快速的三维人手运动跟踪方法研究

三维人手运动跟踪是人机交互领域的一个重要研究方向.提出了一种新的基于模型的三维人手运动跟踪方法,该方法将层次优化嵌入到基于粒子滤波器的跟踪框架中,通过在隐状态空间中对粒子采样来提高粒子滤波器采样效率.首先,提出了采用低维隐状态来描述人手的配置状态,并根据人手的生理运动约束建立人手动态模型;其次,为提高粒子在隐状态空间的采样效率,提出了采用层次遗传优化来快速地在局部寻找好的粒子,并以此作为重要度采样函数修正粒子滤波的采样算法.实验结果表明,该方法可以在人手自遮挡存在时的复杂背景下快速地对人手运动进行跟踪.     

基于二维测地线距离保持映射的人脸识别

人脸空间是嵌套在高维观测空间中的低维流形,为了更好地描述人脸空间的凸起和凹进等细微结构,提出了一种基于二维测地线距离保持映射的人脸识别算法。算法采用矩阵的模式表示人脸空间中的样本图像;基于图像的矩阵表示模型,采用二维测地线距离保持映射算法计算人脸空间的低维嵌套流形;以人脸样本在低维流形空间中的投影为特征进行人脸识别。在CMU PIE人脸数据库上的实验结果验证了算法的合理性和有效性。     

基于局部保持映射的图像隐密检测算法

提出了一种新的基于局部保持映射（Locality Preserving Projections,LPP）降维的图像隐密检测方案。为降低图像特征向量的维数,同时保持其内在低维结构,方便构造更有效的分类器,在经过小波变换形成图像特征后,利用LPP算法得到图像特征集的低维流形,实现对图像高维特征的降维。进而使用支持向量机（SVM）算法将降维后的特征映射到分类特征空间,实现对正常图像和隐密图像分类。实验结果表明,与不采用降维算法的检测方案相比,提出的方案能够显著地提高检测的准确率。     

以多模型融合为特征的三维手势跟踪算法

三维手势跟踪是基于手势交互中的一个基础性研究课题,要实时、高精度地实现三维手势跟踪是一个具有挑战性的热点问题.为提高三维人手跟踪的精确度,提出一种基于多模型融合状态预测的粒子滤波跟踪算法.首先通过对基于数字手套的虚拟烤箱系统进行实验,结合人手的行为理解和描述建立了基于认知实验的人手运动状态预测模型;其次对人手跟踪过程中的数据建立人手运动模型,将Sigma点原理应用到人手模型数据上,得到基于局部分析预测模型;最后将这2个模型按照其与当前帧图像的相似度进行融合,得到粒子滤波过程中的状态预测模型.与退火算法相比较的实验结果表明,在运行时间基本相同的情况下,该算法通过改善粒子滤波过程中状态预测的精度提高了人手跟踪过程中的精度.     

保局投影在图像隐密检测中的应用

为解决图像隐密检测中图像特征维数过高导致的"维数灾难"问题,在保持图像特征内在低维结构的基础上降低特征向量的维数,方便构造更有效的分类器,提出了一种基于保局投影(locality preserving projections,LPP)降维的图像隐密检测算法,对待测图像进行小波变换形成图像特征后,利用LPP算法实现对图像高维特征的降维,得到图像特征集的低维流形.使用支持向量机(SVM)算法将降维后的特征映射到分类特征空间,实现对正常图像和隐密图像分类.实验结果表明,与不使用降维算法的检测方案相比,基于LPP降维的检测算法能够显著地提高检测的准确率.     

基于双向非线性学习的轨迹跟踪和识别

目标的运动轨迹是跟踪和识别目标行为的重要特征之一,在视觉跟踪等领域得到了广泛的应用.然而,由于轨迹数据具有高维和非线性等特点,因而直接建模目标的运动轨迹比较困难.为此,引入一种称为自编码(autoencoder)的双向深层神经网络,并结合粒子滤波提出一种轨迹跟踪识别算法.首先,自编码网络按照一定的学习规则将高维轨迹嵌人到二维平面上,通过该网络的逆向映射得到轨迹的生成模型,由轨迹生成模型可得到一系列可行性轨迹.跟踪过程中,每时刻粒子滤波器的粒子便从这些可行性轨迹中进行抽样,并利用颜色似然函数对抽取的粒子进行加权以及再抽样从而实现对目标状态的估计,最后在二维平面中利用"最小距离分类器"对跟踪轨迹进行识别.特别地,自编码网络提供了高维轨迹空间和低维嵌套结构的双向映射,有效解决了大多数非线性降维方法(例如局部线性嵌入算法(LLE)和等度规映射(ISOMAP))所不具备的逆向映射问题.跟踪和识别手写数字实验表明所提出的方法能在复杂背景下精确跟踪目标并正确识别目标轨迹.     

