best relationship between SOC and spectra after filtering out the influence of green vegetation from mixed spectra. The resulting residual, spectra comprised of stressed vegetation and soil, contained enough information for mapping SOC distribution within the shrub dominated regions of the watershed. This may provide a method to better understand the interaction of soil and vegetation in semiarid ecosystems.
{"title":"Spatial pattern of soil organic carbon acquired from hyperspectral imagery at reynolds creek critical zone observatory (RC-CZO)","authors":"Aihua Li, Ryan Will, N. Glenn, S. Benner, L. Spaete","doi":"10.1109/WHISPERS.2016.8071812","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071812","url":null,"abstract":"Soil Organic Carbon (SOC) is a key soil property and is important for understanding carbon storage and soil-vegetation dynamics. Hyperspectral imagery (imaging spectroscopy) providing detailed spectral signatures of vegetation and soil make it possible to continuously map SOC content over a watershed scale. In this paper, the Next Generation Airborne Visible / Infrared Imaging Spectrometer (AVTPJSng) was used with an unmixing algorithm, the Multiple Endmember Spectral Mixture Analysis, to differentiate fractional cover of healthy vegetation, stressed vegetation and soil at the Reynolds Creek Critical Zone Observatory (PC-CZO). The fractional cover information was used to remove noisy spectra and the resulting residual spectra were used to predict SOC by Partial Least Squares Regression (PLSP). The results showed that the root mean standard error and mean bias of the predicted SOC (%) are 0.75 and 2.4, respectively. We found the best relationship between SOC and spectra after filtering out the influence of green vegetation from mixed spectra. The resulting residual, spectra comprised of stressed vegetation and soil, contained enough information for mapping SOC distribution within the shrub dominated regions of the watershed. This may provide a method to better understand the interaction of soil and vegetation in semiarid ecosystems.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127128360","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071759
S. Keller, A. Braun, S. Hinz, M. Weinmann
In this paper, we address the classification of hyperspectral data which is comparable to the data acquired with the Environmental Mapping and Analysis Program (EnMAP) mission, a hyperspectral satellite mission supposed to be launched into space in the near future. While simulated EnMAP data has already been released, only relatively few studies have focused on investigating the performance of approaches for classifying such EnMAP data. Hence, in a recent paper, a contest for classifying EnMAP data has been initiated to foster research about possible exploitation strategies. Based on the dataset presented therein, we present a framework involving techniques of dimensionality reduction, feature selection and classification. We involve several classifiers for pixelwise classification based on different learning principles and investigate the impact of approaches for dimensionality reduction and feature selection on the classification results. The derived results clearly reveal the potential of respective techniques and provide the basis for further improvements in different research directions.
{"title":"Investigation of the impact of dimensionality reduction and feature selection on the classification of hyperspectral EnMAP data","authors":"S. Keller, A. Braun, S. Hinz, M. Weinmann","doi":"10.1109/WHISPERS.2016.8071759","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071759","url":null,"abstract":"In this paper, we address the classification of hyperspectral data which is comparable to the data acquired with the Environmental Mapping and Analysis Program (EnMAP) mission, a hyperspectral satellite mission supposed to be launched into space in the near future. While simulated EnMAP data has already been released, only relatively few studies have focused on investigating the performance of approaches for classifying such EnMAP data. Hence, in a recent paper, a contest for classifying EnMAP data has been initiated to foster research about possible exploitation strategies. Based on the dataset presented therein, we present a framework involving techniques of dimensionality reduction, feature selection and classification. We involve several classifiers for pixelwise classification based on different learning principles and investigate the impact of approaches for dimensionality reduction and feature selection on the classification results. The derived results clearly reveal the potential of respective techniques and provide the basis for further improvements in different research directions.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127175217","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071702
Hengqian Zhao, Lifu Zhang, Xuesheng Zhao
Diagnostic absorption feature has the potential to be the key factor of mineral information extraction in vegetation-covered region. Reference Spectral Background Removal (RSBR) could simulate the background curve based on the reference spectral background, and eliminate the influence through the background removal process. In this paper, RSBR was introduced into to mineral absorption feature extraction from high vegetation density area. Experiments on simulated data validated its great potential in mineral exploration in vegetation-covered region.
