基于粒子滤波后验概率分布的多特征融合跟踪

在光照和目标形变等外部条件变化的情况下,仅利用目标的单一特征难以鲁棒的跟踪目标。提出了一种基于粒子滤波后验概率分布的多特征融合跟踪算法,在粒子滤波跟踪框架下,用直方图模型表征目标的颜色和边缘特征,通过两种特征后验概率之间的"协作"与"学习"实现特征融合,各种场景的试验结果比较表明,新的融合跟踪算法比仅用单一特征跟踪、现有的多特征融合算法具有更好的稳定性和鲁棒性,特别是针对环境光照和目标背景变化较大的情况更具有优势。     

结合颜色属性的分层结构直方图

目的 为克服单一颜色特征易受光照变化影响,以及图像的空间结构特征对目标形变较为敏感等问题,提出一种结合颜色属性的分层结构直方图。方法 首先,鉴于使用像素灰度值对图像进行分层易受光照变化影响,本文基于颜色属性对图像进行分层,即将输入的彩色图像从RGB空间映射到颜色属性空间,得到11种概率分层图;之后,将图像中的每一个像素仅投影到其概率值最大的分层中,使得各分层之间像素的交集为空,并集为整幅图像;对处理后的每一个分层,通过定义的结构图元来统计像素分布情况,得到每一分层的空间分布信息;最后,将每一分层的像素空间分布信息串联作为输入图像的分层结构直方图,以此来表征图像。结果 为证明本文特征的有效性,将该特征用于图像匹配和视觉跟踪,与参考特征相比,利用本文特征进行图像匹配时,峰值旁瓣比均值提升1.347 9;将本文特征用于视觉跟踪时,采用粒子滤波作为跟踪框架,成功率相对上升4%,精度相对上升4.6%。结论 该特征将图像的颜色特征与空间结构信息相结合,有效解决了单一特征分辨性较差的问题,与参考特征相比,该特征具有更强的分辨性和鲁棒性,因此本文特征可以更好地应用于图像处理应用中。     

Combining object and feature dynamics in probabilistic tracking

Objects can exhibit different dynamics at different spatio-temporal scales, a property that is often exploited by visual tracking algorithms. A local dynamic model is typically used to extract image features that are then used as inputs to a system for tracking the object using a global dynamic model. Approximate local dynamics may be brittle—point trackers drift due to image noise and adaptive background models adapt to foreground objects that become stationary—and constraints from the global model can make them more robust. We propose a probabilistic framework for incorporating knowledge about global dynamics into the local feature extraction processes. A global tracking algorithm can be formulated as a generative model and used to predict feature values thereby influencing the observation process of the feature extractor, which in turn produces feature values that are used in high-level inference. We combine such models utilizing a multichain graphical model framework. We show the utility of our framework for improving feature tracking as well as shape and motion estimates in a batch factorization algorithm. We also propose an approximate filtering algorithm appropriate for online applications and demonstrate its application to tasks in background subtraction, structure from motion and articulated body tracking.     

基于粗糙集理论的图像分割智能决策方法

尽管如今已有多种图像分割算法,但是没有任何一种分割方法能够适用于所有的图像.为了使图像跟踪系统能根据图像特征自适应选取分割算法,给出了一种基于粗糙集理论的图像分割智能决策方法.该方法首先选取若干具代表性的分割算法构成算法库,并用它们对各种样本图像进行分割;然后利用从样本图像中提取出来的各种数值特征,并根据图像分割质量评价标准评判出各样本图像的最优分割算法,用其构成决策信息表;最后应用粗糙集理论来对决策信息表进行离散化处理和属性约简,以生成图像分割算法选取的决策规则.该决策方法解决了图像跟踪系统中分割算法选取的一系列难题.实验证明,该决策方法能比较有效地根据系统所处理图像的特征选取出算法库中最优的分割算法,并可满足车载图像跟踪系统的实时性要求.     

基于声呐图像的水下目标检测、识别与跟踪研究综述

水下目标检测、识别和跟踪是具有重要意义的热点研究问题,在军事和民用领域都有重要的应用.鉴于此,对基于声呐图像的水下目标检测、识别和跟踪原理、方法以及典型算法的研究进展进行全面阐述.首先论述基于声呐图像的水下目标检测、图像去噪、图像分割等方面的主要进展以及典型算法和算法扩展;然后对水下目标声呐图像识别中的特征提取、特征分类方法和主要技术难点进行讨论;最后阐述基于水声信号处理和声呐图像信息的水下目标跟踪方法和算法.通过对水下目标处理过程各个过程的深入讨论和对比分析,指出基于声呐图像的水下目标检测、识别和跟踪中急需解决的关键科学问题及可能的解决思路,并对该领域的未来发展方向做进一步的展望.     

