[目的]为探究作物覆盖坡面地表微地形特征提取的可能性,实现作物覆盖条件下微地形三维建模与动态监测。[方法]采用SfM-MVS(structure from motion with multi-view stereo)方法,以玉米覆盖坡面为研究对象,裸坡作为对照,开展以裸坡及模...[目的]为探究作物覆盖坡面地表微地形特征提取的可能性,实现作物覆盖条件下微地形三维建模与动态监测。[方法]采用SfM-MVS(structure from motion with multi-view stereo)方法,以玉米覆盖坡面为研究对象,裸坡作为对照,开展以裸坡及模拟微地形的标志物为基准,玉米覆盖条件下地表微地形构建提取和精度评估研究。[结果]1)玉米覆盖坡面控制点误差<0.002 m,稀疏点云与密集点云数量分别为裸坡的2.1、2.6倍。2)基于C2C(Cloud to Cloud)的结果表明,玉米覆盖坡面与裸坡点云85.12%距离<0.001 m;DOD(DEM of Difference)的结果表明,玉米覆盖坡面与裸坡DEM有96.07%高差绝对值<0.003 m,整体精度可达毫米级。3)裸坡标志物微地形提取更接近实际值,各方向标志物长、宽、深精度均在98%以上;玉米覆盖坡面标志物虽因植被叠加导致横向标志物精度有所下降,但精度仍在97%以上。[结论]基于SfM-MVS的测量法,可应用于玉米覆盖坡面微地形起伏特征的提取。展开更多
6月29日,电竞世界杯(EWC)发布了2025赛季主题曲《Til My Fingers Bleed》和宣传片MV,由韩国男团SEVENTEEN成员DINO、美国嘻哈歌手Duckwrth以及金属核乐队The Word Alive共同演唱。MV通过视觉语言呈现出多元文化的碰撞。画面不仅包括三...6月29日,电竞世界杯(EWC)发布了2025赛季主题曲《Til My Fingers Bleed》和宣传片MV,由韩国男团SEVENTEEN成员DINO、美国嘻哈歌手Duckwrth以及金属核乐队The Word Alive共同演唱。MV通过视觉语言呈现出多元文化的碰撞。画面不仅包括三位歌手在录音棚录制歌曲的片段,也穿插了大量2024年EWC经典赛事的镜头。展开更多
基于学习的多视图立体匹配算法目前成果显著,但是仍然存在的卷积感受野受限以及忽略图像频率信息导致在低纹理、重复和非兰伯曲面匹配性能不足的问题,针对以上问题提出了上下文增强与图像频率引导的多视图立体匹配网络CAF-MVSNet.首先,...基于学习的多视图立体匹配算法目前成果显著,但是仍然存在的卷积感受野受限以及忽略图像频率信息导致在低纹理、重复和非兰伯曲面匹配性能不足的问题,针对以上问题提出了上下文增强与图像频率引导的多视图立体匹配网络CAF-MVSNet.首先,在特征提取阶段,将上下文增强模块融合到特征金字塔网络中,有效地扩大网络的感受野.然后引入了图像频率引导注意力模块,通过编码图像的不同频率获取图像的线条、形状、纹理和颜色等信息,增强图像的远程上下文联系的同时进一步解决低纹理、重复和非兰伯曲面的精确匹配问题,以实现可靠的特征匹配.在DTU数据集上的实验结果显示,与经典的级联模型Cas MVSNet相比综合误差(overall)提升了12.3%,展现了优秀的性能.此外,在Tanks and Temples数据集上也取得了不错的效果,展现了良好的泛化性能.展开更多
The most crucial requirement in radiation therapy treatment planning is a fast and accurate treatment planning system that minimizes damage to healthy tissues surrounding cancer cells. The use of Monte Carlo toolkits ...The most crucial requirement in radiation therapy treatment planning is a fast and accurate treatment planning system that minimizes damage to healthy tissues surrounding cancer cells. The use of Monte Carlo toolkits has become indispensable for research aimed at precisely determining the dose in radiotherapy. Among the numerous algorithms developed in recent years, the GAMOS code, which utilizes the Geant4 toolkit for Monte Carlo simula-tions, incorporates various electromagnetic physics models and multiple scattering models for simulating particle interactions with matter. This makes it a valuable tool for dose calculations in medical applications and throughout the patient’s volume. The aim of this present work aims to vali-date the GAMOS code for the simulation of a 6 MV photon-beam output from the Elekta Synergy Agility linear accelerator. The simulation involves mod-eling the major components of the accelerator head and the interactions of the radiation beam with a homogeneous water phantom and particle information was collected following the modeling of the phase space. This space was po-sitioned under the X and Y jaws, utilizing three electromagnetic physics mod-els of the GAMOS code: Standard, Penelope, and Low-Energy, along with three multiple scattering models: Goudsmit-Saunderson, Urban, and Wentzel-VI. The obtained phase space file was used as a particle