遮挡情况下刚体位姿估计的自适应无迹卡尔曼分布式融合

针对视觉目标位姿估计系统中常出现的因为特征点遮挡而造成系统估计结果不准确的问题,本文提出了一种利用自适应无迹卡尔曼滤波(AUKF)作为局部滤波器的分布式融合估计方法.通过引入改进的Sage-Husa噪声估计器自适应过程噪声.根据特征点识别量将遮挡情况分为部分遮挡和严重遮挡,对部分遮挡子系统根据先验信息修复缺失观测点后进行局部滤波估计,严重遮挡子系统不参与融合,利用当前时刻整体估计结果对其进行初始化.通过仿真获取了区分遮挡情况的阈值,实验结果表明所提方法能够提升系统在遮挡情况下的估计精度与鲁棒性.     

基于光流的快速人体姿态估计

针对目前深度学习领域人体姿态估计算法计算复杂度高的问题,提出了一种基于光流的快速人体姿态估计算法.在原算法的基础上,首先利用视频帧之间的时间相关性,将原始视频序列分为关键帧和非关键帧分别处理（相邻两关键帧之间的图像和前向关键帧组成一个视频帧组,同一视频帧组内的视频帧相似）,仅在关键帧上运用人体姿态估计算法,并通过轻量级光流场将关键帧识别结果传播到其他非关键帧.其次针对视频中运动场的动态特性,提出一种基于局部光流场的自适应关键帧检测算法,以根据视频的局部时域特性确定视频关键帧的位置.在OutdoorPose和HumanEvaI数据集上的实验结果表明,对于存在背景复杂、部件遮挡等问题的视频序列中,所提算法较原算法检测性能略有提升,检测速度平均可提升89.6%.     

基于轻量级姿态估计的跳绳检测计数算法

针对人体姿态估计算法可实施性低以及基于姿态估计的跳绳计数精度不高的问题, 提出了一种基于轻量级人体姿态估计网络的跳绳计数算法. 该算法首先输入跳绳视频, 接着利用帧间差分法提取关键帧图像并送入人体姿态估计网络进行关节点检测; 同时为了解决轻量级网络检测精度不高的问题, 提出优化的LitePose检测模型, 采用自适应感知解码方法对模型的解码部分进行优化从而减少量化误差; 然后采用卡尔曼滤波对坐标数据进行平滑降噪, 以减小坐标抖动误差; 最终通过关键点坐标变化判断跳绳计数. 实验结果表明, 在相同图像分辨率和环境配置下, 本文提出的算法使用优化的LitePose-S网络模型, 不仅未增加模型参数量和运算复杂度, 同时网络检测精度提高了0.7%, 且优于其他对比网络, 而且本算法在跳绳计数时的平均误差率最低可达1.00%, 可以利用人体姿态估计的结果有效地判断人体起跳和落地情况, 最终得出计数结果.     

Corner validation based on extracted corner properties

We developed a method to validate and filter a large set of previously obtained corner points. We derived the necessary relationships between image derivatives and estimates of corner angle, orientation and contrast. Commonly used cornerness measures of the auto-correlation matrix estimates of image derivatives are expressed in terms of these estimated corner properties. A candidate corner is validated if the cornerness score directly obtained from the image is sufficiently close to the cornerness score for an ideal corner with the estimated orientation, angle and contrast. We tested this algorithm on both real and synthetic images and observed that this procedure significantly improves the corner detection rates based on human evaluations. We tested the accuracy of our corner property estimates under various noise conditions. Extracted corner properties can also be used for tasks like feature point matching, object recognition and pose estimation.     

局部特征表征的6D位姿估计算法

为解决有纹理模型在遮挡条件下6D位姿估计精确度不高的问题,提出了一种局部特征表征的端到端6D位姿估计算法。首先为了得到准确的定位信息,提出了一个空间—坐标注意力机制（spatial and coordinate attention）,通过在YOLOv5网络中加入空间—坐标注意力机制和加权双向特征金字塔网络（bidirectional feature pyramid network）,YOLOv5-CBE算法的精确度（precision）、召回率（recall）、平均精度均值（mAP@0.5）分别提升了3.6%、2.8%、2.5%,局部特征中心点坐标误差最高提升了25%;然后用 YOLOv5-CBE算法检测局部特征关键点,结合3D Harris关键点通过奇异值分解法（singular value decomposition）计算模型的6D位姿,最高遮挡70%的情况下仍然可以保证二维重投影精度（2D reprojection accuracy）和ADD度量精度（ADD accuracy）在95%以上,具有较强的鲁棒性。     

