Learning patterns of activity using real-time tracking

Our goal is to develop a visual monitoring system that passively observes moving objects in a site and learns patterns of activity from those observations. For extended sites, the system will require multiple cameras. Thus, key elements of the system are motion tracking, camera coordination, activity classification, and event detection. In this paper, we focus on motion tracking and show how one can use observed motion to learn patterns of activity in a site. Motion segmentation is based on an adaptive background subtraction method that models each pixel as a mixture of Gaussians and uses an online approximation to update the model. The Gaussian distributions are then evaluated to determine which are most likely to result from a background process. This yields a stable, real-time outdoor tracker that reliably deals with lighting changes, repetitive motions from clutter, and long-term scene changes. While a tracking system is unaware of the identity of any object it tracks, the identity remains the same for the entire tracking sequence. Our system leverages this information by accumulating joint co-occurrences of the representations within a sequence. These joint co-occurrence statistics are then used to create a hierarchical binary-tree classification of the representations. This method is useful for classifying sequences, as well as individual instances of activities in a site     

A system for learning statistical motion patterns

Analysis of motion patterns is an effective approach for anomaly detection and behavior prediction. Current approaches for the analysis of motion patterns depend on known scenes, where objects move in predefined ways. It is highly desirable to automatically construct object motion patterns which reflect the knowledge of the scene. In this paper, we present a system for automatically learning motion patterns for anomaly detection and behavior prediction based on a proposed algorithm for robustly tracking multiple objects. In the tracking algorithm, foreground pixels are clustered using a fast accurate fuzzy k-means algorithm. Growing and prediction of the cluster centroids of foreground pixels ensure that each cluster centroid is associated with a moving object in the scene. In the algorithm for learning motion patterns, trajectories are clustered hierarchically using spatial and temporal information and then each motion pattern is represented with a chain of Gaussian distributions. Based on the learned statistical motion patterns, statistical methods are used to detect anomalies and predict behaviors. Our system is tested using image sequences acquired, respectively, from a crowded real traffic scene and a model traffic scene. Experimental results show the robustness of the tracking algorithm, the efficiency of the algorithm for learning motion patterns, and the encouraging performance of algorithms for anomaly detection and behavior prediction.     

Hierarchical database for a multi-camera surveillance system

This paper presents a framework for event detection and video content analysis for visual surveillance applications. The system is able to coordinate the tracking of objects between multiple camera views, which may be overlapping or non-overlapping. The key novelty of our approach is that we can automatically learn a semantic scene model for a surveillance region, and have defined data models to support the storage of tracking data with different layers of abstraction into a surveillance database. The surveillance database provides a mechanism to generate video content summaries of objects detected by the system across the entire surveillance region in terms of the semantic scene model. In addition, the surveillance database supports spatio-temporal queries, which can be applied for event detection and notification applications.     

Dust particle detection in traffic surveillance video using motion singularity analysis

Dust particle detection in video aims to automatically determine whether the video is degraded by dust particle or not. Dust particles are usually stuck on the camera lends and typically temporally static in the images of a video sequence captured from a dynamic scene. The moving objects in the scene can be occluded by the dusts; consequently, the motion information of moving objects tends to yield singularity. Motivated by this, a dust detection approach is proposed in this paper by exploiting motion singularity analysis in the video. First, the optical model of dust particle is theoretically studied in by simulating optical density of artifacts produced by dust particles. Then, the optical flow is exploited to perform motion singularity analysis for blind dust detection in the video without the need for ground truth dust-free video. More specifically, a singularity model of optical flow is proposed in this paper using the direction of the motion flow field, instead of the amplitude of the motion flow field. The proposed motion singularity model is further incorporated into a temporal voting mechanism to develop an automatic dust particle detection in the video. Experiments are conducted using both artificially-simulated dust-degraded video and real-world dust-degraded video to demonstrate that the proposed approach outperforms conventional approaches to achieve more accurate dust detection.     

