Pub Date : 2016-08-01DOI: 10.1109/WHISPERS.2016.8071810
Jihao Yin, Hui Qv, Xiaoyan Luo
A Conjugated and Augmented Dictionaries (CAD) learning method based on Sparse Auto-Encoder (SAE) is proposed for hyperspectral image classification. The CAD originates from the intention to combine the synthesis model and analysis model. These two models are used to obtain the sparse representation or feature of the pixels. In this paper, CAD has a three-step strategy to learn the dictionaries and classify the pixels of Hyperspectral image. Firstly, we adopt the Sparse Auto-Encoder model to complete the learning process of the suggested dictionaries. Secondly, test samples are reconstructed using the learned dictionaries. Finally, we embed the reconstructed pixels into a linear SVM for classification. Indiana Pine subset is used for the classification experiment, and the classification results show that the reconstructed pixels have the high discrimination characteristics, which makes our method outperforms other hyperspectral image classification algorithms as contrast.
{"title":"A conjugated and augmented dictionary learning method for hyperspectral image classfication","authors":"Jihao Yin, Hui Qv, Xiaoyan Luo","doi":"10.1109/WHISPERS.2016.8071810","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071810","url":null,"abstract":"A Conjugated and Augmented Dictionaries (CAD) learning method based on Sparse Auto-Encoder (SAE) is proposed for hyperspectral image classification. The CAD originates from the intention to combine the synthesis model and analysis model. These two models are used to obtain the sparse representation or feature of the pixels. In this paper, CAD has a three-step strategy to learn the dictionaries and classify the pixels of Hyperspectral image. Firstly, we adopt the Sparse Auto-Encoder model to complete the learning process of the suggested dictionaries. Secondly, test samples are reconstructed using the learned dictionaries. Finally, we embed the reconstructed pixels into a linear SVM for classification. Indiana Pine subset is used for the classification experiment, and the classification results show that the reconstructed pixels have the high discrimination characteristics, which makes our method outperforms other hyperspectral image classification algorithms as contrast.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"15 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":"115219613","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.2016.8071791
Zebin Wu, Qicong Wang, A. Plaza, Jun Li, Jie Wei, Zhihui Wei
The dimensionality of hyperspectral data is very high, and spectral-spatial hyperspectral classification techniques are quite demanding from a computational viewpoint. In this paper, we present a computationally efficient implementation of a spectral-spatial classification method based on weighted Markov random fields. The method learns the spectral information from a sparse multinomial logistic regression (SMLR) classifier, and the spatial information is characterized by modeling the potential function associated with a weighted Markov random field (MRF) as a spatially adaptive vector total variation function. The parallel implementation has been carried out using commodity graphics processing units (GPUs) and the NVIDIA's compute unified device architecture (CUDA), thus exploiting the massively parallel nature of GPUs to achieve significant acceleration factors with regards to the serial version of the same classifier on an NVIDIA Tesla C2075 platform.
高光谱数据的维数非常高，光谱-空间高光谱分类技术从计算角度来说要求很高。本文提出了一种基于加权马尔可夫随机场的光谱空间分类方法的计算效率实现。该方法从稀疏多项式逻辑回归(SMLR)分类器中学习光谱信息，并通过将加权马尔可夫随机场(MRF)相关的势函数建模为空间自适应向量总变分函数来表征空间信息。并行实现使用商用图形处理单元(gpu)和NVIDIA的计算统一设备架构(CUDA)进行，从而利用gpu的大规模并行特性，在NVIDIA Tesla C2075平台上实现与串行版本相同分类器相比的显著加速因子。
{"title":"GPU implementation of hyperspectral image classification based on weighted Markov random fields","authors":"Zebin Wu, Qicong Wang, A. Plaza, Jun Li, Jie Wei, Zhihui Wei","doi":"10.1109/WHISPERS.2016.8071791","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071791","url":null,"abstract":"The dimensionality of hyperspectral data is very high, and spectral-spatial hyperspectral classification techniques are quite demanding from a computational viewpoint. In this paper, we present a computationally efficient implementation of a spectral-spatial classification method based on weighted Markov random fields. The method learns the spectral information from a sparse multinomial logistic regression (SMLR) classifier, and the spatial information is characterized by modeling the potential function associated with a weighted Markov random field (MRF) as a spatially adaptive vector total variation function. The parallel implementation has been carried out using commodity graphics processing units (GPUs) and the NVIDIA's compute unified device architecture (CUDA), thus exploiting the massively parallel nature of GPUs to achieve significant acceleration factors with regards to the serial version of the same classifier on an NVIDIA Tesla C2075 platform.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"48 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":"125657834","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.2016.8071769
R. Witkosky, P. Adams, S. Akciz, K. Buckland, Janet Harvey, P. Johnson, D. Lynch, Frank Sousa, J. Stock, D. Tratt
The 1999 Hector Mine earthquake on the Lavic Lake fault produced a maximum right-lateral displacement of ∼5 m, but the long-term cumulative offset remains unresolved. To identify bedrock that has been offset by the fault, we produced a swath map from airborne hyperspectral imagery. High spatial and spectral resolution, along with a lack of significant vegetation cover helped us differentiate lithologic units and create a geologic map with supervised and unsupervised classifications. Supervised classifications over a small test site had an overall accuracy of 71 ± 1%, and some of the boundaries between units in our unsupervised classification correlate well with lithologic boundaries from a previously published geologic map that covers the same area. Our geologic fault swath map will help to resolve the total tectonic offset of bedrock along the Lavic Lake fault.
