采用运动恢复结构(structure from motion,SFM)算法进行三维人脸建模一直以来受到研究者的关注,但其对错误的匹配点比较敏感,因此,文章提出了一种融合Gabor特征的SFM算法三维人脸建模方法。该方法利用Gabor滤波器提取纹理特征,判别轮廓...采用运动恢复结构(structure from motion,SFM)算法进行三维人脸建模一直以来受到研究者的关注,但其对错误的匹配点比较敏感,因此,文章提出了一种融合Gabor特征的SFM算法三维人脸建模方法。该方法利用Gabor滤波器提取纹理特征,判别轮廓特征点匹配的准确性;针对图像数增多,传统因子分解法不易修正旋转矩阵的问题,利用旋转矩阵的性质求得修正矩阵,避开方程组的求解;提出引入迭代最近点算法将稀疏三维特征点与三维模型进行配准,缩小空间距离,并结合薄板样条函数插值生成特定的三维人脸模型,为增强真实感,进行纹理映射。实验结果表明,该方法有效提高了匹配点的准确性,能够重建出具有较强真实感的三维人脸。展开更多
高精度、高分辨率的地形地貌数据是活动构造定量研究的基础。摄影测量方法的出现和快速发展为获取高精度地形地貌数据提供了一种经济有效的技术手段。相比于传统的测量方法,摄影测量方法可在大范围内同时进行,不受地面通视条件的限制,...高精度、高分辨率的地形地貌数据是活动构造定量研究的基础。摄影测量方法的出现和快速发展为获取高精度地形地貌数据提供了一种经济有效的技术手段。相比于传统的测量方法,摄影测量方法可在大范围内同时进行,不受地面通视条件的限制,且测量成本相对较低。尤其近年来,随着计算机视觉理论及高效的自动特征匹配算法的发展,一种名为"Structure from Motion"(SfM)的三维重建技术被引入摄影测量方法中,极大地提高了摄影测量的自动化程度。文中介绍了摄影测量方法的基本原理及发展历程,并综述了摄影测量方法在活动构造研究中的应用,最后通过SfM摄影测量方法在活动构造研究中的1个具体应用实例,展示了摄影测量方法在活动构造定量研究中的巨大应用潜力。展开更多
为研究视觉SLAM(Simutaneous Localization and Mapping)以及视觉SFM(Structure From Motion)领域中机器人能低成本重建出可进行语义识别三维场景的问题,借鉴Colmap算法,对重建过程中的部分步骤加以改进,对初始化图像的选择进行了精选,...为研究视觉SLAM(Simutaneous Localization and Mapping)以及视觉SFM(Structure From Motion)领域中机器人能低成本重建出可进行语义识别三维场景的问题,借鉴Colmap算法,对重建过程中的部分步骤加以改进,对初始化图像的选择进行了精选,在重建出有误差的三维点云中进行多次基于光束平差法的优化,对无人机拍摄的图像进行重建分析,结果表明,图像在重建的完整度、鲁棒性、精确度以及效率等方面,均取得了较好的效果。展开更多
The spatial distribution of interrill and rill erosion is essential for unravelling soil erosion principles and the application of soil and water conservation practices.To quantify interrill and rill erosion and their...The spatial distribution of interrill and rill erosion is essential for unravelling soil erosion principles and the application of soil and water conservation practices.To quantify interrill and rill erosion and their spatial development,four 30-min rainfalls at 90 mm h^(-1)intensity were consecutively simulated on runoff plots packed with a loess at six slopes of 10°,15°,20°,25°,30°and 35°.The soil surface was measured using the structure from motion(SfM)photogrammetry upon each simulation run,and the runoff and sediment samples were collected and measured at every 10 min.Rills did not develop until the third simulation run.During the initial two runs,the lower third section was more severely eroded than the upper and middle thirds along the slope direction,yet the interrill erosion was statistically uniform from left to right.Rills tended to emerge by both sidewalls and in the lower portion in the third run.The corresponding rill erosion increased with slope from 10°to 20°and then decreased for the slopes steeper,which was consistent with the slope trend of the sediment yield directly measured.The rills expanded substantially primarily via head retreat and to a lesser extent via sideward erosion after receiving another 30-min rainfall.Rill erosion contributed 69.3%of the total erosion loss,and shifted the critical slope corresponding to the maximum loss from 20°to 25°.These findings demonstrate the significance of rill erosion not only in total soil loss but also in its relation to slope,as well as the effectiveness of SfM photogrammetry in quantifying interrill and rill erosion.展开更多
