The recently deployed Transition Region Explorer(TREx)-RGB(red-green-blue)all-sky imager(ASI)is designed to capture“true color”images of the aurora and airglow.Because the 557.7 nm green line is usually the brightes...The recently deployed Transition Region Explorer(TREx)-RGB(red-green-blue)all-sky imager(ASI)is designed to capture“true color”images of the aurora and airglow.Because the 557.7 nm green line is usually the brightest emission line in visible auroras,the green channel of a TREx-RGB camera is usually dominated by the 557.7 nm emission.Under this rationale,the TREx mission does not include a specific 557.7 nm imager and is designed to use the RGB green-channel data as a proxy for the 557.7 nm aurora.In this study,we present an initial effort to establish the conversion ratio or formula linking the RGB green-channel data to the absolute intensity of 557.7 nm auroras,which is crucial for quantitative uses of the RGB data.We illustrate two approaches:(1)through a comparison with the collocated measurement of green-line auroras from the TREx spectrograph,and(2)through a comparison with the modeled green-line intensity according to realistic electron precipitation flux measurements from low-Earth-orbit satellites,with the aid of an auroral transport model.We demonstrate the procedures and provide initial results for the TREx-RGB ASIs at the Rabbit Lake and Lucky Lake stations.The RGB response is found to be nonlinear.Empirical conversion ratios or formulas between RGB green-channel data and the green-line auroral intensity are given and can be applied immediately by TREx-RGB data users.The methodology established in this study will also be applicable to the upcoming SMILE ASI mission,which will adopt a similar RGB camera system in its deployment.展开更多
针对RGB(Red Green Blue)模态与热度模态信息表征形式不一致,特征信息无法有效挖掘、融合问题,提出了一种新的联合注意力强化网络-FCNet(Feature Sharpening and Cross-modal Feature Fusion Net)。首先,通过双维度注意力机制提升图像...针对RGB(Red Green Blue)模态与热度模态信息表征形式不一致,特征信息无法有效挖掘、融合问题,提出了一种新的联合注意力强化网络-FCNet(Feature Sharpening and Cross-modal Feature Fusion Net)。首先,通过双维度注意力机制提升图像特征映射能力;然后,利用跨模态特征融合机制捕获目标区域;最后,利用逐层解码结构消除背景干扰,优化检测目标。实验结果表明,该优化改进算法运算参数更少、运算时间更短,且模型整体检测性能均优于现有多模态检测模型性能。展开更多
针对室内场景图像语义分割结果不精确、显著图粗糙的问题,提出一种基于多模态特征优化提取和双路径引导解码的网络架构(feature regulator and dual-path guidance,FG-Net)。具体来说,设计的特征调节器对每个阶段的多模态特征依次进行...针对室内场景图像语义分割结果不精确、显著图粗糙的问题,提出一种基于多模态特征优化提取和双路径引导解码的网络架构(feature regulator and dual-path guidance,FG-Net)。具体来说,设计的特征调节器对每个阶段的多模态特征依次进行噪声过滤、重加权表示、差异性互补和交互融合,通过强化RGB和深度特征聚合,优化特征提取过程中的多模态特征表示。然后,在解码阶段引入特征交互融合后丰富的跨模态线索,进一步发挥多模态特征的优势。结合双路径协同引导结构,在解码阶段融合多尺度、多层次的特征信息,从而输出更细致的显著图。实验在公开数据集NYUD-v2和SUN RGB-D上进行,在主要评价指标mIoU上达到48.5%,优于其他先进算法。结果表明,该算法实现了更精细的室内场景图像语义分割,表现出了较好的泛化性和鲁棒性。展开更多