Genetic programming extension to APF-based monocular human body pose estimation

New method of the human body pose estimation based on a single camera 2D observation is presented, aimed at smart surveillance related video analysis and action recognition. It employs 3D model of the human body, and genetic algorithm combined with annealed particle filter for searching the global optimum of model state, best matching the object’s 2D observation. Additionally, new motion cost metric is employed, considering current pose and history of the body movement, favouring the estimates with the lowest changes of motion speed comparing to previous poses. The “genetic memory” concept is introduced for the genetic processing of both current and past states of 3D model. State-of-the-art in the field of human body tracking is presented and discussed. Details of implemented method are described. Results of experimental evaluation of developed algorithm are included and discussed.     

基于可变形卷积的多人人体姿态估计

目前针对人体姿态估计的深度神经网络都是在特征图的固定位置上进行采样,无法对人体姿态的几何变换进行建模,当人体实例在尺寸、姿势、拍摄角度等方面发生变化后,网络泛化能力较差.因此,文中提出基于可变形卷积的多人人体姿态估计方法.利用可变形卷积对目标几何变换建模能力较强的特性,设计特征提取模块,可在人体关键点几何变化的条件下保证检测的准确性.为了进一步提高网络性能,利用预训练残差网络.模型的预测值与二维高斯模型生成的真值用于计算损失,并迭代训练模型,能在拍摄视角、附着物及人物尺度变化等复杂条件下有效检测人体关键点.实验表明,文中模型可有效提升人体关键点检测的准确性.     

基于投票决策的实时遮挡处理技术

针对当前基于深度信息的虚实遮挡处理技术面临的实时性差和精度低的问题,提出一种基于局部区域深度估计和基于patch相似性噪声点投票融合的实时虚实遮挡处理算法.该算法将真实场景视频序列作为输入,首先利用局部区域深度估计算法通过稀疏重建估算出稀疏关键点的深度信息,对稀疏深度施加目标区域的约束限制深度向周围像素的传播,从而快速...     

基于神经网络的双目视觉下头部姿态估计

头部姿态估计是人体姿态检测的关键技术之一。本文基于神经网络设计一种在双目视觉下由人脸中的关键点在空间中的相对位置的变化估计头部姿态,并对头部进行定位的方法。将头部姿态分为6种,空间位置关系分为2种。利用改进SDM算法对双目视觉下的人脸关键点进行标记;标记出人脸关键点后利用的POSIT算法对头部姿态角度估计,计算出头部欧拉角;根据左右图像中对应的头部关键点位置的视差由三角测量原理算出其深度信息。并设定阈值对其进行分类。通过实验,该方法的头部姿态估计准确率高,头部空间定位精度良好。     

色噪声下卡尔曼滤波焊缝跟踪算法与试验研究

精确控制激光束使其始终对中并跟踪焊缝是保证激光焊接质量的前提．以大功率光纤激光焊接Type304不锈钢为试验对象,研究一种有色噪声环境下应用卡尔曼滤波最优状态估计预测激光束与焊缝路径偏差的方法．使用高速红外视觉传感器摄取焊接区红外热像,提取焊缝位置参数并构成状态向量,建立基于焊缝位置参数的系统状态方程和焊缝位置测量方程．针对系统动态噪声为有色噪声,通过扩展状态变量的方法建立有色噪声环境下的卡尔曼滤波算法,对焊缝位置进行最优状态估计并得到最小均方差条件下的焊缝偏差最优预测值,消除系统噪声对焊缝偏差测量的影响．焊接试验结果表明新方法可有效抑制有色噪声干扰并提高焊缝跟踪精度．     

Automatic face detection in video sequences using local normalization and optimal adaptive correlation techniques

Automatic human face detection from video sequences is an important component of intelligent human computer interaction systems for video surveillance, face recognition, emotion recognition and face database management. This paper proposes an automatic and robust method to detect human faces from video sequences that combines feature extraction and face detection based on local normalization, Gabor wavelets transform and Adaboost algorithm. The key step and the main contribution of this work is the incorporation of a normalization technique based on local histograms with optimal adaptive correlation (OAC) technique to alleviate a common problem in conventional face detection methods: inconsistent performance due to sensitivity to variation illuminations such as local shadowing, noise and occlusion. The approach uses a cascade of classifiers to adopt a coarse-to-fine strategy for achieving higher detection rates with lower false positives. The experimental results demonstrate a significant performance improvement gains and achieved by local normalization over methods without normalizations in real video sequences with a wide range of facial variations in color, position, scale, and varying lighting conditions.     

