Fast Motion Synthesis of Quadrupedal Animals Using a Minimum Amount of Motion Capture Data

This paper introduces a novel and fast synthesizing method for 3D motions of quadrupedal animals that uses only a small set of motion capture data. Unlike human motions, animal motions are relatively difficult to capture. Also, it is a challenge to synthesize continuously changing animal motions in real time because animals have various gait types according to their speed. The algorithm proposed herein, however, is able to synthesize continuously varying motions with proper limb configuration by using only one single cyclic animal motion per gait type based on the biologically driven Froude number. During the synthesis process, each gait type is automatically determined by its speed parameter, and the transition motions, which have not been entered as input, are synthesized accordingly by the optimized asynchronous motion blending technique. At the start time, given the user's control input, the motion path and spinal joints for turning are adjusted first and then the motion is stitched at any speed with proper transition motions to synthesize a long stream of motions.     

Transferring visual prior for online object tracking

Visual prior from generic real-world images can be learned and transferred for representing objects in a scene. Motivated by this, we propose an algorithm that transfers visual prior learned offline for online object tracking. From a collection of real-world images, we learn an overcomplete dictionary to represent visual prior. The prior knowledge of objects is generic, and the training image set does not necessarily contain any observation of the target object. During the tracking process, the learned visual prior is transferred to construct an object representation by sparse coding and multiscale max pooling. With this representation, a linear classifier is learned online to distinguish the target from the background and to account for the target and background appearance variations over time. Tracking is then carried out within a Bayesian inference framework, in which the learned classifier is used to construct the observation model and a particle filter is used to estimate the tracking result sequentially. Experiments on a variety of challenging sequences with comparisons to several state-of-the-art methods demonstrate that more robust object tracking can be achieved by transferring visual prior.     

Indexing 3-D Human Motion Repositories for Content-Based Retrieval

Content-based retrieval of the similar motions for the human joints has significant impact in the fields of physical medicine, biomedicine, rehabilitation, and motion therapy. In this paper, we propose an efficient indexing approach for 3-D human motion capture data, supporting queries involving both subbody motions as well as whole-body motions.     

Improving multiple pedestrians tracking with semantic information

This work presents an interacting multiple pedestrian tracking method for monocular systems that incorporates a prior knowledge about the environment and about interactions between targets. Pedestrian motion being ruled by both environment and social aspects, we model these complex behaviors by considering four cases of motion: going straight, finding one’s way, walking around and standing still. They are combined within an interacting multiple model particle filter strategy. We model targets interactions with social forces included as potential functions in the weighting process of the particle filter. We use different social force setups within each motion model to handle high-level behaviors (collision avoidance, flocking...). We evaluate our algorithm on challenging datasets and show that such semantic information improves the tracker performance compared to the literature.     

基于三维递归搜索的多级运动估计视频帧率上转换算法

运动补偿插帧是目前主要的帧率上转换方法。为减小内插帧中的块效应,并降低运算量以满足实时高清视频应用,该文提出了一种基于3维递归搜索(3-D Recursive Search, 3-D RS)的多级块匹配运动估计视频帧率上转换算法。该算法将3-D RS与双向运动估计相结合,首先对序列中相邻帧进行由粗到精的三级运动估计,再利用简化的中值滤波器平滑运动矢量场,最后通过线性插值补偿得到内插帧。实验结果表明,与现有的运动补偿插帧算法相比,该算法内插帧的主、客观质量都有所提高,且算法复杂度低,有很强的实用性。     

Multicamera Tracking of Articulated Human Motion Using Shape and Motion Cues

We present a completely automatic algorithm for initializing and tracking the articulated motion of humans using image sequences obtained from multiple cameras. A detailed articulated human body model composed of sixteen rigid segments that allows both translation and rotation at joints is used. Voxel data of the subject obtained from the images is segmented into the different articulated chains using Laplacian Eigenmaps. The segmented chains are registered in a subset of the frames using a single-frame registration technique and subsequently used to initialize the pose in the sequence. A temporal registration method is proposed to identify the partially segmented or unregistered articulated chains in the remaining frames in the sequence. The proposed tracker uses motion cues such as pixel displacement as well as 2-D and 3-D shape cues such as silhouettes, motion residue, and skeleton curves. The tracking algorithm consists of a predictor that uses motion cues and a corrector that uses shape cues. The use of complementary cues in the tracking alleviates the twin problems of drift and convergence to local minima. The use of multiple cameras also allows us to deal with the problems due to self-occlusion and kinematic singularity. We present tracking results on sequences with different kinds of motion to illustrate the effectiveness of our approach. The pose of the subject is correctly tracked for the duration of the sequence as can be verified by inspection.      