基于动态序列图像的流形学习研究

动态场景的外形或表观在很大程度上往往受到一个潜在低维动态过程的控制。基于视频序列之间的时间相干特性,引入一种称为自编码(autoencoder)的特殊双向深层神经网络,采用CRBM（continuous restricted Boltzmann machine）的网络结构,用来学习序列图像的低维流形结构。将autoencoder 用于人体步态序列的实验表明,该方法能提供从高维视频帧到具有一定物理意义过程的低维序列的映射,并能从低维描述中恢复高维图像序列。     

一种基于人工神经网络的快速超分辨率方法

图像的超分辨率是指利用一幅或者多幅的低分辨率图像,通过相应的算法来获得一幅对应的清晰的高分辨率图像.针对现有的基于学习的超分辨率方法低效率的问题,提出一种基于人工神经网络的快速超分辨率方法,该方法试图利用人工神经网络学习到高分辨率图像和低分辨率图像之间的函数关系,其理论基础来自于人工神经网络能够很好地求解流形学习中高维流形和低维流形之间的映射关系.     

一种在源数据稀疏情况下的流形学习算法研究

传统的流形学习算法能有效地学习出高维采样数据的低维嵌入坐标,但也存在一些不足,如不能处理稀疏的样本数据.针对这些缺点,提出了一种基于局部映射的直接求解线性嵌入算法(Solving Directly Linear Embedding,简称SDLE).通过假定低维流形的整体嵌入函数,将流形映射赋予局部光滑的约束,应用核方法将高维空间的坐标投影到特征空间,最后构造出在低维空间的全局坐标.SDLE算法解决了在源数据稀疏情况下的非线性维数约简问题,这是传统的流形学习算法没有解决的问题.通过实验说明了SDLE算法研究的有效性.     

基于Isomap的流形结构重建方法

已有的流形学习方法仅能建立点对点的降维嵌入,而未建立高维数据流形空间与低维表示空间之间的相互映射.此缺陷已限制了流形学习方法在诸多数据挖掘问题中的进一步应用.针对这一问题,文中提出了两种新型高效的流形结构重建算法:快速算法与稳健算法.其均以经典的Isomap方法内在运行机理为出发点,进而推导出高维流形空间与低维表示空间之间双向的显式映射函数关系,基于此函数即可实现流形映射的有效重建.理论分析与实验结果证明,所提算法在计算速度、噪音敏感性、映射表现等方面相对已有方法具有明显优势.     

利用共享动态隐变量模型估计三维人体运动

可视媒体中的基于学习的三维人体运动分析是计算机视觉领域中非常有挑战性的课题.本文在Gauss动态隐变量模型与共享隐结构的基础上,给出一种新的共享动态隐变量模型用于三维人体运动跟踪.该模型针对高维非线性动态系统,可以计算出高维状态向量和高维观测向量的共享动态低维隐变量,同时也能计算出隐变量对高维状态向量、高维观测向量的双向映射、以及隐变量自身的动态关系.使用该模型可以将传统的高维人体运动估计分层为先估计低维隐变量状态,再重建高维人体运动.在实验结果中,用仿真图像序列与真实图像序列证明了方法的有效性.     

Shared latent dynamical structure for three-dimensional human pose estimation

In this paper,a shared latent dynamical model (SLDM) and its application in tracking 3D human motion from monocular videos are proposed by combining the ideas of Gaussian processes dynamical model with shared latent structure.When tracking in high-dimensional space,SLDM can map state space and observation space to a shared latent space of low dimensionality with associated dynamics.During off-line training,three mappings,including dynamical mapping in latent space and mappings from the latent space to both ...     

一种基于局部保持的隐变量模型

隐变量模型是一类有效的降维方法,但是由非线性核映射建立的隐变量模型不能保持数据空间的局部结构。为了克服这个缺点,文中提出一种保持数据局部结构的隐变量模型。该算法充分利用局部保持映射的保局性质,将局部保持映射的目标函数作为低维空间中数据的先验信息,对高斯过程隐变量中的低维数据进行约束,建立局部保持的隐变量。实验结果表明,相比原有的高斯过程隐变量,文中算法较好地保持数据局部结构的效果。     

Generative tracking of 3D human motion in latent space by sequential clonal selection algorithm

High dimensional pose state space is the main challenge in articulated human pose tracking which makes pose analysis computationally expensive or even infeasible. In this paper, we propose a novel generative approach in the framework of evolutionary computation, by which we try to widen the bottleneck with effective search strategy embedded in the extracted state subspace. Firstly, we use ISOMAP to learn the low-dimensional latent space of pose state in the aim of both reducing dimensionality and extracting the prior knowledge of human motion simultaneously. Then, we propose a manifold reconstruction method to establish smooth mappings between the latent space and original space, which enables us to perform pose analysis in the latent space. In the search strategy, we adopt a new evolutionary approach, clonal selection algorithm (CSA), for pose optimization. We design a CSA based method to estimate human pose from static image, which can be used for initialization of motion tracking. In order to make CSA suitable for motion tracking, we propose a sequential CSA (S-CSA) algorithm by incorporating the temporal continuity information into the traditional CSA. Actually, in a Bayesian inference view, the sequential CSA algorithm is in essence a multilayer importance sampling based particle filter. Our methods are demonstrated in different motion types and different image sequences. Experimental results show that our CSA based pose estimation method can achieve viewpoint invariant 3D pose reconstruction and the S-CSA based motion tracking method can achieve accurate and stable tracking of 3D human motion.     