Correspondence-Free Activity Analysis and Scene Modeling in Multiple Camera Views

We propose a novel approach for activity analysis in multiple synchronized but uncalibrated static camera views. In this paper, we refer to activities as motion patterns of objects, which correspond to paths in far-field scenes. We assume that the topology of cameras is unknown and quite arbitrary, the fields of views covered by these cameras may have no overlap or any amount of overlap, and objects may move on different ground planes. Using low-level cues, objects are first tracked in each camera view independently, and the positions and velocities of objects along trajectories are computed as features. Under a probabilistic model, our approach jointly learns the distribution of an activity in the feature spaces of different camera views. Then, it accomplishes the following tasks: 1) grouping trajectories, which belong to the same activity but may be in different camera views, into one cluster; 2) modeling paths commonly taken by objects across multiple camera views; and 3) detecting abnormal activities. Advantages of this approach are that it does not require first solving the challenging correspondence problem, and that learning is unsupervised. Even though correspondence is not a prerequisite, after the models of activities have been learned, they can help to solve the correspondence problem, since if two trajectories in different camera views belong to the same activity, they are likely to correspond to the same object. Our approach is evaluated on a simulated data set and two very large real data sets, which have 22,951 and 14,985 trajectories, respectively.     

Motion priors based on goals hierarchies in pedestrian tracking applications

In this paper, the problem of automated scene understanding by tracking and predicting paths for multiple humans is tackled, with a new methodology using data from a single, fixed camera monitoring the environment. Our main idea is to build goal-oriented prior motion models that could drive both the tracking and path prediction algorithms, based on a coarse-to-fine modeling of the target goal. To implement this idea, we use a dataset of training video sequences with associated ground-truth trajectories and from which we extract hierarchically a set of key locations. These key locations may correspond to exit/entrance zones in the observed scene, or to crossroads where trajectories have often abrupt changes of direction. A simple heuristic allows us to make piecewise associations of the ground-truth trajectories to the key locations, and we use these data to learn one statistical motion model per key location, based on the variations of the trajectories in the training data and on a regularizing prior over the models spatial variations. We illustrate how to use these motion priors within an interacting multiple model scheme for target tracking and path prediction, and we finally evaluate this methodology with experiments on common datasets for tracking algorithms comparison.     

Intelligent visual surveillance — A survey

Detection, tracking, and understanding of moving objects of interest in dynamic scenes have been active research areas in computer vision over the past decades. Intelligent visual surveillance (IVS) refers to an automated visual monitoring process that involves analysis and interpretation of object behaviors, as well as object detection and tracking, to understand the visual events of the scene. Main tasks of IVS include scene interpretation and wide area surveillance control. Scene interpretation aims at detecting and tracking moving objects in an image sequence and understanding their behaviors. In wide area surveillance control task, multiple cameras or agents are controlled in a cooperative manner to monitor tagged objects in motion. This paper reviews recent advances and future research directions of these tasks. This article consists of two parts: The first part surveys image enhancement, moving object detection and tracking, and motion behavior understanding. The second part reviews wide-area surveillance techniques based on the fusion of multiple visual sensors, camera calibration and cooperative camera systems.     

基于SAD与UKF-MeanShift的主动目标跟踪

针对复杂场景下动态目标难以准确分割以及目标难以准确定位的问题,提出将绝对差值和(SAD)方法、无迹卡尔曼滤波(UKF)和Mean shift算法相结合的混合自主跟踪动态目标的方法。首先,采用SAD方法获相邻两帧的视差信息,利用视差实现动态目标的检测,并依此建立目标的核直方图描述模型和状态空间模型,然后UKF算法对状态空间进行滤波估计,最后采用Mean shift 算法精确定位目标。实验结果表明该方法不仅能有效检测场景的动态目标,同时还能获得目标的运动信息。文中所提出的基于UKF-Mean shift的跟踪策略与相关算法相比,体现出较好的跟踪效果与时间性能。     

