基于稠密矢量场及自适应补偿的帧率上变换

为了有效解决视频帧率上变换中物体边缘出现的模糊和块效应等问题,结合运动估计,优化了矢量化后处理和运动补偿算法.在基于块的运动估计中,结合运动的时空相关性优化块匹配算法,提高物体边缘运动矢量的准确性;通过基于像素块相关性的插值算法,获取像素矢量场,并保持物体结构完整性;在运动补偿时,结合矢量后处理的分类信息,自适应地进行加权插值计算,提高内插帧的质量.实验结果表明,与传统方法相比,该方法在主观和客观质量上都有较大的提升.     

一种基于场一致性的H.264整帧丢失掩盖方法

针对H.264视频序列在传输中可能发生的整帧丢失现象,提出了一种新型的整帧丢失掩盖算法,该算法可用于P帧和B帧.对于P帧,该算法需要修改解码器的解码流程,使得掩盖丢失帧时能获取后一帧的运动矢量信息.然后,利用运动矢量映射的方法获取重建丢失帧的参考运动矢量列表,采用基于场一致性的判决方式分别选取前一帧中和后一帧中的最佳运动矢量以取得最终重建矢量.对于B帧,利用后P帧的运动矢量信息,采用基于场一致性的判决方式得到最终重建矢量.实验结果说明此算法和传统的整帧掩盖算法相比能获得更高的PSNR,并且有更好的视觉显示效果.所提算法在H.264视频发生整帧丢失时有较好的掩盖效果.     

基于MPEG的全局运动估计技术

提出了一种基于MPEG(moving picture experts group)的全局运动估计技术: 利用四参数模型来描述由摄像机运动带来的全局运动,再利用基于块匹配的算法来求解4个参数,进而作相应的运动补偿.由于MPEG视频流中的运动矢量蕴含了最优块匹配的信息,所以不需要对MPEG视频流完全解码,即可进行参数估计.实验证明该算法速度较快,并能通过运动补偿排除由摄像机运动造成的伪运动区域的干扰.     

基于粒子群聚类的H.264压缩域运动目标分割

提出了一种基于H.264压缩域的运动目标分割方法.首先从压缩流中提取运动矢量场,采用矢量中值滤波方法滤除运动场的噪声和非真实运动块;再运用前向估计方法进行矢量场的累积,解决了帧内预测时无运动矢量的问题;最后运用基于粒子群聚类算法对运动对象进行自适应分割.试验结果表明,该方法能有效分割出H.264压缩域的运动目标.     

基于运动矢量场的运动目标区域检测

基于稀疏运动矢量场,提出一种动态背景下的运动 目标区域检测方法。根据运动矢量场特性分析进行全局运动 参数估计和全局运动补偿,实现动态场景中的背景校正;利用最大树数据结构, 基于运动矢量补偿误差分级表示视频帧中 运动基本一致的连通区域,进行运动区域初始分类;根据运动目标在空间上的连通性和运动 一致性的特点,选择区域相似性 度量准则,进行区域合并和滤波,将具有相似运动的连通区域合并,实现运动目标区域检测 。将检测出的运动目标区域作为 运动矢量外点反过来又应用于全局运动参数估计过程中,全局运动估计和运动目标区域检测 交替地进行,不仅加快了它们的 计算速度,同时也提高了它们计算和检测的准确性。实验结果表明,本文算法能较好地补偿 序列的全局运动,有效地检测出 局部目标运动区域。     

结合光流法与信息熵的人群状态分析

随着近年来危害公共安全的群体性事件频繁发生,对人群场景下的人群状态分析与异常行为检测成为计算机视觉领域研究的热点问题。目前提出的诸如纯光流法,社会力模型,时空运动模型等算法,在检测准确率方面已能满足需求,但是算法复杂度普遍较高,运算量较大,在实际应用中难以保证实时性。鉴于此,首先,引入了人群移动区域面积的概念并定义了人群状态指数,来描述人群状态的变化,通过光流法获得人群运动矢量场,基于人群运动矢量场定义了人群运动强度指数来描述人群运动强度,基于人群运动矢量场与信息熵定义了人群混乱指数来描述人群运动方向的混乱程度。其次,基于降低算法运算量的考虑,根据上面提到的三个描述人群运动状态的特征变量设计了一种分层处理的人群异常行为检测方案,实验结果证明方案具有很好的效果。      

一种针对彩色图像的泛盲掩密分析方案

基于Farid(2002)年泛盲掩密分析算法对彩色图像检测的局限性,提出了新的泛盲掩密分析算法。该算法利用彩色图像的亮度和色度进行分类特征矢量的选择。通过矢量方向滤波技术(VDF)对图像进行方向化处理,并以滤波前后图像之间的矢量方向相关度为色度的分类特征。该方法大大减少了分类特征的数目,而且克服了Farid算法忽略RGB3个颜色通道间相关性的缺陷。分别利用该算法和Farid的算法对JSteg,F5和EZStego软件掩密后的图像进行检测,比较实验结果表明,对于不同长度的嵌入消息,该算法具有更好的检测性能。     