{"title":"Mineral absorption feature extraction in vegetation covered region based on reference spectral background removal","authors":"Hengqian Zhao, Lifu Zhang, Xuesheng Zhao","doi":"10.1109/WHISPERS.2016.8071702","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071702","url":null,"abstract":"Diagnostic absorption feature has the potential to be the key factor of mineral information extraction in vegetation-covered region. Reference Spectral Background Removal (RSBR) could simulate the background curve based on the reference spectral background, and eliminate the influence through the background removal process. In this paper, RSBR was introduced into to mineral absorption feature extraction from high vegetation density area. Experiments on simulated data validated its great potential in mineral exploration in vegetation-covered region.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127803189","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071755
Jean-Baptiste Courbot, E. Monfrini, V. Mazet, C. Collet
Hyperspectral image processing benefits greatly from using spatial information. Markov field modeling is a well-known statistical model class for considering spatial relationships between sites of an image. Often, the model restricts to Hidden Markov Field, therefore cannot handle non-stationarities in the images. This paper presents a Triplet Markov Field model for hyperspectral image segmentation, allowing the joint retrieving of image classes and local orientations. Segmentation results on synthetic data validate the methods, and results on real astronomical data are presented.
{"title":"Oriented Triplet Markov fields for hyperspectral image segmentation","authors":"Jean-Baptiste Courbot, E. Monfrini, V. Mazet, C. Collet","doi":"10.1109/WHISPERS.2016.8071755","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071755","url":null,"abstract":"Hyperspectral image processing benefits greatly from using spatial information. Markov field modeling is a well-known statistical model class for considering spatial relationships between sites of an image. Often, the model restricts to Hidden Markov Field, therefore cannot handle non-stationarities in the images. This paper presents a Triplet Markov Field model for hyperspectral image segmentation, allowing the joint retrieving of image classes and local orientations. Segmentation results on synthetic data validate the methods, and results on real astronomical data are presented.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124017192","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071785
Chen Mengshuo, Qian Yonggang, Wu Hua, Wang Ning, Ma Lingling, Li Chuanrong, Tang Lingli
Land surface temperature and emissivity separation (TES) is a key problem in thermal infrared (TIR) remote sensing. However, because of the ill-posed problem, the retrieval accuracy still needs to be improved. Through exploring the offset characteristics of atmospheric downward radiance, a temperature and emissivity retrieval algorithm based on atmospheric absorption feature is proposed from hyperspectral thermal infrared data. Furthermore, an optimal channel selection is carried out to improve the efficiency and accuracy of method. The simulated results show that modeling errors less than 0.4K for temperature and 1.5% for relative emissivity for contrast materials and the accuracy is similar to the ISSTES method (Borel, 2008) for high emissivity materials. Furthermore, the proposed method can enhance the retrieval accuracy for low emissivity materials, that is approximately temperature 0.5 K and emissivity 2.1%.
{"title":"A temperature and emissivity retrieval algorithm based on atmospheric absorption feature from hyperspectral thermal infrared data","authors":"Chen Mengshuo, Qian Yonggang, Wu Hua, Wang Ning, Ma Lingling, Li Chuanrong, Tang Lingli","doi":"10.1109/WHISPERS.2016.8071785","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071785","url":null,"abstract":"Land surface temperature and emissivity separation (TES) is a key problem in thermal infrared (TIR) remote sensing. However, because of the ill-posed problem, the retrieval accuracy still needs to be improved. Through exploring the offset characteristics of atmospheric downward radiance, a temperature and emissivity retrieval algorithm based on atmospheric absorption feature is proposed from hyperspectral thermal infrared data. Furthermore, an optimal channel selection is carried out to improve the efficiency and accuracy of method. The simulated results show that modeling errors less than 0.4K for temperature and 1.5% for relative emissivity for contrast materials and the accuracy is similar to the ISSTES method (Borel, 2008) for high emissivity materials. Furthermore, the proposed method can enhance the retrieval accuracy for low emissivity materials, that is approximately temperature 0.5 K and emissivity 2.1%.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"255 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117331637","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071758
P. Lahaie
In applications involving weak light signal like hyperspectral or time distributed signals obtained in applications involving laser induced fluorescence spectral detection, fluorescence lifetime imaging, Raman Spectroscopy or hyperspectral imaging in low light environment, the photons arrive at such a rate that they can be counted or have to be intensified to obtain a usable signal. Detection and classification algorithms need to be designed and evaluated for weak hyperspectral signal processing. A new algorithm, Adaptive Shot Noise (ASN) based on the assumption that a signal respects the Poisson multivariate distribution has been developed using the method of the maximum likelihood. This algorithm demonstrates the capability to be used for detection and classification. Using Monte Carlo simulations its performances are compared with the Adaptive Coherence Estimator (ACE) classification and with an Integrated Signal Algorithm (ISA) and ACE for detection. This new algorithm provides a small increase in performance compared to ACE in very weak signal conditions for classification and in some conditions better performance over both ACE and ISA in detection. The algorithm behavior like ACE shows sensitivity to assumption on the spectral characteristics of the source for the detection, which is not the case for ISA.