文摘采用运动恢复结构(structure from motion,SFM)算法进行三维人脸建模一直以来受到研究者的关注,但其对错误的匹配点比较敏感,因此,文章提出了一种融合Gabor特征的SFM算法三维人脸建模方法。该方法利用Gabor滤波器提取纹理特征,判别轮廓特征点匹配的准确性;针对图像数增多,传统因子分解法不易修正旋转矩阵的问题,利用旋转矩阵的性质求得修正矩阵,避开方程组的求解;提出引入迭代最近点算法将稀疏三维特征点与三维模型进行配准,缩小空间距离,并结合薄板样条函数插值生成特定的三维人脸模型,为增强真实感,进行纹理映射。实验结果表明,该方法有效提高了匹配点的准确性,能够重建出具有较强真实感的三维人脸。
文摘高精度、高分辨率的地形地貌数据是活动构造定量研究的基础。摄影测量方法的出现和快速发展为获取高精度地形地貌数据提供了一种经济有效的技术手段。相比于传统的测量方法,摄影测量方法可在大范围内同时进行,不受地面通视条件的限制,且测量成本相对较低。尤其近年来,随着计算机视觉理论及高效的自动特征匹配算法的发展,一种名为"Structure from Motion"(SfM)的三维重建技术被引入摄影测量方法中,极大地提高了摄影测量的自动化程度。文中介绍了摄影测量方法的基本原理及发展历程,并综述了摄影测量方法在活动构造研究中的应用,最后通过SfM摄影测量方法在活动构造研究中的1个具体应用实例,展示了摄影测量方法在活动构造定量研究中的巨大应用潜力。
文摘为研究视觉SLAM(Simutaneous Localization and Mapping)以及视觉SFM(Structure From Motion)领域中机器人能低成本重建出可进行语义识别三维场景的问题,借鉴Colmap算法,对重建过程中的部分步骤加以改进,对初始化图像的选择进行了精选,在重建出有误差的三维点云中进行多次基于光束平差法的优化,对无人机拍摄的图像进行重建分析,结果表明,图像在重建的完整度、鲁棒性、精确度以及效率等方面,均取得了较好的效果。
基金The study was funded by the National Natural Science Foundation of China(No.42130701,41601277,41571130082)The authors also appreciate the technical support from the Rainfall Simulation Hall of the Fangshan Experimental Field Station of the State Key Laboratory of Earth Surface Processes and Resource Ecology。
文摘The spatial distribution of interrill and rill erosion is essential for unravelling soil erosion principles and the application of soil and water conservation practices.To quantify interrill and rill erosion and their spatial development,four 30-min rainfalls at 90 mm h^(-1)intensity were consecutively simulated on runoff plots packed with a loess at six slopes of 10°,15°,20°,25°,30°and 35°.The soil surface was measured using the structure from motion(SfM)photogrammetry upon each simulation run,and the runoff and sediment samples were collected and measured at every 10 min.Rills did not develop until the third simulation run.During the initial two runs,the lower third section was more severely eroded than the upper and middle thirds along the slope direction,yet the interrill erosion was statistically uniform from left to right.Rills tended to emerge by both sidewalls and in the lower portion in the third run.The corresponding rill erosion increased with slope from 10°to 20°and then decreased for the slopes steeper,which was consistent with the slope trend of the sediment yield directly measured.The rills expanded substantially primarily via head retreat and to a lesser extent via sideward erosion after receiving another 30-min rainfall.Rill erosion contributed 69.3%of the total erosion loss,and shifted the critical slope corresponding to the maximum loss from 20°to 25°.These findings demonstrate the significance of rill erosion not only in total soil loss but also in its relation to slope,as well as the effectiveness of SfM photogrammetry in quantifying interrill and rill erosion.