A hierarchical latent variable model for data visualization

Visualization has proven to be a powerful and widely-applicable tool for the analysis and interpretation of multivariate data. Most visualization algorithms aim to find a projection from the data space down to a two-dimensional visualization space. However, for complex data sets living in a high-dimensional space, it is unlikely that a single two-dimensional projection can reveal all of the interesting structure. We therefore introduce a hierarchical visualization algorithm which allows the complete data set to be visualized at the top level, with clusters and subclusters of data points visualized at deeper levels. The algorithm is based on a hierarchical mixture of latent variable models, whose parameters are estimated using the expectation-maximization algorithm. We demonstrate the principle of the approach on a toy data set, and we then apply the algorithm to the visualization of a synthetic data set in 12 dimensions obtained from a simulation of multiphase flows in oil pipelines, and to data in 36 dimensions derived from satellite images     

Gaussian process dynamical models for human motion

We introduce Gaussian process dynamical models (GPDM) for nonlinear time series analysis, with applications to learning models of human pose and motion from high-dimensionalmotion capture data. A GPDM is a latent variable model. It comprises a low-dimensional latent space with associated dynamics, and a map from the latent space to an observation space. We marginalize out the model parameters in closed-form, using Gaussian process priors for both the dynamics and the observation mappings. This results in a non-parametric model for dynamical systems that accounts for uncertainty in the model. We demonstrate the approach, and compare four learning algorithms on human motion capture data in which each pose is 50-dimensional. Despite the use of small data sets, the GPDM learns an effective representation of the nonlinear dynamics in these spaces.     

Transformer-based hierarchical latent space VAE for interpretable remaining useful life prediction

Data-driven prediction of remaining useful life (RUL) has emerged as one of the most sought-after research in prognostics and health management (PHM). Nevertheless, most RUL prediction methods based on deep learning are black-box models that lack a visual interpretation to understand the RUL degradation process. To remedy the deficiency, we propose an intrinsically interpretable RUL prediction method based on three main modules: a temporal fusion separable convolutional network (TF-SCN), a hierarchical latent space variational auto-encoder (HLS-VAE), and a regressor. TF-SCN is used to extract the local feature information of the temporal signal. HLS-VAE is based on a transformer backbone that mines long-term temporal dependencies and compresses features into a hierarchical latent space. To enhance the streaming representation of the latent space, the temporal degradation information, i.e., health indicators (HI), is incorporated into the latent space in the form of inductive bias by using intermediate latent variables. The latent space can be used as a visual representation with self-interpretation to evaluate RUL degradation patterns visually. Experiments based on turbine engines show that the proposed approach achieves the same high-quality RUL prediction as black-box models while providing a latent space in which degradation rate can be captured to provide the interpretable evaluation.     

Shared Kernel Information Embedding for discriminative inference

Latent variable models, such as the GPLVM and related methods, help mitigate overfitting when learning from small or moderately sized training sets. Nevertheless, existing methods suffer from several problems: 1) complexity, 2) the lack of explicit mappings to and from the latent space, 3) an inability to cope with multimodality, and 4) the lack of a well-defined density over the latent space. We propose an LVM called the Kernel Information Embedding (KIE) that defines a coherent joint density over the input and a learned latent space. Learning is quadratic, and it works well on small data sets. We also introduce a generalization, the shared KIE (sKIE), that allows us to model multiple input spaces (e.g., image features and poses) using a single, shared latent representation. KIE and sKIE permit missing data during inference and partially labeled data during learning. We show that with data sets too large to learn a coherent global model, one can use the sKIE to learn local online models. We use sKIE for human pose inference.     

Visual gesture interfaces for virtual environments

Virtual environments provide a whole new way of viewing and manipulating 3D data. Current technology moves the images out of desktop monitors and into the space immediately surrounding the user. Users can literally put their hands on the virtual objects. Unfortunately, techniques for interacting with such environments are yet to mature. Gloves and sensor-based trackers are unwieldy, constraining and uncomfortable to use. A natural, more intuitive method of interaction would be to allow the user to grasp objects with their hands and manipulate them as if they were real objects.We are investigating the use of computer vision in implementing a natural interface based on hand gestures. A framework for a gesture recognition system is introduced along with results of experiments in colour segmentation, feature extraction and template matching for finger and hand tracking, and simple hand pose recognition. Implementation of a gesture interface for navigation and object manipulation in virtual environments is presented.