基于无载波超宽带雷达的小样本人体动作识别

随着雷达硬件平台尺寸越来越小、成本越来越低,室内基于雷达的人体动作识别应用已经成为现实,能够在具有简单架构的低成本设备中实现.无载波超宽带雷达具有极高的分辨力,能够捕获人体细微动作变化并且对室内复杂环境具有很强的抗干扰能力.与基于视频人体动作识别研究相比,超宽带雷达还具有穿透家具、墙体以及保护个人隐私等优点.针对雷达回波信号利用传统时频分析方法实现人体动作识别比较耗时、实时性不好的缺陷,引入机器学习方法对不同类型人体动作进行分类识别.引入机器学习方法用于超宽带雷达人体动作识别最大难点是只有少量可用的超宽带雷达实测数据样本,针对该问题提出基于主成分分析法（PCA）和离散余弦变换（DCT）相结合的人体动作特征提取方法,并利用改进网格搜索算法优化的支持向量机在小样本数据下对人体动作进行识别,最后根据实测数据采取三种不同方案进行仿真实验,结果表明即使在训练数据样本只有5组的条件下,基于PCA和DCT相结合特征提取方法在不同类型人体动作的平均识别率均能达到96%以上.     

Motion Capture of the Human Body Using Multiple Depth Sensors

The movements of the human body are difficult to capture owing to the complexity of the three‐dimensional skeleton model and occlusion problems. In this paper, we propose a motion capture system that tracks dynamic human motions in real time. Without using external markers, the proposed system adopts multiple depth sensors (Microsoft Kinect) to overcome the occlusion and body rotation problems. To combine the joint data retrieved from the multiple sensors, our calibration process samples a point cloud from depth images and unifies the coordinate systems in point clouds into a single coordinate system via the iterative closest point method. Using noisy skeletal data from sensors, a posture reconstruction method is introduced to estimate the optimal joint positions for consistent motion generation. Based on the high tracking accuracy of the proposed system, we demonstrate that our system is applicable to various motion‐based training programs in dance and Taekwondo.     

基于双隐变量空间局部粒子搜索的人体运动形态估计

该文提出一种双隐变量空间局部粒子搜索(DLVSLPS)算法,可以从多视角图像序列的轮廓特征较准确地估计出3维人体运动形态序列。该算法用高斯过程动态模型(GPDM)降维建立双隐变量空间和低维隐变量数据到高维数据的映射关系后,然后对双隐变量空间使用近邻权重先验条件搜索(NWPCS),实现局部低维粒子搜索来生成较优高维数据,从而估计相应帧的3维人体运动形态,解决传统粒子滤波算法直接在高维数据空间采样较难获取有效正确数据进行估计的问题。经仿真实验验证,所提出的算法比传统粒子滤波算法在实现多视角非连续帧估计,克服轮廓图像数据歧义,减小估计误差有明显优势。     

基于MPEG压缩域的快速、鲁棒的全局运动信息提取方法

全局运动信息在视频分析中起着重要的作用.本文根据MPEG编码特点,提出了一种从MPEG压缩域中快速有效地进行全局运动参数估计的算法.该算法充分利用了MPEG压缩码流中的信息,通过提取预测残差DC图像的运动背景区域,估计全局运动参数,从而保证了参数估计的准确性,有效地克服了已有文献中仅仅采用运动矢量进行全局运动估计的局限性.根据不同的MPEG测试序列的对比分析,结果表明,本算法可快速准确地对MPEG视频序列进行全局运动信息估计,同时具有很高的鲁棒性.     