{"title":"Classification and anomaly detection algorithms for weak hyperspectral signal processing","authors":"P. Lahaie","doi":"10.1109/WHISPERS.2016.8071758","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071758","url":null,"abstract":"In applications involving weak light signal like hyperspectral or time distributed signals obtained in applications involving laser induced fluorescence spectral detection, fluorescence lifetime imaging, Raman Spectroscopy or hyperspectral imaging in low light environment, the photons arrive at such a rate that they can be counted or have to be intensified to obtain a usable signal. Detection and classification algorithms need to be designed and evaluated for weak hyperspectral signal processing. A new algorithm, Adaptive Shot Noise (ASN) based on the assumption that a signal respects the Poisson multivariate distribution has been developed using the method of the maximum likelihood. This algorithm demonstrates the capability to be used for detection and classification. Using Monte Carlo simulations its performances are compared with the Adaptive Coherence Estimator (ACE) classification and with an Integrated Signal Algorithm (ISA) and ACE for detection. This new algorithm provides a small increase in performance compared to ACE in very weak signal conditions for classification and in some conditions better performance over both ACE and ISA in detection. The algorithm behavior like ACE shows sensitivity to assumption on the spectral characteristics of the source for the detection, which is not the case for ISA.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132353288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071788
E. Carpena-Colon, Luis O. Jimenez-Rodriguez, Emmanuel Arzuaga, M. Velez-Reyes
This paper presents the development and enhancement of a subsurface (underwater) linear unmixing algorithm, called LIGU, specially conceived to determine individual contributions to the measured signal of given spectral reflectance of objects at the bottom of coastal shallow waters. This algorithm is part of a Hyperspectral Coastal Image Analysis Toolbox (HyCIAT), which is a repository of tools to be used to retrieve information from object embedded in a diffusive and murky medium. This paper discusses mathematical formulations behind the subsurface unmixing algorithm LIGU and presents enhancements made to the algorithm. Finally, quantitative and qualitative results will be presented using a hyperspectral data set from a controlled and well known environment. These results provide noticeable quantitative improvement when LIGU is compared with other linear unmixing algorithm not developed for subsurface (underwater) applications.
{"title":"Subsurface linear unmixing on a controlled underwater enviroment","authors":"E. Carpena-Colon, Luis O. Jimenez-Rodriguez, Emmanuel Arzuaga, M. Velez-Reyes","doi":"10.1109/WHISPERS.2016.8071788","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071788","url":null,"abstract":"This paper presents the development and enhancement of a subsurface (underwater) linear unmixing algorithm, called LIGU, specially conceived to determine individual contributions to the measured signal of given spectral reflectance of objects at the bottom of coastal shallow waters. This algorithm is part of a Hyperspectral Coastal Image Analysis Toolbox (HyCIAT), which is a repository of tools to be used to retrieve information from object embedded in a diffusive and murky medium. This paper discusses mathematical formulations behind the subsurface unmixing algorithm LIGU and presents enhancements made to the algorithm. Finally, quantitative and qualitative results will be presented using a hyperspectral data set from a controlled and well known environment. These results provide noticeable quantitative improvement when LIGU is compared with other linear unmixing algorithm not developed for subsurface (underwater) applications.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"89 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123593859","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071669
Cencen Pan, Kun Tan, Q. Du, Qinwu Yan, Jianwei Ding
Stripes in hyperspectral imagery reduce image quality and limit further applications. In this paper, we propose a novel destriping method. In this method, reference spectra is extracted in VNIR bands and linear unmixing is performed to denoise these bands, and abundance maps derived by VNIR bands are then used to repair SWIR bands. The error term of all the SWIR bands is also calculated, and the kriging interpolation method is used to interpolate error term, deriving the final destriped SWIR images. Destriping results shown that the proposed method outperforms the traditional kriging interpolation with visual inspection and quantitative assessment.
