Propagating adaptive-weighted vector median filter for motion-field smoothing

In the field of predictive video coding and format conversion, there is an increasing attention towards estimation of the true inter-frame motion. The restoration of motion vector field computed by 3-D RS is addressed and a propagating adaptive-weighted vector median (PAWVM) post-filter is proposed. This approach decomposes blocks to make a better estimation on object borders and propagates good vectors in the scanning direction. Furthermore, a hard-thresholding method is introduced into calculating vector weights to improve the propagating. By exploiting both the spatial correlation of the vector field and the matching error of candidate vectors, PAWVM makes a good balance between the smoothness of vector field and the prediction error, and the output vector field is more valid to reflect the true motion.     

Object detection based on combination of local and spatial information

A method of object detection based on combination of local and spatial information is proposed.Firstly,the categorygiven representative images are chosen through clustering to be templates,and the local and spatial information of template are extracted and generalized as the template feature.At the same time,the codebook dictionary of local contour is also built up.Secondly,based on the codebook dictionary,sliding-window mechanism and the vote algorithm are used to select initial candidate object windows.Lastly,the final object windows are got from initial candidate windows based on local and spatial structure feature matching.Experimental results demonstrate that the proposed approach is able to consistently identify and accurately detect the objects with better performance than the existing methods.     

Fault detection and diagnosis for data incomplete industrial systems with new Bayesian network approach

For the fault detection and diagnosis problem in largescale industrial systems,there are two important issues: the missing data samples and the non-Gaussian property of the data.However,most of the existing data-driven methods cannot be able to handle both of them.Thus,a new Bayesian network classifier based fault detection and diagnosis method is proposed.At first,a non-imputation method is presented to handle the data incomplete samples,with the property of the proposed Bayesian network classifier,and the missing values can be marginalized in an elegant manner.Furthermore,the Gaussian mixture model is used to approximate the non-Gaussian data with a linear combination of finite Gaussian mixtures,so that the Bayesian network can process the non-Gaussian data in an effective way.Therefore,the entire fault detection and diagnosis method can deal with the high-dimensional incomplete process samples in an efficient and robust way.The diagnosis results are expressed in the manner of probability with the reliability scores.The proposed approach is evaluated with a benchmark problem called the Tennessee Eastman process.The simulation results show the effectiveness and robustness of the proposed method in fault detection and diagnosis for large-scale systems with missing measurements.     

Elevation estimation for low-angle target based on reflection paths suppression

In the signal processing for metrewave radar, the reflection paths of target echoes can cause severe error in the elevation estimation for the low-angle target tracking. The exact angles of the reflection paths are unknown beforehand, and therefore, the reflection paths can not be suppressed easily. Therefore, in this article, an improved reflection paths suppression approach is presented. A block matrix aggregate is constructed based on the possible angles of the reflection paths. Combined with the beamforming-like processing, a generalized maximum likelihood estimation is derived to optimize the estimation. Moreover, the noise reduction method based on the Toeplitz covariance matrix is used for better performance. This approach is applied to the real data collected by the low-angle tracking radar with 8-channel vertical array. The experiment results show that the reflection effects are reduced and the accuracy of the elevation estimate is improved.     

New approach to eliminate structural redundancy in case resource pools using α mutual information

Structural redundancy elimination in case resource pools (CRP) is critical for avoiding performance bottlenecks and maintaining robust decision capabilities in cloud computing services. For these purposes, this paper proposes a novel approach to ensure redundancy elimination of a reasoning system in CRP. By using α entropy and mutual information, functional measures to eliminate redundancy of a system are developed with respect to a set of outputs. These measures help to distinguish both the optimal feature and the relations among the nodes in reasoning networks from the redundant ones with the elimination criterion. Based on the optimal feature and its harmonic weight, a model for knowledge reasoning in CRP (CRPKR) is built to complete the task of query matching, and the missing values are estimated with Bayesian networks. Moreover, the robustness of decisions is verified through parameter analyses. This approach is validated by the simulation with benchmark data sets using cloud SQL. Compared with several state-of-the-art techniques, the results show that the proposed approach has a good performance and boosts the robustness of decisions.     

Human activity recognition based on HMM by improved PSO and event probability sequence

This paper proposes a hybrid approach for recognizing human activities from trajectories.First,an improved hidden Markov model(HMM) parameter learning algorithm,HMM-PSO,is proposed,which achieves a better balance between the global and local exploitation by the nonlinear update strategy and repulsion operation.Then,the event probability sequence(EPS) which consists of a series of events is computed to describe the unique characteristic of human activities.The analysis on EPS indicates that it is robust to the changes in viewing direction and contributes to improving the recognition rate.Finally,the effectiveness of the proposed approach is evaluated by data experiments on current popular datasets.     

