Human Arm Motion Tracking by Inertial/Magnetic Sensors Using Unscented Kalman Filter and Relative Motion Constraint

Human motion tracking has many applications in biomedical and industrial services. Low-cost inertial/magnetic sensors are widely used in human motion capture systems to obtain the orientation of the human body segments. In this paper, we have presented a quaternion-based unscented Kalman filter algorithm to fuse inertial/magnetic sensors measurements for tracking human arm movements. In order to have a better estimation of the orientation of the forearm and the upper arm, a constraint equation was developed based on the relative velocity of the elbow joint with respect to the inertial sensors attached to the forearm and the upper arm. Also to compensate for fast body motions, we adapted the measurement covariance matrix in such a way that the filter implements gyroscopes when large accelerations are involved. The proposed algorithm was evaluated experimentally by an optical tracking system as the ground truth reference. The results showed the effectiveness and good performance of the proposed algorithm.     

单帧图像人体姿态估计综述

人体姿态估计是指从图像中检测人体各部分的位置并计算其方向和尺度信息,姿态估计的结果分二维和三维两种情况,而估计的方法分基于模型和无模型两种途径。本文首先介绍了人体姿态估计的研究背景和应用方向,然后对姿态估计的相关概念作了阐述,分析了姿态估计的输出表示,接着从人体目标检测和姿态估计两大类进行了详细分析和讨论,从实际应用的角度对各种方法做了理论上的比较和分析。最后,对相关研究还存在的问题和进一步研究的趋势作了归纳和总结。     

Multi-view 3D Human Pose Estimation in Complex Environment

We introduce a framework for unconstrained 3D human upper body pose estimation from multiple camera views in complex environment. Its main novelty lies in the integration of three components: single-frame pose recovery, temporal integration and model texture adaptation. Single-frame pose recovery consists of a hypothesis generation stage, in which candidate 3D poses are generated, based on probabilistic hierarchical shape matching in each camera view. In the subsequent hypothesis verification stage, the candidate 3D poses are re-projected into the other camera views and ranked according to a multi-view likelihood measure. Temporal integration consists of computing K-best trajectories combining a motion model and observations in a Viterbi-style maximum-likelihood approach. Poses that lie on the best trajectories are used to generate and adapt a texture model, which in turn enriches the shape likelihood measure used for pose recovery. The multiple trajectory hypotheses are used to generate pose predictions, augmenting the 3D pose candidates generated at the next time step.     

着装场景下双分支网络的人体姿态估计

目的 人体姿态估计旨在识别和定位不同场景图像中的人体关节点并优化关节点定位精度。针对由于服装款式多样、背景干扰和着装姿态多变导致人体姿态估计精度较低的问题,本文以着装场景下时尚街拍图像为例,提出一种着装场景下双分支网络的人体姿态估计方法。方法 对输入图像进行人体检测,得到着装人体区域并分别输入姿态表示分支和着装部位分割分支。姿态表示分支通过在堆叠沙漏网络基础上增加多尺度损失和特征融合输出关节点得分图,解决服装款式多样以及复杂背景对关节点特征提取干扰问题,并基于姿态聚类定义姿态类别损失函数,解决着装姿态视角多变问题;着装部位分割分支通过连接残差网络的浅层特征与深层特征进行特征融合得到着装部位得分图。然后使用着装部位分割结果约束人体关节点定位,解决服装对关节点遮挡问题。最后通过姿态优化得到最终的人体姿态估计结果。结果 在构建的着装图像数据集上验证了本文方法。实验结果表明,姿态表示分支有效提高了人体关节点定位准确率,着装部位分割分支能有效避免着装场景中人体关节点误定位。在结合着装部位分割优化后,人体姿态估计精度提高至92.5%。结论 本文提出的人体姿态估计方法能够有效提高着装场景下的人体姿态估计精度,较好地满足虚拟试穿等实际应用需求。     

Passivity-based Visual Motion Observer with Panoramic Camera for Pose Control

This paper considers the vision-based estimation and pose control with a panoramic camera via passivity approach. First, a hyperbolic projection of a panoramic camera is presented. Next, using standard body-attached coordinate frames (the world frame, mirror frame, camera frame and object frame), we represent the body velocity of the relative rigid body motion (position and orientation). After that, we propose a visual motion observer to estimate the relative rigid body motion from the measured camera data. We show that the estimation error system with a panoramic camera has the passivity which allows us to prove stability in the sense of Lyapunov. The visual motion error system which consists of the estimation error system and the pose control error system preserves the passivity. After that, stability and L ₂-gain performance analysis for the closed-loop system are discussed via Lyapunov method and dissipative systems theory, respectively. Finally, simulation and experimental results are shown in order to confirm the proposed method.     