分裂合并运动分割的多运动视觉里程计方法

目的 视觉里程计（visual odometry,VO）仅需要普通相机即可实现精度可观的自主定位,已经成为计算机视觉和机器人领域的研究热点,但是当前研究及应用大多基于场景为静态的假设,即场景中只有相机运动这一个运动模型,无法处理多个运动模型,因此本文提出一种基于分裂合并运动分割的多运动视觉里程计方法,获得场景中除相机运动外多个运动目标的运动状态。方法 基于传统的视觉里程计框架,引入多模型拟合的方法分割出动态场景中的多个运动模型,采用RANSAC（random sample consensus）方法估计出多个运动模型的运动参数实例;接着将相机运动信息以及各个运动目标的运动信息转换到统一的坐标系中,获得相机的视觉里程计结果,以及场景中各个运动目标对应各个时刻的位姿信息;最后采用局部窗口光束法平差直接对相机的姿态以及计算出来的相机相对于各个运动目标的姿态进行校正,利用相机运动模型的内点和各个时刻获得的相机相对于运动目标的运动参数,对多个运动模型的轨迹进行优化。结果 本文所构建的连续帧运动分割方法能够达到较好的分割结果,具有较好的鲁棒性,连续帧的分割精度均能达到近100%,充分保证后续估计各个运动模型参数的准确性。本文方法不仅能够有效估计出相机的位姿,还能估计出场景中存在的显著移动目标的位姿,在各个分段路径中相机自定位与移动目标的定位结果位置平均误差均小于6%。结论 本文方法能够同时分割出动态场景中的相机自身运动模型和不同运动的动态物体运动模型,进而同时估计出相机和各个动态物体的绝对运动轨迹,构建出多运动视觉里程计过程。     

基于视觉的增强现实运动跟踪算法

增强现实系统不仅具有虚拟现实的特点同时具有虚实结合的新特性，为实现虚拟物体与真实物体间的完善结合，必须实时地动态跟踪摄像与真实物体间的相对位置和方向，建立观测模,墼是而通过动态三维显示技术迅速地将虚拟物体添加到真实物体之上，然而目前大多数增强现实系统的注册对象均匀静物体，运动物体的注册跟踪尚很少有人涉足。该算法通过标志点的光流场估计真实环境中运动物体的运动参数，根据透视投影原理和刚体的运动特性确定摄像机与运动物体间的相对位置和方向，实现增强现实系统的运动目标跟踪注册。该算法构架简单、实时性强，易于实现，扩展了增强现实系统的应用范围。     

Structure from Motion Using Sequential Monte Carlo Methods

In this paper, the structure from motion (SfM) problem is addressed using sequential Monte Carlo methods. A new SfM algorithm based on random sampling is derived to estimate the posterior distributions of camera motion and scene structure for the perspective projection camera model. Experimental results show that challenging issues in solving the SfM problem, due to erroneous feature tracking, feature occlusion, motion/structure ambiguity, mixed-domain sequences, mismatched features, and independently moving objects, can be well modeled and effectively addressed using the proposed method.     

目标跟踪研究综述

在监控系统中获取感兴趣的目标是目前极具挑战性的研究热点之一,在视频监控、人机交互和军事领域都具有巨大的应用前景。目标追踪问题主要的技术难点是实时、准确和稳定地将感兴趣的目标表现出来,但是由于目标运动方式、运动场景和目标外在特征的突然改变以及光照变化和拍摄时的抖动等问题都会导致监测追踪系统准确率和稳定性的下降。本文对国内外目标追踪问题的热点和难点进行了详细的分析论述,将目标跟踪归结为目标识别和追踪两个部分来详细讨论,同时考虑到场景对目标追踪的直接影响,将场景理解作为目标追踪的重要技术难点进行了探讨,指出了视觉跟踪模型的具体问题。     