Novel approach to GPS/SINS integration for IMU alignment

Strapdown inertial navigation system(SINS) requires an initialization process that establishes the relationship between the body frame and the local geographic reference.This process,called alignment,is generally used to estimate the initial attitude angles.This is possible because an accurate determination of the inertial measurement unit(IMU) motion is available based on the measurement obtained from global position system(GPS).But the update frequency of GPS is much lower than SINS.Due to the non-synchronous data streams from GPS and SINS,the initial attitude angles may not be computed accurately enough.In addition,the estimated initial attitude angles may have relatively large uncertainties that can affect the accuracy of other navigation parameters.This paper presents an effective approach of matching the velocities which are provided by GPS and SINS.In this approach,a digital high-pass filter,which implements a pre-filtering scheme of the measured signal,is used to filter the Schuler cycle of discrete velocity difference between the SINS and GPS.Simulation results show that this approach improves the accuracy greatly and makes the convergence time satisfy the required accuracy.     

STAP method based on atomic norm minimization with array amplitude-phase error calibration

In this paper,a space-time adaptive processing(STAP)method is proposed for the airborne radar with the array amplitude-phase error considered,which is based on atomic norm minimization(ANM).In the conventional ANM-based STAP method,the influence of the array amplitude-phase error is not considered and restrained,which inevitably causes performance deterioration.To solve this problem,the array amplitude-phase error is firstly estimated.Then,by pre-estimating the array amplitude-phase error information,a modified ANM model is built,in which the array amplitude-phase error factor is separated from the clutter response and the clutter covariance matrix(CCM)to improve the estimation accuracy of the CCM.To prove that the atomic norm theory is applicable in the presence of the array amplitude-phase error,the clutter sparsity is analyzed in this paper.Meanwhile,simulation results demonstrate that the proposed method is superior to the state-of-the-art STAP method.Moreover,the measured data is used to verify the effectiveness of the proposed method.     

Improved Integrated Deep Model for Pap-Smear Cell Analysis

Cervical cancer is the fourth most common malignancy to strike a woman globally. If discovered early enough, it can be effectively treated. Although there is a chance of error owing to human error, the Pap smear is a good tool for first screening for cervical cancer. It also takes a lot of time and effort to complete. The aim of this study was to reduce the possibility of error by automating the process of classifying cervical cancer using Pap smear images. For the purpose of this study,dual con...     

Fully Distributed Cooperative Motion of Group Robots

This paper is focused on the fully distributed cooperative motion of group robots and proposes a new approach. Each robot has a local sensing ability and a simple action selection strategy. Computational complexity is decreased by the fully distributed architecture and the information insufficiency is solved by the interaction between the robots and the environment. Variable loop and random method are used to deal with the fluctuation and equity selection problem and the rapidity and reasonabiliiy are guaranteed. Some simulations have proved the effectiveness of the proposed approach.     

光学运动捕捉散乱数据处理的一种方法

阐述了一种适用于被动式光学人体运动捕捉散乱数据处理的方法,该方法基于模版匹配的原理,根据人体标示的特征点的局部相对刚性结构以及整体拓扑结构关系,利用刚性结构匹配、运动轨迹追踪、拓扑结构校验实现散乱数据的处理,从而得到人体运动参数及骨架拓扑结构。该方法解决了追踪算法在特征点重合时的追踪出错问题,处理方法经优化后数据处理过程能够满足运动捕捉的实时性要求,具备错误自动检测和纠正能力,处理过程无须人工干涉。     

稀疏场景下SAR方位向随机丢失数据的迭代成像算法

针对合成孔径雷达(synthetic aperture radar, SAR)方位向随机丢失部分数据导致目标模糊和能量分散的问题, 提出基于稀疏优化理论的重建成像方法。该方法主要针对稀疏观测场景的SAR方位向随机缺失数据的回波信号进行成像处理, 利用SAR回波模拟算子, 避免了二维回波信号矩阵变成向量的操作, 从而减小了内存占用和计算量。所提出的基于SAR近似观测模型的迭代优化重建算法能够实现对观测目标幅度的高精度重建。和传统基于匹配滤波的SAR成像算法相比, 提出的算法能够明显地消除SAR方位向随机丢失部分数据引起的目标模糊和目标能量分散。和迭代软阈值收缩算法相比, 提出的算法重建的目标幅度误差更小。理想的点目标回波数据和真实的星载SAR稀疏观测场景的回波数据处理表明了所提算法在减少重建目标误差、提高观测目标区域的目标背景比等方面有较大的优势。     

一种视频驱动的三维人体动画合成方法

三维人体动画的生成是计算机动画领域的热点问题。已有的运动捕获方法虽然可以获得逼真的运动数据，但使用不便、代价昂贵的缺点限制了其应用。利用已有的运动数据库，提出了一种视频驱动的三维人体动画合成方法。该方法通过建立运动库概率转移和统计模型，挖掘运动数据的时空结构，能够根据视频内容合成出新的动画数据，从而提高已有运动数据的可重用性，大大降低动画制作成本。采用广播体操运动视频验证了方法的有效性。     