{"title":"Hyperspectral image destriping using unmixing-based kriging interpolation","authors":"Cencen Pan, Kun Tan, Q. Du, Qinwu Yan, Jianwei Ding","doi":"10.1109/WHISPERS.2016.8071669","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071669","url":null,"abstract":"Stripes in hyperspectral imagery reduce image quality and limit further applications. In this paper, we propose a novel destriping method. In this method, reference spectra is extracted in VNIR bands and linear unmixing is performed to denoise these bands, and abundance maps derived by VNIR bands are then used to repair SWIR bands. The error term of all the SWIR bands is also calculated, and the kriging interpolation method is used to interpolate error term, deriving the final destriped SWIR images. Destriping results shown that the proposed method outperforms the traditional kriging interpolation with visual inspection and quantitative assessment.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126046168","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In hyperspectral remote sensing imagery, the sensor, atmosphere, topography and other factors often bring about some degradations, such as noises, blurring, aliasing, clouding, shadowing, etc. Compensating for these degradations through quality improvement is a key preprocessing step in the exploitation of hyperspectral imagery. In this paper, a comprehensive analysis of the quality improvement techniques for hyperspectral images is presented. In order to embody the differences with those used for other types of images, the methods are classified according to their special processing strategies for hyperspectral images. Except for the description of the theory and methods, some experiments on hyperspectral images, including denoisng, deblurring, inpainting, destriping are illustrated. Some potential methods about this interesting topic are also discussed.
{"title":"Quality improvement of hyperspectral remote sensing images: A technical overview","authors":"Huifang Li, Huanfeng Shen, Q. Yuan, Hongyan Zhang, Lefei Zhang, Liangpei Zhang","doi":"10.1109/WHISPERS.2016.8071695","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071695","url":null,"abstract":"In hyperspectral remote sensing imagery, the sensor, atmosphere, topography and other factors often bring about some degradations, such as noises, blurring, aliasing, clouding, shadowing, etc. Compensating for these degradations through quality improvement is a key preprocessing step in the exploitation of hyperspectral imagery. In this paper, a comprehensive analysis of the quality improvement techniques for hyperspectral images is presented. In order to embody the differences with those used for other types of images, the methods are classified according to their special processing strategies for hyperspectral images. Except for the description of the theory and methods, some experiments on hyperspectral images, including denoisng, deblurring, inpainting, destriping are illustrated. Some potential methods about this interesting topic are also discussed.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127142122","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071716
Utsav B. Gewali, S. Monteiro
In this paper, we propose and compare two spectral angle based approaches for spatial-spectral classification. Our methods use the spectral angle to generate unary energies in a grid-structured Markov random field defined over the pixel labels of a hyperspectral image. The first approach is to use the exponential spectral angle mapper (ESAM) kernel/covariance function, a spectral angle based function, with the support vector machine and the Gaussian process classifier. The second approach is to directly use the minimum spectral angle between the test pixel and the training pixels as the unary energy. We compare the proposed methods with the state-of-the-art Markov random field methods that use support vector machines and Gaussian processes with squared exponential kernel/covariance function. In our experiments with two datasets, it is seen that using minimum spectral angle as unary energy produces better or comparable results to the existing methods at a smaller running time.
{"title":"Spectral angle based unary energy functions for spatial-spectral hyperspectral classification using Markov random fields","authors":"Utsav B. Gewali, S. Monteiro","doi":"10.1109/WHISPERS.2016.8071716","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071716","url":null,"abstract":"In this paper, we propose and compare two spectral angle based approaches for spatial-spectral classification. Our methods use the spectral angle to generate unary energies in a grid-structured Markov random field defined over the pixel labels of a hyperspectral image. The first approach is to use the exponential spectral angle mapper (ESAM) kernel/covariance function, a spectral angle based function, with the support vector machine and the Gaussian process classifier. The second approach is to directly use the minimum spectral angle between the test pixel and the training pixels as the unary energy. We compare the proposed methods with the state-of-the-art Markov random field methods that use support vector machines and Gaussian processes with squared exponential kernel/covariance function. In our experiments with two datasets, it is seen that using minimum spectral angle as unary energy produces better or comparable results to the existing methods at a smaller running time.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127479652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：