{"title":"Geologic swath map of the lavic lake fault from airborne thermal hyperspectral imagery","authors":"R. Witkosky, P. Adams, S. Akciz, K. Buckland, Janet Harvey, P. Johnson, D. Lynch, Frank Sousa, J. Stock, D. Tratt","doi":"10.1109/WHISPERS.2016.8071769","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071769","url":null,"abstract":"The 1999 Hector Mine earthquake on the Lavic Lake fault produced a maximum right-lateral displacement of ∼5 m, but the long-term cumulative offset remains unresolved. To identify bedrock that has been offset by the fault, we produced a swath map from airborne hyperspectral imagery. High spatial and spectral resolution, along with a lack of significant vegetation cover helped us differentiate lithologic units and create a geologic map with supervised and unsupervised classifications. Supervised classifications over a small test site had an overall accuracy of 71 ± 1%, and some of the boundaries between units in our unsupervised classification correlate well with lithologic boundaries from a previously published geologic map that covers the same area. Our geologic fault swath map will help to resolve the total tectonic offset of bedrock along the Lavic Lake fault.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"21 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":"124445138","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.2016.8071680
Wenzi Liao, Daniel Erick Ochoa Donoso, F. V. Coillie, Jie Li, C. Qi, S. Gautama, W. Philips
Hyperspectral (HS) imagery contains a wealth of spectral and spatial information that can improve target detection and recognition performance. Conventional spectral-spatial classification methods cannot fully exploit both spectral and spatial information of HS image. In this paper, we propose a new method to fuse the spectral and spatial information for HS image classification. Our approach transfers the spatial structures of the whole morphological profile into the original HS image by using bilateral filtering, and obtains an enhanced HS image enriching both spectral and spatial information. Meanwhile, the enhanced HS image has the same spectral and spatial dimensions as the original HS image, which may provide a new input to improve the performances of existing HS image classification methods. Experimental results on real HS images are very encouraging. Compared to the methods using only single feature and stacking all the features together, the proposed fusion method improves the overall classification accuracy more than 10% and 5%, respectively.
{"title":"Spectral-spatial classification for hyperspectral image by bilateral filtering and morphological features","authors":"Wenzi Liao, Daniel Erick Ochoa Donoso, F. V. Coillie, Jie Li, C. Qi, S. Gautama, W. Philips","doi":"10.1109/WHISPERS.2016.8071680","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071680","url":null,"abstract":"Hyperspectral (HS) imagery contains a wealth of spectral and spatial information that can improve target detection and recognition performance. Conventional spectral-spatial classification methods cannot fully exploit both spectral and spatial information of HS image. In this paper, we propose a new method to fuse the spectral and spatial information for HS image classification. Our approach transfers the spatial structures of the whole morphological profile into the original HS image by using bilateral filtering, and obtains an enhanced HS image enriching both spectral and spatial information. Meanwhile, the enhanced HS image has the same spectral and spatial dimensions as the original HS image, which may provide a new input to improve the performances of existing HS image classification methods. Experimental results on real HS images are very encouraging. Compared to the methods using only single feature and stacking all the features together, the proposed fusion method improves the overall classification accuracy more than 10% and 5%, respectively.","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":"121829493","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.2016.8071717
J. Theiler, G. Grosklos
We examine the Mahalanobis-distance based kernel-RX (KRX) algorithm for anomaly detection, and find that it can exhibit an unfortunate phenomenon: the anomalousness, for points far from the training data, can decrease with increasing distance. We demonstrate this directly for a few special cases, and provide a more general argument that applies in the large bandwidth regime.