Robust segment-based object tracking using generalized hyperplane approximation

Tracking based on gradient descent algorithm using image gradient is one of the popular object tracking method. However, it easily fails to track when illumination changes. Although several illumination invariant features have been proposed, applying the invariant feature to the gradient descent method is not easy because the invariant feature is represented as a non-linear function of image pixel values and its Jacobian cannot be calculated in a closed-form. To make it possible, we introduce the generalized hyperplane approximation technique and apply it to histogram of oriented gradient (HOG) feature, one of the well-known illumination invariant feature. In addition, we achieve partial occlusion invariance using image segments. The hyperplanes are calculated from training segment images obtained by perturbing the motion parameter around the target region. Then, it is used to map the difference in non-linear feature of image onto the increment of alignment parameters. This process is mathematically same to the gradient descent method. The information from each segment is integrated by a simple weighted linear combination with confidence weights of segments. Compared to the previous tracking algorithms, our method shows very fast and stable tracking results in experiments on several practical image sequences.     

Real-time model-based tracking combining spatial and temporal features

This paper describes a method for tracking moving image features by combining spatial and temporal edge information with model-based feature information. The algorithm updates the two-dimensional position of object features by correlating predicted model features with current image data. The results of the correlation process are used to compute an updated model. The algorithm makes use of a high temporal sampling rate with respect to spatial changes of the image features and operates in a real-time multi-processing environment. Preliminary results demonstrate successful tracking for image feature velocities between 1.1 and 4.5 pixels every image frame.     

A real-time visual object tracking system based on Kalman filter and MB-LBP feature matching

Visual tracking has very important applications in practice. Many proposed visual trackers are not suitable for real-time applications because of their huge computational loads or sensitivities against changing environments such as illumination variation. In this paper, we propose a new tracker which uses modified Multi-scale Block Local Binary Patterns (MB-LBP) like feature to characterize the tracked object. Such feature has low computational load and robustness against illumination variation. An updated appearance model is build based on the modified MB-LBP feature. The model is updated in every frame by replacing the appearance model with the features extracted from the most current detected image patch of target. Moreover, we use the predicted information about the target to constructed a smaller searching area for target in new frame. It greatly reduces computational load for target searching. Numerical experiments show that the drift effect of tracker is greatly avoided and the tracker has very effective and robust performance on various test videos.     

基于视觉传达的多帧图像特征目标跟踪仿真

针对当前方法提取的多帧图像目标特征精度较差,导致多帧图像特征目标跟踪准确率较低、跟踪时间较长的问题,提出了基于视觉传达的多帧图像特征目标跟踪方法。采用稀疏表示方法采集多帧图像目标特征,利用高斯分布构建图像运动模型,小波分析多帧图像灰度及细节特征,根据灰度投影法提取多帧图像目标特征,并匹配多帧图像特征点,获取多帧图像轮廓轨迹追踪目标,运用图像运动目标状态模型,求解前景轮廓的目标函数权重,实现多帧图像特征目标跟踪。实验结果表明,所提方法提取的多帧图像目标特征精度较好,能够有效降低多帧图像特征目标跟踪时间,提高多帧图像特征目标跟踪准确率。     

在线矩形特征选择的压缩跟踪算法

针对压缩跟踪算法无法选择合适的矩形特征,易出现目标漂移、丢失现象,提出了一种基于在线矩形特征选择的压缩跟踪算法。首先,在初始化阶段生成投影矩阵,利用该投影矩阵提取特征来构造候选特征池,在特征池中使用矩形特征来表示目标特性,并去除与目标差异较大的矩形特征,最后计算分类分数最大的窗口,并将其作为目标窗口,从而实现跟踪。实验结果表明,该算法特征总数量比压缩跟踪算法特征总数量减少了13%,且跟踪精度和鲁棒性方面得到了改善,对于320pixel×240pixel大小的视频平均处理帧速为20frame/s,满足实时性要求。     