基于运动显著性的移动镜头下的运动目标检测

针对移动镜头下的运动目标检测中的背景建模复杂、计算量大等问题,提出一种基于运动显著性的移动镜头下的运动目标检测方法,在避免复杂的背景建模的同时实现准确的运动目标检测。该方法通过模拟人类视觉系统的注意机制,分析相机平动时场景中背景和前景的运动特点,计算视频场景的显著性,实现动态场景中运动目标检测。首先,采用光流法提取目标的运动特征,用二维高斯卷积方法抑制背景的运动纹理;然后采用直方图统计衡量运动特征的全局显著性,根据得到的运动显著图提取前景与背景的颜色信息;最后,结合贝叶斯方法对运动显著图进行处理,得到显著运动目标。通用数据库视频上的实验结果表明,所提方法能够在抑制背景运动噪声的同时,突出并准确地检测出场景中的运动目标。     

Multi-feature hierarchical topic models for human behavior recognition

Human behavior recognition is one important task of image processing and surveillance system. One main challenge of human behavior recognition is how to effectively model behaviors on condition of unconstrained videos due to tremendous variations from camera motion,background clutter,object appearance and so on. In this paper,we propose two novel Multi-Feature Hierarchical Latent Dirichlet Allocation models for human behavior recognition by extending the bag-of-word topic models such as the Latent Dirichlet Allocation model and the Multi-Modal Latent Dirichlet Allocation model. The two proposed models with three hierarchies including low-level visual features,feature topics,and behavior topics can effectively fuse two different types of features including motion and static visual features,avoid detecting or tracking the motion objects,and improve the recognition performance even if the features are extracted with a great amount of noise. Finally,we adopt the variational EM algorithm to learn the parameters of these models. Experiments on the YouTube dataset demonstrate the effectiveness of our proposed models.     

Rigid body segmentation and shape description from dense opticalflow under weak perspective

We present an algorithm for identifying and tracking independently moving rigid objects from optical flow. Some previous attempts at segmentation via optical flow have focused on finding discontinuities in the flow field. While discontinuities do indicate a change in scene depth, they do not in general signal a boundary between two separate objects. The proposed method uses the fact that each independently moving object has a unique epipolar constraint associated with its motion. Thus motion discontinuities based on self-occlusion can be distinguished from those due to separate objects. The use of epipolar geometry allows for the determination of individual motion parameters for each object as well as the recovery of relative depth for each point on the object. The algorithm assumes an affine camera where perspective effects are limited to changes in overall scale. No camera calibration parameters are required. A Kalman filter based approach is used for tracking motion parameters with time     

Unordered rule discovery using Ant Colony Optimization

In this article,a novel unordered classification rule list discovery algorithm is presented based on Ant Colony Optimization（ACO）. The proposed classifier is compared empirically with two other ACO-based classification techniques on 26 data sets,selected from miscellaneous domains,based on several performance measures. As opposed to its ancestors,our technique has the flexibility of generating a list of IF-THEN rules with unrestricted order. It makes the generated classification model more comprehensible and easily interpretable.The results indicate that the performance of the proposed method is statistically significantly better as compared with previous versions of AntMiner based on predictive accuracy and comprehensibility of the classification model.     