为推进富含花青素的紫苏品种选育,指导逆境胁迫下的紫苏生产管理,以紫苏为研究对象,采集田间叶片并使用数码相机拍照,结合红绿蓝色彩空间(red green blue color space,RGB)和CIELab色彩空间(CIELab color space)2种图像色彩分析手段处...为推进富含花青素的紫苏品种选育,指导逆境胁迫下的紫苏生产管理,以紫苏为研究对象,采集田间叶片并使用数码相机拍照,结合红绿蓝色彩空间(red green blue color space,RGB)和CIELab色彩空间(CIELab color space)2种图像色彩分析手段处理图片,与叶片花青素含量进行相关性和显著性分析,筛选出相关系数较高的色彩参数,建立单变量回归反演模型,最终综合建模得到预测效果最优的紫苏叶片花青素含量预测模型。结果表明,在RGB色彩空间中,红光标准化值(normalized redness intensity,NRI)、绿光标准化值(normalized greenness intensity,NGI)与花青素含量呈极显著相关,其中NGI的相关系数大于NRI。当叶片正反面色彩贡献比为2∶1时,NGI与花青素含量的相关性最大,相关系数为0.8532。对比不同模型发现,以NGI为自变量建立的指数模型拟合效果最好,相关系数为0.8381,决定系数(R^(2))达0.7550。在CIELab色彩空间中,红度(a^(*))与花青素含量的相关性最好,且相关系数同样在叶片正反面色彩贡献比为2∶1时达最大,为0.7356。基于a^(*)建立的幂模型拟合效果最好,相关系数和R^(2)分别为0.7438和0.6798。分别使用NGI模型和a^(*)模型对叶片花青素含量进行估测,验证后发现a^(*)模型的预测效果更好,准确性和稳定性更高,因此以a^(*)模型为预测紫苏叶片花青素含量的最优模型。展开更多
基金jointly funded by the Canada Foundation for Innovationthe Alberta Economic Development and Trade organization+1 种基金the University of Calgarysupported by the Canadian Space Agency。
文摘The recently deployed Transition Region Explorer(TREx)-RGB(red-green-blue)all-sky imager(ASI)is designed to capture“true color”images of the aurora and airglow.Because the 557.7 nm green line is usually the brightest emission line in visible auroras,the green channel of a TREx-RGB camera is usually dominated by the 557.7 nm emission.Under this rationale,the TREx mission does not include a specific 557.7 nm imager and is designed to use the RGB green-channel data as a proxy for the 557.7 nm aurora.In this study,we present an initial effort to establish the conversion ratio or formula linking the RGB green-channel data to the absolute intensity of 557.7 nm auroras,which is crucial for quantitative uses of the RGB data.We illustrate two approaches:(1)through a comparison with the collocated measurement of green-line auroras from the TREx spectrograph,and(2)through a comparison with the modeled green-line intensity according to realistic electron precipitation flux measurements from low-Earth-orbit satellites,with the aid of an auroral transport model.We demonstrate the procedures and provide initial results for the TREx-RGB ASIs at the Rabbit Lake and Lucky Lake stations.The RGB response is found to be nonlinear.Empirical conversion ratios or formulas between RGB green-channel data and the green-line auroral intensity are given and can be applied immediately by TREx-RGB data users.The methodology established in this study will also be applicable to the upcoming SMILE ASI mission,which will adopt a similar RGB camera system in its deployment.
文摘针对RGB(Red Green Blue)模态与热度模态信息表征形式不一致,特征信息无法有效挖掘、融合问题,提出了一种新的联合注意力强化网络-FCNet(Feature Sharpening and Cross-modal Feature Fusion Net)。首先,通过双维度注意力机制提升图像特征映射能力;然后,利用跨模态特征融合机制捕获目标区域;最后,利用逐层解码结构消除背景干扰,优化检测目标。实验结果表明,该优化改进算法运算参数更少、运算时间更短,且模型整体检测性能均优于现有多模态检测模型性能。
文摘为推进富含花青素的紫苏品种选育,指导逆境胁迫下的紫苏生产管理,以紫苏为研究对象,采集田间叶片并使用数码相机拍照,结合红绿蓝色彩空间(red green blue color space,RGB)和CIELab色彩空间(CIELab color space)2种图像色彩分析手段处理图片,与叶片花青素含量进行相关性和显著性分析,筛选出相关系数较高的色彩参数,建立单变量回归反演模型,最终综合建模得到预测效果最优的紫苏叶片花青素含量预测模型。结果表明,在RGB色彩空间中,红光标准化值(normalized redness intensity,NRI)、绿光标准化值(normalized greenness intensity,NGI)与花青素含量呈极显著相关,其中NGI的相关系数大于NRI。当叶片正反面色彩贡献比为2∶1时,NGI与花青素含量的相关性最大,相关系数为0.8532。对比不同模型发现,以NGI为自变量建立的指数模型拟合效果最好,相关系数为0.8381,决定系数(R^(2))达0.7550。在CIELab色彩空间中,红度(a^(*))与花青素含量的相关性最好,且相关系数同样在叶片正反面色彩贡献比为2∶1时达最大,为0.7356。基于a^(*)建立的幂模型拟合效果最好,相关系数和R^(2)分别为0.7438和0.6798。分别使用NGI模型和a^(*)模型对叶片花青素含量进行估测,验证后发现a^(*)模型的预测效果更好,准确性和稳定性更高,因此以a^(*)模型为预测紫苏叶片花青素含量的最优模型。