基于运动结构图的无人机序列影像三维重建

无人机序列影像具有重叠度高、拓扑联系强的特点。针对现有增量式三维重建算法在处理大规模无人机影像时效率低、输出不稳定的问题,提出基于运动结构图的无人机序列影像三维重建方法。顾及地理位置信息与IMU(Inertial measurement unit) 数据建立图像索引集,提高图像匹配效率,并使用随机抽样一致性算法计算相对运动;使用运动结构图表示影像之间的“ 关联”,采用闭环检测剔除错误边;借助李代数与李群的映射关系, 融合1 L 范式与M 估计一次性求解影像全局运动,期间只进行一次光束法平差,降低了计算复杂度。实验验证本文方法在效率和效果上得到了提高。     

被动式光学运动捕捉系统丢点检测与补偿

针对虚拟人技术中被动式光学捕捉系统丢点或误匹配问题,利用标志点运动的连续性和局部相关性提出了一种检测和补偿的方法。注意到标志点运动的连续性,利用轨迹的二阶导数检测标志点畸变的时刻。由于标志点在人体分布的局部相关性,从刚性模型和线性模型两个角度,利用与畸变点相邻的标志点修正和补偿畸变点的真正位置。实验结果验证了所提方法的有效性。     

基于多参量模型的全局运动估计算法

提出了一种基于特征点匹配的视频序列多参量运动估计算法,即将初始特征点域,进行梯度方向上的极坐标变换,并依据正交特征,作方向轴上的一维投影,构建出新的特征曲线,从而将孤立的特征点匹配转化成特征曲线相关.依据曲线的相关性,完成特征点对的匹配,最后使用最小二乘解算平移、旋转、缩放等变换参量.实验结果表明:算法的特征点误匹配率<5%,缩放误差<0.1%,平移误差<0.1像素;各参量估计范围:缩放因子s:0.7130°,平移参量|d|>35.算法在保证了高精度的同时,具有更为宽松的使用条件及适用性.     

基于中心距离特征的人体运动序列关键帧提取

运动捕获数据的关键帧是原始运动序列的简洁表示,对于运动压缩、运动检索和运动分割起着重要的作用。提出了一种基于中心距离特征的人体运动捕获数据关键帧提取方法,通过提取四肢到中心点ROOT的距离,得到一组中心距离特征,将特征分为上肢和下肢来分别表示,并提取上下肢的距离模,得到二维的特征向量模;然后采用主成分分析得到一维特征,并提取其局部极值点作为初始关键帧;最后通过对初始关键帧的重新筛选与插入得到最终关键帧序列。实验结果表明,该方法提取的关键帧序列在视觉上能够很好的概括原始运动序列的内容,且具有高压缩率。     

一种基于变形模板的椭圆跟踪算法

给出一种新的基于活动轮廓与光流约束的椭圆跟踪算法。椭圆变形模板在引入对目标模型严格的全局约束的同时,降低了计算复杂度;基于模型的光流技术被用于提取目标的运动信息并引导轮廓的变形;借助扩展Kalman(EKF)滤波器有机组合目标的形状与运动信息;最后利用光流测量方程给出的误差测度及EKF给出的估计方差对虚假样本点作出判断与舍弃,从而保证对图像噪声、遮挡及伪边缘具有较强的克服能力。算法除可用于跟踪刚体运动之外,也可用于非刚体(例如人头)运动的跟踪。计算实例表明了算法的有效性。     

强化结构的数字壁画病害修复算法研究

针对北京地区法海寺壁画块状缺失,且缺失区域结构信息丰富的特点,提出了一种强化结构的数字图像修复算法,解决了Criminisi算法修复时对图像结构信息考虑不足的问题。首先在计算填充块的优先权函数时,将线性卷积的曲率计算融入数据项中,同时增加结构信息的权重,实现了优先修复结构信息丰富区域的目的;其次在搜索匹配块的相似度计算中引入区域协方差方法,确保修复后图像结构的一致性,减少了匹配错误率。实验结果表明：该算法较好地解决了结构性较强的壁画块状信息缺失填充错位问题。     

融合视惯传感数据的非接触式物体尺寸获取方法

从视觉数据和惯性测量单元(Inertial Measurement Unit, IMU)数据各自的特点出发,利用计算机视觉三维重建技术从视觉数据中恢复视觉传感器的六维位姿信息,分析计算惯性测量数据和视觉位姿数据的时空关系,实现融合视惯传感数据的场景重建及尺度估计,进而获取待测对象的绝对物理尺寸数据。上述方法在实验数据上取得了相对误差约为3%的非接触式物体尺寸估计结果,由于该方法无需在待测场景中嵌入额外标定物或辅助定位装置,具有较好的灵活性和适用范围。