{"title":"Cracks in KRX: When more distant points are less anomalous","authors":"J. Theiler, G. Grosklos","doi":"10.1109/WHISPERS.2016.8071717","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071717","url":null,"abstract":"We examine the Mahalanobis-distance based kernel-RX (KRX) algorithm for anomaly detection, and find that it can exhibit an unfortunate phenomenon: the anomalousness, for points far from the training data, can decrease with increasing distance. We demonstrate this directly for a few special cases, and provide a more general argument that applies in the large bandwidth regime.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"7 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":"124755004","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.2016.8071794
Xia Zhang, Xing Wu, Honglei Lin
Phyllosilicate is a principal form of hydrous minerals on the martian surface. It's also an indicative mineral in comparing different sediments and degree of aqueous alteration. Shortwave infrared (SWIR) and thermal infrared (TIR) spectral bands have distinct spectral response to the mineral groups and ions. However, combining SWIR and TIR to recognize phyllosilicate has been rarely studied. Based on the USGS spectral library, facing sensors of Mars: Compact Reconnaissance Imaging Spectrometer for Mars(CRISM) and Thermal Emission Imaging System(THEMIS), we conducted the research on the mechanis m of the spectral response of phyllosilicate, and established the SWIR and TIR identification model respectively, then combined the SWIR and TIR spectral features to build the combined recognition model of phyllosilicate by Fisher discriminant analysis. The results show that the identification accuracy of the combined model is the highest, which can correctly classify 90.6% of the mineral samples and improve the identification accuracy of phyllosilicate effectively.
{"title":"Combining SWIR and TIR spectral features for regnizaion of phyllosilicate of martian surface","authors":"Xia Zhang, Xing Wu, Honglei Lin","doi":"10.1109/WHISPERS.2016.8071794","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071794","url":null,"abstract":"Phyllosilicate is a principal form of hydrous minerals on the martian surface. It's also an indicative mineral in comparing different sediments and degree of aqueous alteration. Shortwave infrared (SWIR) and thermal infrared (TIR) spectral bands have distinct spectral response to the mineral groups and ions. However, combining SWIR and TIR to recognize phyllosilicate has been rarely studied. Based on the USGS spectral library, facing sensors of Mars: Compact Reconnaissance Imaging Spectrometer for Mars(CRISM) and Thermal Emission Imaging System(THEMIS), we conducted the research on the mechanis m of the spectral response of phyllosilicate, and established the SWIR and TIR identification model respectively, then combined the SWIR and TIR spectral features to build the combined recognition model of phyllosilicate by Fisher discriminant analysis. The results show that the identification accuracy of the combined model is the highest, which can correctly classify 90.6% of the mineral samples and improve the identification accuracy of phyllosilicate effectively.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"39 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":"122136718","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}
Band selection (BS) is an important dimensionality reduction procedure in hyperspectral data processing, which selects a subset of original bands that contain the most useful information about objects. Ant Colony Optimization (ACO) algorithm was recently introduced for band selection from hyperspectral images. This algorithm has been demonstrated it could select satisfactory results in experimental analysis. However, the ACO-based band selection (ACOBS) is time-consuming for hyperspectral image analysis due to its high computational amount. In this paper, the high-performance computing technology based on the Graphics Processing Units (GPUs) was utilized to improve the computational efficiency of the ACOBS algorithm. The experimental results showed that the computational performance of ACOBS based on GPU was significantly improved in the analysis of real hyperspectral data.
{"title":"GPU implementation of ant colony optimization-based band selections for hyperspectral data classification","authors":"Jianwei Gao, Zhengchao Chen, Lianru Gao, Bing Zhang","doi":"10.1109/WHISPERS.2016.8071720","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071720","url":null,"abstract":"Band selection (BS) is an important dimensionality reduction procedure in hyperspectral data processing, which selects a subset of original bands that contain the most useful information about objects. Ant Colony Optimization (ACO) algorithm was recently introduced for band selection from hyperspectral images. This algorithm has been demonstrated it could select satisfactory results in experimental analysis. However, the ACO-based band selection (ACOBS) is time-consuming for hyperspectral image analysis due to its high computational amount. In this paper, the high-performance computing technology based on the Graphics Processing Units (GPUs) was utilized to improve the computational efficiency of the ACOBS algorithm. The experimental results showed that the computational performance of ACOBS based on GPU was significantly improved in the analysis of real hyperspectral data.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"176 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":"130577530","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.2016.8071763
Mingming Xu, Liangpei Zhang, Bo Du, Lefei Zhang
Spectral unmixing is one of the most important techniques for analyzing hyperspectral images and many hyperspectral unmixing algorithms were developed under an assumption that pure pixels exist in recent years. However, the pure-pixel assumption may be seriously violated for highly mixed data. Endmember extraction can be regards as an optimization problem no matter whether pure-pixel exists or not. In this paper, we incorporate linear mixed model and particle swarm optimization to develop a linear mixed model constrained particle swarm optimization (LMMC-PSO) for endmember extraction from highly mixed data. Each particle in LMMC-PSO moves in search space according to linear mixed model rather than with a velocity, which is dynamically adjusted according to its own optimal position and global optimum of all particles. The experimental results indicated that the proposed method obtained better results than the algorithms of VCA, MVC-NMF, MVSA, MVES, and SISAL.
