Articulated Motion Estimation With Moving Camera Decoupling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately decouple the motion of articulated objects from the motion of a moving camera in video, leading to entangled estimations that hinder precise motion analysis.
Innovation Solution
A method using a neural network to jointly estimate the motion of a camera and articulated objects in video, employing supervised gradient descent and global optimizations to refine the initial estimation, incorporating human motion priors and background scene features to disentangle the motions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If motion estimation is performed on video captured by a moving camera, then the ability to analyze articulated object motion is improved, but the accuracy of motion estimation deteriorates due to entangled camera and object motions
Solution Approach 1:
The patent segments the motion estimation problem into two independent components: camera motion estimation and articulated object motion estimation. By separately estimating camera pose (position and orientation) and object pose, the method decouples the entangled motions that occur in dynamic video capture scenarios, allowing each component to be optimized independently for accuracy.
Solution Approach 2:
The patent introduces an intermediary coordinate system (world coordinate system) that serves as a reference frame to separate camera motion from object motion. By transforming both camera and object positions into this common reference frame, the system can independently track each entity's motion relative to the world, rather than relative to each other, thereby resolving the entanglement problem.
2Productivity
If current motion estimation solutions are applied to dynamic camera videos, then motion analysis can be performed, but the decoupling of camera and articulated object motion fails
Solution Approach 1:
The patent employs dynamic pose estimation that adapts to changing camera conditions throughout video sequences. The system continuously updates camera pose and object pose estimates frame-by-frame, allowing it to handle dynamic scenarios where camera motion varies over time. This dynamic approach maintains reliable decoupling even when camera movement patterns change, unlike static methods that assume fixed camera conditions.
Solution Approach 2:
The patent implements a feedback mechanism where the estimated camera pose is used to correct and refine the articulated object motion estimation, and vice versa. By iteratively adjusting both estimates based on their mutual consistency with observed video data, the system improves the reliability of motion decoupling, ensuring that neither camera nor object motion is incorrectly attributed to the other.
Data Source
AI summary
Estimating motion of a human or other object in video is a common computer task with applications in robotics, sports, mixed reality, etc. However, motion estimation becomes difficult when the camera capturing the video is moving, because the observed object and camera motions are entangled. The present disclosure provides for joint estimation of the motion of a camera and the motion of articulated objects captured in video by the camera.