source to simulate dose distributions (depth-dose and dose profile) for field sizes of 5 × 5 cm<sup>2</sup> and 10 × 10 cm<sup>2</sup> at depths of 10 cm and 20 cm in a water phantom, with a source-surface distance (SSD) of 90 cm from the target. We compared the three electromagnetic physics models and the three multiple scattering mod-els of the GAMOS code to experimental results. Validation of our results was performed using the gamma index, with an acceptability criterion of 3% for the dose difference (DD) and 3 mm for the distance-to-agreement (DTA). We achieved agreements of 94% and 96%, respectively, between simulation and experimentation for the three electromagnetic physics models and three mul-tiple scattering models, for field sizes of 5 × 5 cm<sup>2</sup> and 10 × 10 cm<sup>2</sup> for depth-dose curves. For dose profile curves, a good agreement of 100% was found between simulation and experimentation for the three electromagnetic physics models, as well as for the three multiple scattering models for a field size of 5 × 5 cm<sup>2</sup> at 10 cm and 20 cm depths. For a field size of 10 × 10 cm<sup>2</sup>, the Penelope model dominated with 98% for 10 cm, along with the three multiple scattering models. The Penelope model and the Standard model, along with the three multiple scattering models, dominated with 100% for 20 cm. Our study, which compared these different GAMOS code models, can be crucial for enhancing the accuracy and quality of radiotherapy, contributing to more effective patient treatment. Our research compares various electro-magnetic physics models and multiple scattering models with experimental measurements, enabling us to choose the models that produce the most reli-able results, thereby directly impacting the quality of simulations. This en-hances confidence in using these models for treatment planning. Our re-search consistently contributes to the progress of Monte Carlo simulation techniques in radiation therapy, enriching the scientific literature.展开更多
文摘[目的]为探究作物覆盖坡面地表微地形特征提取的可能性,实现作物覆盖条件下微地形三维建模与动态监测。[方法]采用SfM-MVS(structure from motion with multi-view stereo)方法,以玉米覆盖坡面为研究对象,裸坡作为对照,开展以裸坡及模拟微地形的标志物为基准,玉米覆盖条件下地表微地形构建提取和精度评估研究。[结果]1)玉米覆盖坡面控制点误差<0.002 m,稀疏点云与密集点云数量分别为裸坡的2.1、2.6倍。2)基于C2C(Cloud to Cloud)的结果表明,玉米覆盖坡面与裸坡点云85.12%距离<0.001 m;DOD(DEM of Difference)的结果表明,玉米覆盖坡面与裸坡DEM有96.07%高差绝对值<0.003 m,整体精度可达毫米级。3)裸坡标志物微地形提取更接近实际值,各方向标志物长、宽、深精度均在98%以上;玉米覆盖坡面标志物虽因植被叠加导致横向标志物精度有所下降,但精度仍在97%以上。[结论]基于SfM-MVS的测量法,可应用于玉米覆盖坡面微地形起伏特征的提取。
文摘6月29日,电竞世界杯(EWC)发布了2025赛季主题曲《Til My Fingers Bleed》和宣传片MV,由韩国男团SEVENTEEN成员DINO、美国嘻哈歌手Duckwrth以及金属核乐队The Word Alive共同演唱。MV通过视觉语言呈现出多元文化的碰撞。画面不仅包括三位歌手在录音棚录制歌曲的片段,也穿插了大量2024年EWC经典赛事的镜头。
文摘基于学习的多视图立体匹配算法目前成果显著,但是仍然存在的卷积感受野受限以及忽略图像频率信息导致在低纹理、重复和非兰伯曲面匹配性能不足的问题,针对以上问题提出了上下文增强与图像频率引导的多视图立体匹配网络CAF-MVSNet.首先,在特征提取阶段,将上下文增强模块融合到特征金字塔网络中,有效地扩大网络的感受野.然后引入了图像频率引导注意力模块,通过编码图像的不同频率获取图像的线条、形状、纹理和颜色等信息,增强图像的远程上下文联系的同时进一步解决低纹理、重复和非兰伯曲面的精确匹配问题,以实现可靠的特征匹配.在DTU数据集上的实验结果显示,与经典的级联模型Cas MVSNet相比综合误差(overall)提升了12.3%,展现了优秀的性能.此外,在Tanks and Temples数据集上也取得了不错的效果,展现了良好的泛化性能.