An adaptive unscented Kalman filter approach to secure state estimation for wireless sensor networks

Wireless sensor networks are vulnerable to false data injection attacks, which may mislead the state estimation. To solve this problem, this paper presents a chi-square test-based adaptive secure state estimation (CTASSE) algorithm for state estimation and attack detection. Taking advantage of Kalman filters, attack signal together with process noise or measurement noise are described as total white Gaussian noise with uncertain covariance matrix. The chi-square test method is used in the adaptation of the total noise covariance and attack detection. Then, a standard adaptive unscented Kalman filter (UKF) is used for the state estimation. Finally, simulation results show that the proposed CTASSE algorithm performs better than other UKFs in state estimation and is also effective in real-time attack detection.     

A robust object tracking framework based on a reliable point assignment algorithm

Visual tracking, which has been widely used in many vision fields, has been one of the most active research topics in computer vision in recent years. However, there are still challenges in visual tracking, such as illumination change, object occlusion, and appearance deformation. To overcome these difficulties, a reliable point assignment (RPA) algorithm based on wavelet transform is proposed. The reliable points are obtained by searching the location that holds local maximal wavelet coefficients. Since the local maximal wavelet coefficients indicate high variation in the image, the reliable points are robust against image noise, illumination change, and appearance deformation. Moreover, a Kalman filter is applied to the detection step to speed up the detection processing and reduce false detection. Finally, the proposed RPA is integrated into the tracking-learning-detection (TLD) framework with the Kalman filter, which not only improves the tracking precision, but also reduces the false detections. Experimental results showed that the new framework outperforms TLD and kernelized correlation filters with respect to precision, f-measure, and average overlap in percent.     

基于双分支时空步态特征融合的深度学习步态识别

针对现有步态识别方法易受拍摄视角、着装变化影响的问题,提出一种融合二维无肩姿态拓扑能量图(shoulderless pose topological energy maps, SPTEM)和三维局部骨骼步态特征(local skeleton gait features, LSGF)的深度学习步态识别方法.首先,利用轻量级BlazePose姿态估计算法提取步态视频序列中的人体姿态拓扑图以生成SPTEM,在提高检测速度的同时减弱衣物变化带来的影响;然后,引入LSGF以弥补单一能量图特征在多变视角情况下识别准确率较低的不足;最后,提出结合注意力机制的时空特征提取网络模型,并在全连接层将双流特征进行一致融合.在CASIA-B数据集上对所提出方法进行验证,并与当前主流的步态识别方法进行比较,结果表明,所提出方法在跨视角和穿大衣/棉衣条件下的步态识别率都有明显提升.     

Automatic Feature Point Detection and Tracking of Human Actions in Time-of-flight Videos

Detecting feature points on the human body in video frames is a key step for tracking human movements. There have been methods developed that leverage models of human pose and classification of pixels of the body image. Yet, occlusion and robustness are still open challenges. In this paper, we present an automatic, model-free feature point detection and action tracking method using a time-of-flight camera. Our method automatically detects feature points for movement abstraction. To overcome errors caused by miss-detection and occlusion, a refinement method is devised that uses the trajectory of the feature points to correct the erroneous detections. Experiments were conducted using videos acquired with a Microsoft Kinect camera and a publicly available video set and comparisons were conducted with the state-of-the-art methods. The results demonstrated that our proposed method delivered improved and reliable performance with an average accuracy in the range of 90%. The trajectorybased refinement also demonstrated satisfactory effectiveness that recovers the detection with a success rate of 93.7%. Our method processed a frame in an average time of 71.1 ms.      