运动显著性概率图提取及目标检测

目的 动态场景图像中所存在的静态目标、背景纹理等静态噪声,以及背景运动、相机抖动等动态噪声,极易导致运动目标检测误检或漏检。针对这一问题,本文提出了一种基于运动显著性概率图的目标检测方法。方法 该方法首先在时间尺度上构建包含短期运动信息和长期运动信息的构建时间序列组;然后利用TFT（temporal Fourier transform）方法计算显著性值。基于此,得到条件运动显著性概率图。接着在全概率公式指导下得到运动显著性概率图,确定前景候选像素,突出运动目标的显著性,而对背景的显著性进行抑制;最后以此为基础,对像素的空间信息进行建模,进而检测运动目标。结果 对提出的方法在3种典型的动态场景中与9种运动目标检测方法进行了性能评价。3种典型的动态场景包括静态噪声场景、动态噪声场景及动静态噪声场景。实验结果表明,在静态噪声场景中,F_score提高到92.91%,准确率提高到96.47%,假正率低至0.02%。在动态噪声场景中,F_score提高至95.52%,准确率提高到95.15%,假正率低至0.002%。而在这两种场景中,召回率指标没有取得最好的性能的原因是,本文所提方法在较好的包络目标区域的同时,在部分情况下易将部分目标区域误判为背景区域的,尤其当目标区域较小时,这种误判的比率更为明显。但是,误判的比率一直维持在较低的水平,且召回率的指标也保持在较高的值,完全能够满足于实际应用的需要,不能抵消整体性能的显著提高。另外,在动静态噪声场景中,4种指标均取得了最优的性能。因此,本文方法能有效地消除静态目标干扰,抑制背景运动和相机抖动等动态噪声,准确地检测出视频序列中的运动目标。结论 本文方法可以更好地抑制静态背景噪声和由背景变化（水波荡漾、相机抖动等）引起的动态噪声,在复杂的噪声背景下准确地检测出运动目标,提高了运动目标检测的鲁棒性和普适性。     

Motion segmentation and qualitative dynamic scene analysis from an image sequence

This article deals with analysis of the dynamic content of a scene from an image sequence irrespective of the static or dynamic nature of the camera. The tasks involved can be the detection of moving objects in a scene observed by a mobile camera, or the identification of the movements of some relevant components of the scene relatively to the camera. This problem basically requires a motion-based segmentation step. We present a motion-based segmentation method relying on 2-D affine motion models and a statistical regularization approach which ensures stable motion-based partitions. This can be done without the explicit estimation of optic flow fields. Besides these partitions are linked in time. Therefore, the motion interpretation process can be performed on more than two successive frames. The ability to follow a given coherently moving region within an interval of several images of the sequence makes the interpretation process more robust and more comprehensive. Identification of the kinematic components of the scene is induced from an intermediate layer accomplishing a generic qualitative motion labeling. No 3-D measurements are required. Results obtained on several real-image sequences corresponding to complex outdoor situations are reported.     

Multi-object detection and tracking by stereo vision

This paper presents a new stereo vision-based model for multi-object detection and tracking in surveillance systems. Unlike most existing monocular camera-based systems, a stereo vision system is constructed in our model to overcome the problems of illumination variation, shadow interference, and object occlusion. In each frame, a sparse set of feature points are identified in the camera coordinate system, and then projected to the 2D ground plane. A kernel-based clustering algorithm is proposed to group the projected points according to their height values and locations on the plane. By producing clusters, the number, position, and orientation of objects in the surveillance scene can be determined for online multi-object detection and tracking. Experiments on both indoor and outdoor applications with complex scenes show the advantages of the proposed system.     

Robust foreground detection in video using pixel layers

A framework for robust foreground detection that works under difficult conditions such as dynamic background and moderately moving camera is presented in this paper. The proposed method includes two main components: coarse scene representation as the union of pixel layers, and foreground detection in video by propagating these layers using a maximum-likelihood assignment. We first cluster into "layers" those pixels that share similar statistics. The entire scene is then modeled as the union of such non-parametric layer-models. An in-coming pixel is detected as foreground if it does not adhere to these adaptive models of the background. A principled way of computing thresholds is used to achieve robust detection performance with a pre-specified number of false alarms. Correlation between pixels in the spatial vicinity is exploited to deal with camera motion without precise registration or optical flow. The proposed technique adapts to changes in the scene, and allows to automatically convert persistent foreground objects to background and re-convert them to foreground when they become interesting. This simple framework addresses the important problem of robust foreground and unusual region detection, at about 10 frames per second on a standard laptop computer. The presentation of the proposed approach is complemented by results on challenging real data and comparisons with other standard techniques.