基于Kinect的人体动作识别算法研究

针对现有的复杂背景下人体动作识别中存在识别准确率不高和实时性不强等问题,提出基于Kinect骨骼数据的改进动作识别算法。通过Kinect获取骨骼数据,提取出人体关节的特征向量,然后用模板匹配的方法对人体动作进行识别。通过搭建机器人体感控制系统验证了算法的可行性。在相同实验条件下测得算法的平均识别率为95.2%,平均识别时间为32.5ms。与其它动作识别算法比较,证明了算法的识别率较高、实时性较好。     

基于高斯增量降维与流形Boltzmann优化的人体运动形态估计

为了从多视角轮廓图像估计出含空间位置信息的三维人体运动形态,该文提出高斯增量降维与流形Boltzmann优化（GIDRMBO）算法.该算法把表示三维人体运动形态的高维数据分成表示空间位置信息和姿态信息两段子向量后,用高斯增量降维模型（GIDRM）分别对其样本进行降维,建立相应的低维空间及映射关系,然后在相应的低维空间使用流形Boltzmann优化算法来对轮廓匹配目标函数进行优化,从而实现估计.其中,所提算法分别利用了两段子向量样本的低维数据作为先验信息,可较好的避免陷入局部最优区域进行搜索,最终生成与各视角原始运动图像匹配且含空间位置信息的三维人体运动形态.经仿真实验验证,所提算法与常用粒子滤波算法相比,其估计误差小,并且还能起到消除轮廓数据歧义和克服短时遮挡的作用.     

Alignment of 3-D optical coherence tomography scans to correct eye movement using a particle filtering

Eye movement artifacts occurring during 3-D optical coherence tomography (OCT) scanning is a well-recognized problem that may adversely affect image analysis and interpretation. A particle filtering algorithm is presented in this paper to correct motion in a 3-D dataset by considering eye movement as a target tracking problem in a dynamic system. The proposed particle filtering algorithm is an independent 3-D alignment approach, which does not rely on any reference image. 3-D OCT data is considered as a dynamic system, while the location of each A-scan is represented by the state space. A particle set is used to approximate the probability density of the state in the dynamic system. The state of the system is updated frame by frame to detect A-scan movement. The proposed method was applied on both simulated data for objective evaluation and experimental data for subjective evaluation. The sensitivity and specificity of the x-movement detection were 98.85% and 99.43%, respectively, in the simulated data. For the experimental data (74 3-D OCT images), all the images were improved after z-alignment, while 81.1% images were improved after x-alignment. The proposed algorithm is an efficient way to align 3-D OCT volume data and correct the eye movement without using references.     

Super-Resolution Without Explicit Subpixel Motion Estimation

The need for precise (subpixel accuracy) motion estimates in conventional super-resolution has limited its applicability to only video sequences with relatively simple motions such as global translational or affine displacements. In this paper, we introduce a novel framework for adaptive enhancement and spatiotemporal upscaling of videos containing complex activities without explicit need for accurate motion estimation. Our approach is based on multidimensional kernel regression, where each pixel in the video sequence is approximated with a 3-D local (Taylor) series, capturing the essential local behavior of its spatiotemporal neighborhood. The coefficients of this series are estimated by solving a local weighted least-squares problem, where the weights are a function of the 3-D space-time orientation in the neighborhood. As this framework is fundamentally based upon the comparison of neighboring pixels in both space and time, it implicitly contains information about the local motion of the pixels across time, therefore rendering unnecessary an explicit computation of motions of modest size. The proposed approach not only significantly widens the applicability of super-resolution methods to a broad variety of video sequences containing complex motions, but also yields improved overall performance. Using several examples, we illustrate that the developed algorithm has super-resolution capabilities that provide improved optical resolution in the output, while being able to work on general input video with essentially arbitrary motion.      

姿态估计和跟踪结合的运动视频关键帧提取

提出姿态估计和特定部位跟踪相结合的动作视频关键帧提取算法.首先利用非确定人体部位的时间连续性保持提高基于柔性部件铰接人体模型的单帧图像人体姿态估计准确率,通过实施数据降维得到局部拓扑结构表达能力强的判别性运动特征向量,采用极值判定原理确定候选关键帧集合.然后利用ISODATA动态聚类算法,通过初始聚类中心优化、基于语义的关键帧集合增强等策略确定关键帧.实验表明文中算法具有较高的关键帧提取准确率和召回率,支持基于语义的关键帧提取.提取的视频关键帧可以用于运动视频压缩和批注审阅.     