基于MPEG的全局运动估计技术

提出了一种基于MPEG（moving Picture experts group)的全局运动估计技术;利用四参数模型来描述由摄像机运动带来的全局运动,再利用基于块匹配的算法来求解4个参数,进而作相应的运动补偿,由于MPEG视频流中的运动矢量蕴含了最优块匹配的信息,所以不需要对MPEG视频流完全解码,即可进行参数估计,实验证明该算法速度较快,并能通过运动补偿排除由摄像机运动造成的伪运动区域的干扰。     

一种新型基于运动矢量检测的去隔行算法

去隔行是电视图像增强处理和视屏格式转换的一项关键技术.传统的去隔行算法存在着难以克服的缺陷.考虑到由运动估计求得的真实矢量场的特性,提出一种新的基于运动矢量检测的去隔行算法,不需要消耗太多硬件资源.实验证明该算法能够较高质量的恢复图像,客观和主观的测评显示了该算法的健壮性.     

基于运动矢量场和方向自适应的快速运动估计算法

该文提出了一种基于运动矢量场和方向自适应的快速搜索算法。算法针对序列图像的运动矢量场所具有的中心偏置性和时空相关性进行预判,对静止块设定阈值直接中止搜索;根据运动类型自适应选择搜索起始点和搜索策略;采用了两种新的混合搜索方法,对小运动和大运动宏块使用线性-菱形搜索,对中等运动块使用六边形-菱形搜索算法,搜索模板具有强烈的方向特性。实验结果表明,该文算法的搜索速度和搜索精度优于现有的快速运动估计算法,而搜索精度非常接近于全搜索法。     

SVM及其鲁棒性研究

多数人脸识别方法是利用大量正确标记的训练样本来学习精度足够高的识别模型。收集人脸图像并对其进行正确的标记会耗费大量的人力、物力,为了给已有的图像进行标注,研究者进行了大量的工作,但由于多种原因,标记的图像不一定全部正确,称这种标记错误为类别噪声。文中针对含类别噪声的人脸识别问题,指出SVM适用于这类问题,并通过分析位于不同位置的样本对分类的影响从理论上解释了SVM对噪声具有鲁棒性的原因。在SVM基础上,删除一定比例的被判定为噪声的样本后,鲁棒性能有所提高。PubFig数据集上的量实验验证了SVM及改进算法在含类别噪声学习中的有效性。     

Support Vector Machine and Neural Network Classification of Metallic Objects Using Coefficients of the Spheroidal MQS Response Modes

Two different supervised learning algorithms, support vector machine (SVM) and neural networks (NN), are applied in classifying metallic objects according to size using the expansion coefficients of their magneto-quasistatic response in the spheroidal coordinate system. The classified objects include homogeneous spheroids and composite metallic assemblages meant to resemble unexploded ordnance. An analytical model is used to generate the necessary training data for each learning method. SVM and NN are shown to be successful in classifying three different types of objects on the basis of size. They are capable of fast classification, making them suitable for real-time application. Furthermore, both methods are robust and have a good tolerance of 20-dB SNR additive Gaussian noise. SVM shows promise in dealing with noise due to uncertainty in the object's position and orientation.     

基于SVM的信号解调算法

在小样本、低信噪比条件下,同步参数估计会存在较大误差,从而导致信号解调性能的降低。为了解决该问题,将信号解调看成有限长度采样样本的学习问题,并利用支持向量机( SVM)良好的学习性能,在存在同步误差的条件下,通过提高判决端的处理能力来改善系统的接收性能,提出了基于SVM的信号解调算法。在Matlab环境下,对提出算法在精准同步、残存同步误差、高斯白噪声和高斯色噪声等情况进行了计算机仿真,结果表明,相比于匹配滤波器算法,基于SVM的信号解调算法能较好地克服定时误差和相位误差以及色噪声对解调性能的影响。     

Recognition of dynamic video contents with global probabilistic models of visual motion.

The exploitation of video data requires methods able to extract high-level information from the images. Video summarization, video retrieval, or video surveillance are examples of applications. In this paper, we tackle the challenging problem of recognizing dynamic video contents from low-level motion features. We adopt a statistical approach involving modeling, (supervised) learning, and classification issues. Because of the diversity of video content (even for a given class of events), we have to design appropriate models of visual motion and learn them from videos. We have defined original parsimonious global probabilistic motion models, both for the dominant image motion (assumed to be due to the camera motion) and the residual image motion (related to scene motion). Motion measurements include affine motion models to capture the camera motion and low-level local motion features to account for scene motion. Motion learning and recognition are solved using maximum likelihood criteria. To validate the interest of the proposed motion modeling and recognition framework, we report dynamic content recognition results on sports videos.     