文摘The most crucial requirement in radiation therapy treatment planning is a fast and accurate treatment planning system that minimizes damage to healthy tissues surrounding cancer cells. The use of Monte Carlo toolkits has become indispensable for research aimed at precisely determining the dose in radiotherapy. Among the numerous algorithms developed in recent years, the GAMOS code, which utilizes the Geant4 toolkit for Monte Carlo simula-tions, incorporates various electromagnetic physics models and multiple scattering models for simulating particle interactions with matter. This makes it a valuable tool for dose calculations in medical applications and throughout the patient’s volume. The aim of this present work aims to vali-date the GAMOS code for the simulation of a 6 MV photon-beam output from the Elekta Synergy Agility linear accelerator. The simulation involves mod-eling the major components of the accelerator head and the interactions of the radiation beam with a homogeneous water phantom and particle information was collected following the modeling of the phase space. This space was po-sitioned under the X and Y jaws, utilizing three electromagnetic physics mod-els of the GAMOS code: Standard, Penelope, and Low-Energy, along with three multiple scattering models: Goudsmit-Saunderson, Urban, and Wentzel-VI. The obtained phase space file was used as a particle source to simulate dose distributions (depth-dose and dose profile) for field sizes of 5 × 5 cm<sup>2</sup> and 10 × 10 cm<sup>2</sup> at depths of 10 cm and 20 cm in a water phantom, with a source-surface distance (SSD) of 90 cm from the target. We compared the three electromagnetic physics models and the three multiple scattering mod-els of the GAMOS code to experimental results. Validation of our results was performed using the gamma index, with an acceptability criterion of 3% for the dose difference (DD) and 3 mm for the distance-to-agreement (DTA). We achieved agreements of 94% and 96%, respectively, between simulation and experimentation for the three electromagnetic physics models and three mul-tiple scattering models, for field sizes of 5 × 5 cm<sup>2</sup> and 10 × 10 cm<sup>2</sup> for depth-dose curves. For dose profile curves, a good agreement of 100% was found between simulation and experimentation for the three electromagnetic physics models, as well as for the three multiple scattering models for a field size of 5 × 5 cm<sup>2</sup> at 10 cm and 20 cm depths. For a field size of 10 × 10 cm<sup>2</sup>, the Penelope model dominated with 98% for 10 cm, along with the three multiple scattering models. The Penelope model and the Standard model, along with the three multiple scattering models, dominated with 100% for 20 cm. Our study, which compared these different GAMOS code models, can be crucial for enhancing the accuracy and quality of radiotherapy, contributing to more effective patient treatment. Our research compares various electro-magnetic physics models and multiple scattering models with experimental measurements, enabling us to choose the models that produce the most reli-able results, thereby directly impacting the quality of simulations. This en-hances confidence in using these models for treatment planning. Our re-search consistently contributes to the progress of Monte Carlo simulation techniques in radiation therapy, enriching the scientific literature.