背景对齐差分的机场跑道异物分块检测与跟踪算法

针对机场跑道异物的无人自主检测与识别过程中存在的检测结果受环境影响大、小目标检测困难以及漏检率高等问题,提出基于背景差分的机场跑道异物分块检测与跟踪算法.首先利用速度辅助的位置信息线性插值,获取背景模板图像库与待检测图像序列最相关帧;然后将图像分为不同子块,利用各子块内图像纹理复杂度设置自适应权值,结合ORB算法进行特征点提取,将各子块特征点归一化至原始图像,并与背景模板库最相关帧对齐作差,获取异物检测结果;最后引入核相关滤波器对检测结果进行多目标跟踪,采用各子块局部跟踪算法降低运算时间,并对跟踪结果进行可靠性检验.在3种实验场景下,与5种主流检测算法的对比实验结果表明,与目前已有的基于图像的机场异物检测算法相比,在保证算法处理速度的基础上,该算法将异物整体错检率降低了70%以上,并在异物尺寸大于1cm×1cm的情况下,将整体漏检率降低至0,获得了较好的效果.     

基于卷积受限玻尔兹曼机的医学图像分类新方法

利用数据挖掘方法对医学图像做分析是目前研究的热点之一,常用的挖掘方法首先需要从医学图像中提取特征,然后进行分类分析。目前,应用最多的是提取图像的统计特征,这种方法对所提取的特征有很强的依赖性。采用一种深度学习的新方法——卷积受限玻尔兹曼机模型,并且采用改进的快速持续对比散度算法对模型进行训练。该方法直接从乳腺X光图像中自主学习特征并利用学习到的特征对图像进行分类。实验结果显示,新方法对医学图像的分类精度相对于已有方法有明显的提升。     

自适应卷积特征选择的实时跟踪算法

目的 针对深度卷积特征相关滤波跟踪算法因特征维度多造成的跟踪速度慢及其在目标发生形变、遮挡等情况时存在跟踪失败的问题,提出了一种自适应卷积特征选择的实时跟踪算法。方法 该算法先分析结合深度卷积特征的相关滤波跟踪算法定位目标的特性,然后提出使用目标区域和搜索区域的特征均值比来评估卷积操作,选取满足均值比大于阈值的特征通道数最多的卷积层,减少卷积特征的层数及维度,并提取该卷积层的有效卷积特征来训练相关滤波分类器,最后采用稀疏的模型更新策略提高跟踪速度。结果 在OTB-100标准数据集上进行算法测试,本文算法的平均距离精度值达86.4%,平均跟踪速度达29.9帧/s,比分层卷积相关滤波跟踪算法平均距离精度值提高了2.7个百分点,速度快将近3倍。实验结果表明,本文自适应特征选择的方式在保证跟踪精度的同时有效地提升了跟踪的速度,且优于当前使用主成分分析降维的方式;与现有前沿跟踪算法对比,本文算法的整体性能优于实验中对比的9种算法。结论 该算法采用自适应卷积通道和卷积层选择的方式有效地减少了卷积层数和特征维度,降低了模型的复杂度,提升了跟踪速度,利用稀疏模型更新策略进一步提升了跟踪的速度,减少了模型漂移现象,当目标发生快速运动、遇到遮挡、光照变化等复杂场景时,仍可实时跟踪到目标,具有较强的鲁棒性和适应性。     

Local Kernel Feature Analysis (LKFA) for object recognition

This paper proposes a new Local Kernel Feature Analysis (LKFA) method for object recognition. LKFA captures the nonlinear local relationship in an image via kernel functions. Different from traditional kernel methods for object recognition, the proposed method does not need to reserve the training samples. LKFA is designed to extract the eigenvalue features from the Hermite matrix of a local feature representation, which we have theoretically proven its robustness to noise and perturbations. Experiment results on palmprint and face recognitions demonstrated the effectiveness of the proposed LKFA that significantly improved the performance of the local feature based object recognition method.     

Kernelized correlation tracker on smartphones

This paper shows the implementation of a KC tracker (high-speed kernelized correlation tracker) on an Android smartphone. The image processing part is implemented with the Android-NDK in C/C++. Some parts of the tracking algorithm, which can be parallelized very well, are partitioned and calculated on the GPU with OpenGL ES and OpenCL. Other parts, such as the Discrete Fourier Transform (DFT), are calculated on the CPU (partly with the ARM-NEON features). With these hardware acceleration steps we could reach real-time performance (at least 20–30 FPS) on up-to-date smartphones, such as Samsung Galaxy S4, S5 or Google Nexus 5.Beyond that, we present some new color features and compare their tracking quality to the HOG features using the KC tracker and show that their tracking quality is mostly superior compared to the HOG features.If an object gets lost by the tracker which is the case e.g. if the object is totally hidden or outside the viewing range, there should be a possibility to perform a re-detection. In this paper, we show a basic approach to determine the tracking quality and search for the tracking object in the entire images of the subsequent video-frames.     