基于卷积神经网络的响应自适应跟踪

针对目标跟踪中的遮挡、旋转、快速运动、形变等问题,本文提出基于卷积神经网络的响应自适应跟踪算法。首先,通过卷积神经网络提取目标的多层卷积特征,利用粒子滤波算法获取目标的多模板响应图,自适应学习目标的期望响应;然后通过构造目标函数的对偶形式解决多模板联合优化问题,计算多模板情况下每层卷积特征的最优滤波参数;最后通过相关滤波算法计算多层滤波响应,通过响应加权融合的方式计算最终响应图,以此估计目标位置。本文利用OTB-2013数据集中的方法测试我们提出的算法,实验表明该算法的整体成功率和精确度为0.884和0.915。本文算法在距离准确度、成功率和平均跟踪误差方面均优于传统的相关滤波跟踪算法,有一定研究价值。     

Variable structure Human Intention Estimator with mobility and vision constraints as model selection criteria

In this paper, a novel method for early estimation of a human’s action intention is presented. Human intention, modeled as a goal location associated with a hand motion and eye gaze dynamics, is inferred by fusing information from collected hand motion and gaze motion data. The algorithm, called Human Intention Estimator with Variable Structure (HIEVS), uses two variable structure Interacting Multiple Model (VS-IMM) filters in parallel to process the hand motion and gaze data and generate posterior model probabilities associated with a finite set of action models. The posterior model probabilities from each filter are then fused at the end of each iteration and the current intention is estimated as the model, which has the highest fused posterior model probability. Two model set augmentation (MSA) algorithms are presented to select the active models for each VS-IMM during each iteration. For the hand motion filter, an MSA algorithm which computes the human’s reachable workspace is used. The MSA algorithm for the gaze filter utilizes the human’s visual span to determine the active models. This method allows for accurate early prediction of the human’s intention even when the total model set is large. A real world experiment is performed to validate the proposed method.     

Motion estimation in the 3-D Gabor domain.

Motion estimation methods can be broadly classified as being spatiotemporal or frequency domain in nature. The Gabor representation is an analysis framework providing localized frequency information. When applied to image sequences, the 3-D Gabor representation displays spatiotemporal/spatiotemporal-frequency (st/stf) information, enabling the application of robust frequency domain methods with adjustable spatiotemporal resolution. In this work, the 3-D Gabor representation is applied to motion analysis. We demonstrate that piecewise uniform translational motion can be estimated by using a uniform translation motion model in the st/stf domain. The resulting motion estimation method exhibits both good spatiotemporal resolution and substantial noise resistance compared to existing spatiotemporal methods. To form the basis of this model, we derive the signature of the translational motion in the 3-D Gabor domain. Finally, to obtain higher spatiotemporal resolution for more complex motions, a dense motion field estimation method is developed to find a motion estimate for every pixel in the sequence.     

基于刚体几何不变性的雷达目标运动路径拟合和三维优化重建方法

对于飞机、船舰等刚体雷达目标,其在运动过程中具有空间几何不变性。利用这一约束条件,可以通过雷达回波中提取出的目标散射点的一维距离史重建出目标的三维形状和运动路径。鉴于现有的基于几何不变性的雷达目标三维重建算法存在鲁棒性差的问题,本文利用雷达目标的运动惰性,对初步重建后得到的目标运动路径进行了拟合,并利用拟合后的运动路径对目标散射点的三维坐标进行了优化重建。文中对重建的误差进行了分析,提出了仿射扰动和欧式重建误差的误差模型。仿真实验证明,经仿射匹配校正后的拟合路径与真实路径基本吻合,从而可以有效获得目标的运动特征;同时,利用本文提出的优化重建方法能够有效抑制目标的欧式重建误差,提高了重建算法的准确性。      

Joint Blind Equalization and Decoding in OFDM Systems Using Particle Filtering

In this paper a sequential algorithm is proposed for joint blind channel equalization and decoding for orthogonal frequency-division multiplexing (OFDM) in frequency selective channels. This algorithm offers a recursive method to sequentially calculate the posterior probability for maximum a posteriori detection. Recursive calculations are done along the indexes in each OFDM symbol using a particle filter. By defining an appropriate importance function, and a proper prior probability distribution function for the channel tap coefficients (and marginalizing it), an efficient method is presented for joint equalization and channel decoding in OFDM based systems. Performance of the proposed detector is evaluated using computer simulations and its bit error rate is compared with the trained turbo equalizer and a conventional particle filter-based method. The results show that the proposed method outperforms the previously presented particle filter-based method without a need for training data.