{"title":"The linear mixed model constrained particle swarm optimization for hyperspectral endmember extraction from highly mixed data","authors":"Mingming Xu, Liangpei Zhang, Bo Du, Lefei Zhang","doi":"10.1109/WHISPERS.2016.8071763","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071763","url":null,"abstract":"Spectral unmixing is one of the most important techniques for analyzing hyperspectral images and many hyperspectral unmixing algorithms were developed under an assumption that pure pixels exist in recent years. However, the pure-pixel assumption may be seriously violated for highly mixed data. Endmember extraction can be regards as an optimization problem no matter whether pure-pixel exists or not. In this paper, we incorporate linear mixed model and particle swarm optimization to develop a linear mixed model constrained particle swarm optimization (LMMC-PSO) for endmember extraction from highly mixed data. Each particle in LMMC-PSO moves in search space according to linear mixed model rather than with a velocity, which is dynamically adjusted according to its own optimal position and global optimum of all particles. The experimental results indicated that the proposed method obtained better results than the algorithms of VCA, MVC-NMF, MVSA, MVES, and SISAL.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"70 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":"123112002","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.2016.8071760
N. Yokoya, Xiaoxiang Zhu, A. Plaza
A new methodology that solves unmixing problems involving a set of multisensor time-series spectral images is proposed in order to understand dynamic changes of the surface at a subpixel scale. The proposed methodology couples multiple unmixing problems via regularization on graphs between the multisensor time-series data to obtain robust and stable unmixing solutions beyond data modalities owing to different sensor characteristics and the effects of non-optimal atmospheric correction. A synthetic dataset that includes seasonal and trend changes on the surface and the residuals of non-optimal atmospheric correction is used for numerical validation. Experimental results demonstrate the effectiveness of the proposed methodology.
{"title":"Graph-regularized coupled spectral unmixing for multisensor time-series analysis","authors":"N. Yokoya, Xiaoxiang Zhu, A. Plaza","doi":"10.1109/WHISPERS.2016.8071760","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071760","url":null,"abstract":"A new methodology that solves unmixing problems involving a set of multisensor time-series spectral images is proposed in order to understand dynamic changes of the surface at a subpixel scale. The proposed methodology couples multiple unmixing problems via regularization on graphs between the multisensor time-series data to obtain robust and stable unmixing solutions beyond data modalities owing to different sensor characteristics and the effects of non-optimal atmospheric correction. A synthetic dataset that includes seasonal and trend changes on the surface and the residuals of non-optimal atmospheric correction is used for numerical validation. Experimental results demonstrate the effectiveness of the proposed methodology.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"14 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":"115091792","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.2016.8071739
L. Huo, Lijun Zhao, Ping Tang
Sparse representation classifiers have been widely studied for hyperspectral image classification. The success of sparse representation classifiers depends highly on the training dictionary. However, the definition of training samples, often in the form of field investigations, is time consuming and costly. To mitigate the problem, active learning tries to iteratively define the most informative training samples based on the outputs of the classifiers, thus reducing the quantities of samples to be labeled. For different classification models, several different active learning strategies have been proposed. In this paper, we studied one active learning strategy for sparse representation classifiers. The main idea of the proposed algorithm is to select the samples with most similar reconstruction errors for two different classes. The experiments are performed on two public hyperspectral data. The results show the effectiveness of the proposed algorithm.
{"title":"Hyperspectral image classification with sparse representation classifier and active learning","authors":"L. Huo, Lijun Zhao, Ping Tang","doi":"10.1109/WHISPERS.2016.8071739","DOIUrl":"https://doi.org/10.1109/WHISPERS.2016.8071739","url":null,"abstract":"Sparse representation classifiers have been widely studied for hyperspectral image classification. The success of sparse representation classifiers depends highly on the training dictionary. However, the definition of training samples, often in the form of field investigations, is time consuming and costly. To mitigate the problem, active learning tries to iteratively define the most informative training samples based on the outputs of the classifiers, thus reducing the quantities of samples to be labeled. For different classification models, several different active learning strategies have been proposed. In this paper, we studied one active learning strategy for sparse representation classifiers. The main idea of the proposed algorithm is to select the samples with most similar reconstruction errors for two different classes. The experiments are performed on two public hyperspectral data. The results show the effectiveness of the proposed algorithm.","PeriodicalId":369281,"journal":{"name":"2016 8th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS)","volume":"35 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":"129587237","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：