Subduction zones involve many complex geological processes,including the release of slabderived fluids,fluid/rock interactions,partial melting,isotopic fractionations,elemental transporting,and crust/mantle interactio...Subduction zones involve many complex geological processes,including the release of slabderived fluids,fluid/rock interactions,partial melting,isotopic fractionations,elemental transporting,and crust/mantle interactions.Lithium(Li)isotopes(~6Li and~7Li)have relative mass difference up to 16%,being the largest among metal elements.Thus,Li isotopes have advantage to interprete trace various geological processes.Most importantly,during crust/mantle interactions in deep subduction zones,surface materials and mantle rocks usually have distinct Li isotopic compositions.Li isotopes can be potential tracer for subduction processes,from the onset of subduction to the release of Li from subducted slabs and interaction with mantle wedge,as well as the fate of Li in slab-derived fluids and residual slabs.Moreover,the Li isotopic composition of subducting output materials can provide useful information for understanding global Li circulation.With developments in measurement and expansion of Li isotopic database,Li isotopic geochemistry will provide more inference and be a powerful tracer for understanding subduction-related processes.This work retrospected the application of Li isotopes in tracing successive subduction processes,and made some prospects for further studies of Li isotopes.展开更多
为了提升弱纹理区域无监督多视图深度估计性能,文中提出一种基于邻域自适应无监督多视图深度估计算法。算法采用双分支结构,深度估计分支首先采用邻域自适应深度分布方法改善弱纹理区域深度分布;其次采用深度变化概率引导的深度假设范...为了提升弱纹理区域无监督多视图深度估计性能,文中提出一种基于邻域自适应无监督多视图深度估计算法。算法采用双分支结构,深度估计分支首先采用邻域自适应深度分布方法改善弱纹理区域深度分布;其次采用深度变化概率引导的深度假设范围细化后续阶段深度估计。为了提高对场景边缘的识别,采用基于标准差的深度平滑约束。神经渲染分支用于提高深度估计能力,为了增强与深度估计分支间的几何一致性,采用融合图像颜色与深度信息的采样方法。由实验结果可知,该算法在DTU数据集测试完整度误差和整体精度误差优于其他无监督算法,且完整度误差比DS⁃MVSNet减小16.71%。可视化结果表明,针对弱纹理区域深度估计性能提升明显。在Tanks and Temples数据集上进行泛化性验证,整体性能(Mean)为56.22,证明了所提算法的有效性。展开更多
基金granted by the National Natural Science Foundation of China(NSFC 41273037,41673031,41473033)
文摘Subduction zones involve many complex geological processes,including the release of slabderived fluids,fluid/rock interactions,partial melting,isotopic fractionations,elemental transporting,and crust/mantle interactions.Lithium(Li)isotopes(~6Li and~7Li)have relative mass difference up to 16%,being the largest among metal elements.Thus,Li isotopes have advantage to interprete trace various geological processes.Most importantly,during crust/mantle interactions in deep subduction zones,surface materials and mantle rocks usually have distinct Li isotopic compositions.Li isotopes can be potential tracer for subduction processes,from the onset of subduction to the release of Li from subducted slabs and interaction with mantle wedge,as well as the fate of Li in slab-derived fluids and residual slabs.Moreover,the Li isotopic composition of subducting output materials can provide useful information for understanding global Li circulation.With developments in measurement and expansion of Li isotopic database,Li isotopic geochemistry will provide more inference and be a powerful tracer for understanding subduction-related processes.This work retrospected the application of Li isotopes in tracing successive subduction processes,and made some prospects for further studies of Li isotopes.
文摘为了提升弱纹理区域无监督多视图深度估计性能,文中提出一种基于邻域自适应无监督多视图深度估计算法。算法采用双分支结构,深度估计分支首先采用邻域自适应深度分布方法改善弱纹理区域深度分布;其次采用深度变化概率引导的深度假设范围细化后续阶段深度估计。为了提高对场景边缘的识别,采用基于标准差的深度平滑约束。神经渲染分支用于提高深度估计能力,为了增强与深度估计分支间的几何一致性,采用融合图像颜色与深度信息的采样方法。由实验结果可知,该算法在DTU数据集测试完整度误差和整体精度误差优于其他无监督算法,且完整度误差比DS⁃MVSNet减小16.71%。可视化结果表明,针对弱纹理区域深度估计性能提升明显。在Tanks and Temples数据集上进行泛化性验证,整体性能(Mean)为56.22,证明了所提算法的有效性。