Arm motion analysis using genetic algorithm for rehabilitation and healthcare

The worlds population is quickly aging. With an aging society, an increase in patients with brain damage is predicted. In rehabilitation, the analysis of arm motion is vital as various day to day activities relate to arm movements. The therapeutic approach and evaluation method are generally selected by therapists based on his/her experience, which can be an issue for quantitative evaluation in any specific movement task. In this paper, we develop a measurement system for arm motion analysis using a 3D image sensor. The method of upper body posture estimation based on a steady-state genetic algorithm (SSGA) is proposed. A continuous model of generation for an adaptive search in dynamical environment using an adaptive penalty function and island model is applied. Experimental results indicate promising results as compared with the literature.     

Human Pose Tracking in Monocular Sequence Using Multilevel Structured Models

Tracking human body poses in monocular video has many important applications. The problem is challenging in realistic scenes due to background clutter, variation in human appearance and self-occlusion. The complexity of pose tracking is further increased when there are multiple people whose bodies may inter-occlude. We proposed a three-stage approach with multi-level state representation that enables a hierarchical estimation of 3D body poses. Our method addresses various issues including automatic initialization, data association, self and inter-occlusion. At the first stage, humans are tracked as foreground blobs and their positions and sizes are coarsely estimated. In the second stage, parts such as face, shoulders and limbs are detected using various cues and the results are combined by a grid-based belief propagation algorithm to infer 2D joint positions. The derived belief maps are used as proposal functions in the third stage to infer the 3D pose using data-driven Markov chain Monte Carlo. Experimental results on several realistic indoor video sequences show that the method is able to track multiple persons during complex movement including sitting and turning movements with self and inter-occlusion.     

Marker-based human pose tracking using adaptive annealed particle swarm optimization with search space partitioning

Pose estimation and tracking of an articulated structure based on data from multiple cameras has seen numerous applications in recent years. In this paper, a marker-based human pose tracking algorithm from multi view video sequences is proposed. The purpose of the proposed algorithm is to present a low cost motion capture system that can be used as an alternative to high cost available commercial human motion capture systems. The problem is defined as the optimization of 45 parameters which define body pose model and is solved using a modified version of particle swarm optimization (PSO) algorithm. The objective of this optimization is to maximize a fitness function which formulates how much the body model matches with 2D marker coordinates in video frames. A sampling covariance matrix is used in the first part of the velocity equation of PSO and is annealed with iterations. The sampling covariance matrix is computed adaptively, based on variance of parameters in the swarm. One of the concerns in this algorithm is the high number of parameters to define the model of body pose. To tackle this problem, we partition the optimization state space into six stages that exploit the hierarchical structure of the skeletal model. The first stage optimizes the six parameters that define the global orientation and position of the body. Other stages relate to optimization of right and left hand, right and left leg and head orientation. In the proposed partitioning method previously optimized parameters are allowed some variation in each step that is called soft partitioning. Experimental results on Pose Estimation and Action Recognition (PEAR) database indicate that the proposed algorithm achieves lower estimation error in tracking human motion compared with Annealed Particle Filter (APF) and Parametric Annealing (PA) methods.     

Error growth in position estimation from noisy relative pose measurements

We examine how the estimation error grows with time when a mobile robot estimates its location from relative pose measurements without global position or orientation sensors. We show that, in both two-dimensional and three-dimensional space, both the bias and the variance of the position estimation error grows at most linearly with time asymptotically. Non-asymptotic bounds on the bias and variance are obtained, which provide insight into the mechanism of error growth. The bias is crucially dependent on the trajectory of the robot. Conclusions on the asymptotic growth rate of the bias continue to hold even with unbiased measurements or error-free translation measurements. Exact formulas for the bias and the variance of the position estimation error are provided for two specific two-dimensional trajectories–straight line and periodic. Experiments with a P3-DX wheeled robot and Monte Carlo simulations are provided to verify the theoretical predictions. A method to reduce the bias is proposed based on the lessons learned.