基于非局部高分辨率网络的人体姿态估计方法

人体姿态估计是计算机视觉中的基础任务,其可应用于动作识别、游戏、动画制作等。受非局部均值方法的启发,设计了非局部高分辨率网络（non-local high-resolution,NLHR）,在原始图像1/32分辨率的网络阶段融合非局部网络模块的,使网络有了获取全局特征的能力,从而提高人体姿态估计的准确率。NLHR网络在MPII数据集上训练,在MPII验证集上测试,PCKh@0.5评价标准下的平均准确率为90.5%,超过HRNet基线0.2个百分点;在COCO人体关键点检测数据集上训练,在COCO验证集上测试,平均准确率为76.7%,超过HRNet基线2.3个百分点。通过3组消融实验,验证NLHR网络针对人体姿态估计在精度上能够超过现有的人体姿态估计网络。     

A framework for heading-guided recognition of human activity

A combined 2D, 3D approach is presented that allows for robust tracking of moving people and recognition of actions. It is assumed that the system observes multiple moving objects via a single, uncalibrated video camera. Low-level features are often insufficient for detection, segmentation, and tracking of non-rigid moving objects. Therefore, an improved mechanism is proposed that integrates low-level (image processing), mid-level (recursive 3D trajectory estimation), and high-level (action recognition) processes. A novel extended Kalman filter formulation is used in estimating the relative 3D motion trajectories up to a scale factor. The recursive estimation process provides a prediction and error measure that is exploited in higher-level stages of action recognition. Conversely, higher-level mechanisms provide feedback that allows the system to reliably segment and maintain the tracking of moving objects before, during, and after occlusion. Heading-guided recognition (HGR) is proposed as an efficient method for adaptive classification of activity. The HGR approach is demonstrated using “motion history images” that are then recognized via a mixture-of-Gaussians classifier. The system is tested in recognizing various dynamic human outdoor activities: running, walking, roller blading, and cycling. In addition, experiments with real and synthetic data sets are used to evaluate stability of the trajectory estimator with respect to noise.     

基于信息补偿的自适应联邦滤波算法

针对采用标准卡尔曼滤波器必须知道系统噪声统计特性的局限性,研究了一类系统噪声未知情况下的自适应联邦滤波方法,指出了自适应滤波方法应用于联邦结构时应当注意的问题,提出了一种基于信息补偿的自适应联邦滤波算法。SINS／BDS／GPS组合导航系统的仿真结果表明,该方法可以有效抑制系统噪声未知情况下的滤波发散现象,提高了滤波的稳定性和估计性能。     

On the choice of importance distributions for unconstrained and constrained state estimation using particle filter

Recursive state estimation of constrained nonlinear dynamical system has attracted the attention of many researchers in recent years. For nonlinear/non-Gaussian state estimation problems, particle filters have been widely used (Arulampalam et al. [1]). As pointed out by Daum [2], particle filters require a proposal distribution and the choice of proposal distribution is the key design issue. In this paper, a novel approach for generating the proposal distribution based on a constrained Extended Kalman filter (C-EKF), Constrained Unscented Kalman filter (C-UKF) and constrained Ensemble Kalman filter (C-EnkF) has been proposed. The efficacy of the proposed state estimation algorithms using a particle filter is illustrated via a successful implementation on a simulated gas-phase reactor, involving constraints on estimated state variables and another example problem, which involves constraints on the process noise (Rao et al. [10]). We also propose a state estimation scheme for estimating state variables in an autonomous hybrid system using particle filter with Unscented Kalman filter as a proposal and unconstrained Ensemble Kalman filter (EnKF) as a proposal. The efficacy of the proposed state estimation scheme for an autonomous hybrid system is demonstrated by conducting simulation studies on a three-tank hybrid system. The simulation studies underline the crucial role played by the choice of proposal distribution in formulation of particle filters.     

Least squares estimation of nonstationary covariance parameters in linear systems

Least squares estimation techniques are employed to overcome previous difficulties encountered in accurately estimating the state and measurement noise covariance parameters in linear stochastic systems. In the past accurate and rapidly converging covariance parameter estimates have been achieved with complex estimation algorithms only after specifying the statistical nature of the noise in the system and constraining the time variation of the covariance parameters. Weighted least squares estimation allows these restrictions to be removed while achieving near optimal accuracy using a filter on the same order of complexity as a Kalman filter. Allowing the covariance parameters to vary in as general a manner in time as the state in a linear discrete time stochastic system, and assuming that a Kalman filter is applied to this system using incorrect knowledge of the a priori statistics, it is shown how a covariance system is developed similar to the original system. Unbiased least squares estimates of the covariance parameters and of the original state are obtained without the necessity of specifying the distribution on the noise in either system. The accuracy of these estimates approaches optimal accuracy with increasing measurements when adaptive Kalman filters are applied to each system.