目的为了解决基于卷积神经网络的算法对高光谱图像小样本分类精度较低、模型结构复杂和计算量大的问题,提出了一种变维卷积神经网络。方法变维卷积神经网络对高光谱分类过程可根据内部特征图维度的变化分为空—谱信息融合、降维、混合...目的为了解决基于卷积神经网络的算法对高光谱图像小样本分类精度较低、模型结构复杂和计算量大的问题,提出了一种变维卷积神经网络。方法变维卷积神经网络对高光谱分类过程可根据内部特征图维度的变化分为空—谱信息融合、降维、混合特征提取与空—谱联合分类的过程。这种变维结构通过改变特征映射的维度,简化了网络结构并减少了计算量,并通过对空—谱信息的充分提取提高了卷积神经网络对小样本高光谱图像分类的精度。结果实验分为变维卷积神经网络的性能分析实验与分类性能对比实验,所用的数据集为Indian Pines和Pavia University Scene数据集。通过实验可知,变维卷积神经网络对高光谱小样本可取得较高的分类精度,在Indian Pines和Pavia University Scene数据集上的总体分类精度分别为87. 87%和98. 18%,与其他分类算法对比有较明显的性能优势。结论实验结果表明,合理的参数优化可有效提高变维卷积神经网络的分类精度,这种变维模型可较大程度提高对高光谱图像中小样本数据的分类性能,并可进一步推广到其他与高光谱图像相关的深度学习分类模型中。展开更多
To minimize the low classification accuracy and low utilization of spatial information in traditional hyperspectral image classification methods, we propose a new hyperspectral image classification method, which is ba...To minimize the low classification accuracy and low utilization of spatial information in traditional hyperspectral image classification methods, we propose a new hyperspectral image classification method, which is based on the Gabor spatial texture features and nonparametric weighted spectral features, and the sparse representation classification method（Gabor–NWSF and SRC）, abbreviated GNWSF–SRC. The proposed（GNWSF–SRC） method first combines the Gabor spatial features and nonparametric weighted spectral features to describe the hyperspectral image, and then applies the sparse representation method. Finally, the classification is obtained by analyzing the reconstruction error. We use the proposed method to process two typical hyperspectral data sets with different percentages of training samples. Theoretical analysis and simulation demonstrate that the proposed method improves the classification accuracy and Kappa coefficient compared with traditional classification methods and achieves better classification performance.展开更多
Interference alignment(IA) is suitable for cognitive radio networks(CRNs).However, in IA spectrum sharing(SS) process of general underlay CRNs, transmit power of cognitive radio transmitters usually should be reduced ...Interference alignment(IA) is suitable for cognitive radio networks(CRNs).However, in IA spectrum sharing(SS) process of general underlay CRNs, transmit power of cognitive radio transmitters usually should be reduced to satisfy interference constraint of primary user(PU), which may lead to low signalto-noise-ratio at cognitive radio receivers(CRRs). Consequently, sum rate of cognitive users(CUs) may fall short of the theoretical maximum through IA. To solve this problem,we propose an adaptive IA SS method for general distributed multi-user multi-antenna CRNs. The relationship between interference and noise power at each CRR is analyzed according to channel state information, interference requirement of PU, and power budget of CUs. Based on the analysis, scenarios of the CRN are classified into 4 cases, and corresponding IA SS algorithms are properly designed. Transmit power adjustment, CU access control and adjusted spatial projection are used to realize IA among CUs. Compared with existing methods, the proposed method is more general because of breaking the restriction that CUs can only transmit on the idle sub-channels. Moreover, in comparison to other five IA SS methods applicable in general CRN, the proposed method leads to improved achievable sum rate of CUs while guarantees transmission of PU.展开更多
文摘目的为了解决基于卷积神经网络的算法对高光谱图像小样本分类精度较低、模型结构复杂和计算量大的问题,提出了一种变维卷积神经网络。方法变维卷积神经网络对高光谱分类过程可根据内部特征图维度的变化分为空—谱信息融合、降维、混合特征提取与空—谱联合分类的过程。这种变维结构通过改变特征映射的维度,简化了网络结构并减少了计算量,并通过对空—谱信息的充分提取提高了卷积神经网络对小样本高光谱图像分类的精度。结果实验分为变维卷积神经网络的性能分析实验与分类性能对比实验,所用的数据集为Indian Pines和Pavia University Scene数据集。通过实验可知,变维卷积神经网络对高光谱小样本可取得较高的分类精度,在Indian Pines和Pavia University Scene数据集上的总体分类精度分别为87. 87%和98. 18%,与其他分类算法对比有较明显的性能优势。结论实验结果表明,合理的参数优化可有效提高变维卷积神经网络的分类精度,这种变维模型可较大程度提高对高光谱图像中小样本数据的分类性能,并可进一步推广到其他与高光谱图像相关的深度学习分类模型中。
基金supported by the National Natural Science Foundation of China(No.61275010)the Ph.D.Programs Foundation of Ministry of Education of China(No.20132304110007)+1 种基金the Heilongjiang Natural Science Foundation(No.F201409)the Fundamental Research Funds for the Central Universities(No.HEUCFD1410)
文摘To minimize the low classification accuracy and low utilization of spatial information in traditional hyperspectral image classification methods, we propose a new hyperspectral image classification method, which is based on the Gabor spatial texture features and nonparametric weighted spectral features, and the sparse representation classification method（Gabor–NWSF and SRC）, abbreviated GNWSF–SRC. The proposed（GNWSF–SRC） method first combines the Gabor spatial features and nonparametric weighted spectral features to describe the hyperspectral image, and then applies the sparse representation method. Finally, the classification is obtained by analyzing the reconstruction error. We use the proposed method to process two typical hyperspectral data sets with different percentages of training samples. Theoretical analysis and simulation demonstrate that the proposed method improves the classification accuracy and Kappa coefficient compared with traditional classification methods and achieves better classification performance.
基金supported by National Natuvertexesral Science Foundation of China under Grant 61201233 61271262 and 61701043
文摘Interference alignment(IA) is suitable for cognitive radio networks(CRNs).However, in IA spectrum sharing(SS) process of general underlay CRNs, transmit power of cognitive radio transmitters usually should be reduced to satisfy interference constraint of primary user(PU), which may lead to low signalto-noise-ratio at cognitive radio receivers(CRRs). Consequently, sum rate of cognitive users(CUs) may fall short of the theoretical maximum through IA. To solve this problem,we propose an adaptive IA SS method for general distributed multi-user multi-antenna CRNs. The relationship between interference and noise power at each CRR is analyzed according to channel state information, interference requirement of PU, and power budget of CUs. Based on the analysis, scenarios of the CRN are classified into 4 cases, and corresponding IA SS algorithms are properly designed. Transmit power adjustment, CU access control and adjusted spatial projection are used to realize IA among CUs. Compared with existing methods, the proposed method is more general because of breaking the restriction that CUs can only transmit on the idle sub-channels. Moreover, in comparison to other five IA SS methods applicable in general CRN, the proposed method leads to improved achievable sum rate of CUs while guarantees transmission of PU.