基于二层SVM多分类器的桩基缺陷诊断

 为了研究桩基缺陷的快速和准确的分类方法,依据支持向量机(SVM)理论,采用多层分类的方法,改进了一对一SVM多分类器结构,构建了二层一对一SVM多分类器模型,提出了二层多分类计算方法.与BP神经网络相比,二层SVM分类器的学习训练快捷,分类处理的实时性能好,对小样本测试环境的适应能力强,并且具有较好的分类准确率.该方法适合于分析训练样本数量少、分类精度要求高和分类输入输出变量较多的桩基缺陷诊断等多分类问题,对桩基多处缺陷识别的研究也具有重要支持.     

Edge-based 3-D camera motion estimation with application to videocoding

Two classes of algorithms for modeling camera motion in video sequences captured by a camera are proposed. The first class can be applied when there is no camera translation and the motion of the camera can be adequately modeled by zoom, pan, and rotation parameters. The second class is more general in that it can be applied when the camera is undergoing a translation motion, as well as a rotation and zoom and pan. This class uses seven parameters to describe the motion of the camera and requires the depth map to be known at the receiver. The salient feature of both algorithms is that the camera motion is estimated using binary matching of the edges in successive frames. The rate distortion characteristics of the algorithms are compared with that of the block matching algorithm and show that the former provide performance characteristics similar to those of the latter with reduced computational complexity.     

基于多层运动历史图像的飞行时间相机人体运动识别

该文利用飞行时间(Time-Of-Fligh, TOF)相机提供的距离图像,在运动历史图像的基础上提出一种基于多层运动历史图像的人体运动识别方法。计算距离轮廓序列的运动能量图作为整体运动信息,同时根据距离变化量,计算前向、后向的多层运动历史图像作为局部运动信息,共同组成多层运动历史图像。为了解决Hu矩对不连续或具有噪声的形状较为敏感的问题,引入R变换对每层运动历史图像进行特征提取,串联形成特征向量送入SVM进行分类识别。实验结果表明,该识别方法可以有效识别人体运动。     

基于SVM及其改进算法的fMRI图像分类性能研究

为了提出一种更适用于分析fMRI图像特征的机器学习算法,引入机器学习近年提出的、具有较好的泛化能力、并能够保证极值解是全局最优解的新方法支持向量机（SVM）算法,具体选择了PSVM、SSVM、LPSVM、NSVM 4种SVM改进算法以及基本SVM算法应用于fMRI图像的分类问题,在MATLAB平台上进行了算法仿真实现。在对各种算法的分类计算时间、分类精确度两个方面进行比较和讨论后,得到PSVM算法在fMRI图像的分类问题上,有较好的综合性能。     

Light field imaging based on a parallel SVM method for recognizing 2D fake pedestrians

It is novel to apply three-dimensional (3D) light field imaging technology to recognize two-dimensional (2D) fake pedestrians. In this paper, we propose a parallel support vector machine (SVM) method based on 3D light field imaging (light field camera) and machine learning techniques. A light field (LF) camera with robust sensors, which is able to record rich 3D information, is used as hardware equipment. Histogram of oriented gradient (HOG) feature extraction algorithm and SVM classification method are used to recognize the real and 2D fake pedestrians efficiently. Besides, we carry out an experiment on our improved LF pedestrian dataset. The experimental results of parameter optimization study show that in the case of 400 training samples (200 positive samples and 200 negative samples), 120 to 420 testing samples, and an HOG cellsize as 8×8, the best recognition accuracy with polynomial kernel function is improved by more than 2% compared with the previous method. The best accuracy is 99.17%. Otherwise, the recognition accuracy of more than 98.00% will be obtained even under other experimental conditions.     

Fine‐Motion Estimation Using Ego/Exo‐Cameras

Robust motion estimation for human–computer interactions played an important role in a novel method of interaction with electronic devices. Existing pose estimation using a monocular camera employs either ego‐motion or exo‐motion, both of which are not sufficiently accurate for estimating fine motion due to the motion ambiguity of rotation and translation. This paper presents a hybrid vision‐based pose estimation method for fine‐motion estimation that is specifically capable of extracting human body motion accurately. The method uses an ego‐camera attached to a point of interest and exo‐cameras located in the immediate surroundings of the point of interest. The exo‐cameras can easily track the exact position of the point of interest by triangulation. Once the position is given, the ego‐camera can accurately obtain the point of interest's orientation. In this way, any ambiguity between rotation and translation is eliminated and the exact motion of a target point (that is, ego‐camera) can then be obtained. The proposed method is expected to provide a practical solution for robustly estimating fine motion in a non‐contact manner, such as in interactive games that are designed for special purposes (for example, remote rehabilitation care systems).