一种多特征融合的粒子滤波跟踪新算法

仅利用单一的目标特征进行跟踪是大多数跟踪算法鲁棒性不高的重要原因。提出了一种有效的多特征融合跟踪方法,该方法同时结合了颜色和运动边缘特征,并通过粒子滤波方法合理地进行概率融合。实验结果表明,算法能够在一种特征受到背景干扰导致目标鉴别能力丧失时,其它特征仍能稳定可靠地跟踪目标,算法简单,鲁棒性高,能够有效适用于复杂背景下的目标跟踪。     

Unusual human behavior recognition using evolutionary technique

The vision for surveillance is an important task in many computer vision applications. The monitoring system concerns the tracking and recognition of people, and more generally, the understanding of human behaviors, from image sequences involving humans. Several methods for human tracking and human behavior recognition have been proposed by various researchers. But most of those do not have versatility and flexibility. In this paper, we propose an efficient and robust object tracking algorithm which use the color features, the distance features and count feature based on an evolutionary techniques to measure the observation similarity. And then we will track each person and classify their behavior properties by analyzing their trajectory pattern. We propose multi-layer perceptron based on hybrid genetic algorithm using Gaussian synapse make the recognition algorithm very efficient and robust for classify human behavior by trajectory pattern.     

带时延补偿的视觉跟踪控制

在实际的视觉伺服系统中, 由于摄像机到图像处理设备的传输延迟和图像处理本身占用的时间, 视觉信息的获取会产生时延. 对此, 给出了一个带有时延补偿的视觉跟踪控制方法. 通过实时拟合图像雅可比矩阵, 实现了对机械手末端执行器图像特征信息的实时预测, 从而减小了估计误差. 在此基础上, 设计了一个带有时延补偿的控制方案. 通过对运动目标进行跟踪的仿真实验, 验证了本文时延补偿方法的有效性.     

级联特征融合孪生网络目标跟踪算法研究

在光照变化、遮挡、背景相似、变形等复杂情况下,目标跟踪过程中难以精确地提取丰富的特征信息,容易导致目标跟踪出现漂移或者跟踪丢失.由于多层神经网络的浅层特征具有高分辨率,适合于目标定位;深层特征具有丰富的语义信息,适合于目标分类.充分利用这一优势,提出了一种级联特征融合的孪生网络目标跟踪算法.对ResNet-50网络进行...     

融合特征的卫星视频车辆单目标跟踪

目的 卫星视频作为新兴遥感数据,可以提供观测区域高分辨率的空间细节信息与丰富的时序变化信息,为交通监测与特定车辆目标跟踪等应用提供了不同于传统视频视角的信息。相较于传统视频数据,卫星视频中的车辆目标分辨率低、尺度小、包含的信息有限。因此,当目标边界不明、存在部分遮挡或者周边环境表观模糊时,现有的目标跟踪器往往存在严重的目标丢失问题。对此,本文提出一种基于特征融合的卫星视频车辆核相关跟踪方法。方法 对车辆目标使用原始像素和方向梯度直方图（histogram of oriented gradient,HOG）方法提取包含互补判别能力的特征,利用核相关目标跟踪器分别得到具备不变性和判别性的响应图;通过响应图融合的方式结合两种特征的互补信息,得到目标位置;使用响应分布指标（response distribution criterion,RDC）判断当前目标特征的稳定性,决定是否更新跟踪器的表征模型。本文使用的相关滤波方法具有计算量小且运算速度快的特点,具备跟踪多个车辆目标的拓展能力。结果 在8个卫星视频序列上与主流的6种相关滤波跟踪器进行比较,实验数据涵盖光照变化、快速转弯、部分遮挡、阴影干扰、道路颜色变化和相似目标临近等情况,使用准确率曲线和成功率曲线的曲线下面积（area under curve,AUC）对车辆跟踪的精度进行评价。结果表明,本文方法较好地均衡了使用不同特征的基础跟踪器（性能排名第2）的判别能力,准确率曲线AUC提高了2.9%,成功率曲线AUC下降了4.1%,成功跟踪车辆目标,不发生丢失,证明了本文方法的先进性和有效性。结论 本文提出的特征融合的卫星视频车辆核相关跟踪方法,均衡了不同特征提取器的互补信息,较好解决了卫星视频中车辆目标信息不足导致的目标丢失问题,提升了精度。