The measurement and interpretation of 2D and 3D image motion

2D 和 3D 图像运动的测量和解释

• 批准号: 41810-2006
• 负责人: Barron, John
• 金额: $ 2.33万
• 依托单位: University of Western Ontario
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2007
• 资助国家: 加拿大
• 起止时间: 2007-01-01 至 2008-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=364613
• 关键词: measurement; interpretation; 2D; 3D; image

The central theme of Dr. Barron's research deals with the measurement and interpretation of 2D and 3D image image and 3D range motion. 2D image motion, approximated by 2D optical flow, is the local 2D motion at each image point in an image sequence. 3D image motion, approximated by 3D optical flow, is the volumetric motions of voxels in volume sequences. 3D Range motion, approximated by range flow, is the local 3D motion at surface points (and as such can be considered 3D optical flow on a surface). Some of the applications of his research could be: (1) Recovery of camera translation and rotation parameters and scene depth maps from a sequence of images generated by the moving camera using time-varying 2D optical flow in a Kalman filtering framework. (2) Measurement of 2D/3D corn seedling growth from 2D optical flow (including in the dark using near-infrared imagery). (3) Measurement of leaf motion and growth using 3D range flow computed from time-varying 3D surface range datsets. (4) Measurement of voxel motion in sequences of gated MRI cardiac datasets leading to analysis, say, of the motion of the left ventricular heart wall. (5) Tracking of severe weather storms in 3D Doppler precipitation density and radial velocity datasets. This research contributes to the areas of autonomous vehicle navigation, monitoring plant growth in controlled environments (greenhouses), diagnosis of heart disease, and automation of Doppler radar weather nowcasting/forecasting.

巴伦博士研究的中心主题涉及2D和3D图像和3D范围运动的测量和解释。2D图像运动，由2D光流近似，是图像序列中每个图像点处的局部2D运动。3D图像运动由3D光流近似，是体积序列中体素的体积运动。3D范围运动，近似为范围流，是表面点处的局部3D运动（因此可以被认为是表面上的3D光流）。他的研究的一些应用可以是：（1）在卡尔曼滤波框架中使用时变2D光流从移动相机生成的图像序列中恢复相机平移和旋转参数以及场景深度图。(2)从2D光流测量2D/3D玉米幼苗生长（包括在黑暗中使用近红外成像）。(3)利用从随时间变化的三维表面范围数据集计算的三维范围流测量叶运动和生长。(4)测量门控MRI心脏数据集序列中的体素运动，从而分析左心室心脏壁的运动。(5)在三维多普勒降水密度和径向速度数据集中对严重天气风暴的跟踪。这项研究有助于自动驾驶汽车导航，监测植物生长在受控环境（温室），心脏病的诊断，以及多普勒雷达天气临近预报/预报的自动化等领域。