Detection of passenger flow on and off buses based on video images and YOLO algorithm

Bus passenger flow information is very important as a reference data for bus company line optimization, schedule scheduling basis, and passenger travel mode arrangement. With the development of image processing technology, it has become a current research trend to count passenger flow with the help of surveillance video of passengers getting on and off the bus. The specific research contents of this paper based on video image detection and statistics of passengers are as follows:(1) Collect head target image samples through a variety of ways, including 3960 positive head target samples and 4150 negative head target samples, which together constitute the head target feature database. (2) Established a head target detection model based on deep learning. First, the labeling of the head target training data set is completed. Then, after 15,000 iterations of model training, the YOLOv3 head target detection network model was obtained, with a recall rate of 92.12% and an accuracy rate of 89.71%. (3) A multi-target matching tracking algorithm based on the combination of Cam-shift and YOLOv3 is proposed. First, the Cam-shift algorithm is used to track the head target. Secondly, the head target tracking data and the YOLOv3 detection data are combined to solve the problem of drift during the tracking of the Cam-shift algorithm through the data association matching method based on the minimum distance, and then combined with the time constraint, a passenger location information judgment rule is proposed. Optimize the error and missed detection in the process of head target detection and tracking, and improve the reliability of passenger trajectory tracking. (4) A statistical algorithm for the detection of passengers getting on and off the bus is proposed. First, the trajectory of passengers in the bus boarding and disembarking area is analyzed, and a process for judging passengers’ boarding and boarding behavior is proposed. At the same time, a passenger position information judgment rule is proposed according to the different situations of whether there are new passengers or missing passengers, so as to optimize the problem of wrong detection and missing detection in the process of head target detection and tracking. (5) Finally, experiments are carried out in actual bus scenes and simulation scenes. The experiment proves that the statistical algorithm for the detection of passengers getting on and off the bus proposed in this paper has good detection, tracking and statistics effects in bus scenes and simulation scenes.

     

面向智能家居的挥手识别算法

目的 用手势控制家电是智能家居发展的趋势之一,传统的静态手势识别算法难以适应复杂的居家环境,特别当使用广角相机或环境干扰大时,为此提出一种动态的挥手识别算法,可以对视频序列中的挥手动作做出响应,以达到控制家电的目的。方法 挥手动作具有周期性且频率相对稳定,算法首先调整长滤波器和短滤波器使其检测到视频内周期性运动的区域,然后利用人手识别算法对周期性运动区域进行验证并确认人手。结果 通过与主流的手势识别算法的对比,在复杂环境下,本文算法将成功次数提高了3%,误触发次数降低了44%,响应时间也降低了近0.4 s。结论 实验结果表明,算法能够满足实际应用需求。此外,算法不基于运动目标检测,运算量极低,可以在较高的图像分辨率下实时运行,并能被移植到嵌入式平台下。     

基于mean-shift算法的人脸实时跟踪方法*

复杂环境下实时人脸跟踪方法在视觉监控系统中具有很重要的意义,但目前的跟踪算法普遍存在目标遮挡、尺寸变化等过于敏感的不足,限制了其应用范围。提出了一种人脸检测、mean-shift算法与卡尔曼滤波器相结合的实时全自动人脸跟踪算法。实验结果表明该算法实时性很强,可以实现对运动人脸的快速跟踪,同时对目标遮挡也有很好的鲁棒性。     

基于mean-shift算法的人脸实时跟踪方法*

复杂环境下实时人脸跟踪方法在视觉监控系统中具有很重要的意义,但目前的跟踪算法普遍存在目标遮挡、尺寸变化等过于敏感的不足,限制了其应用范围。提出了一种人脸检测、mean-shift算法与卡尔曼滤波器相结合的实时全自动人脸跟踪算法。实验结果表明该算法实时性很强,可以实现对运动人脸的快速跟踪,同时对目标遮挡也有很好的鲁棒性。     

基于蛙眼视觉特性的运动目标模糊化区域理解跟踪方法

动态场景下的运动目标检测与跟踪是计算机视觉研究的前沿方向, 对场景的背景突变和目标的外观突变 的鲁棒性是当前研究的难点所在. 针对这种情形, 本文提出一种基于蛙眼视觉特性的鲁棒跟踪方法. 该方法利用蛙眼视觉认知的生理特性和外部特性, 设计了一种与之相应的模糊化区域理解的运动目标跟踪方法. 针对实验室环境下的动态序列的实验结果验证了方法的有效性; 并进一步将该方法与传统的Canny算子理解结 果及经典的Mean shift算法理解结果进行对比, 显示了方法的优越性.     

基于L2范数最小化的实时目标跟踪

在贝叶斯推理框架下,基于稀疏表示的跟踪算法能够较好地处理目标在视频场景中的各种复杂的外观变化,取得较为鲁棒的跟踪效果,但算法的计算复杂度很高,很难满足实时性要求。针对稀疏跟踪算法的这一问题,提出了一种基于l2范数最小化的实时目标跟踪算法。将PCA子空间目标表示与l2范数最小化进行结合,去除稀疏跟踪算法中常用的琐碎模板集,建立了基于l2范数最小化的目标表示模型以及将遮挡等因素考虑在内的观测似然度函数。在大量的实验测试集上的对比实验结果显示,该算法和多个非常优秀的跟踪算法相比,可以达到相同甚至更高的跟踪精度,而且在多个测试集上可以达到每秒20帧的速度。该算法可以很好地应对视频监控场景中遮挡、光线突变、尺度变化和非刚性形变等干扰,同时算法复杂度低,满足了实时要求。     

深度学习辅助的多行人跟踪算法

目的 目标的长距离跟踪一直是视频监控中最具挑战性的任务之一。现有的目标跟踪方法在存在遮挡、目标消失再出现等情况下往往会丢失目标,无法进行持续有效的跟踪。一方面目标消失后再次出现时,将其作为新的目标进行跟踪的做法显然不符合实际需求;另一方面,在跟踪过程中当相似的目标出现时,也很容易误导跟踪器把该相似对象当成跟踪目标,从而导致跟踪失败。为此,提出一种基于目标识别辅助的跟踪算法来解决这个问题。方法 将跟踪问题转化为寻找帧间检测到的目标之间对应关系问题,从而在目标消失再现后,采用深度学习网络实现有效的轨迹恢复,改善长距离跟踪效果,并在一定程度上避免相似目标的干扰。结果 通过在标准数据集上与同类算法进行对比实验,本文算法在目标受到遮挡、交叉运动、消失再现的情况下能够有效地恢复其跟踪轨迹,改善跟踪效果,从而可以对多个目标进行持续有效的跟踪。结论 本文创新性地提出了一种结合基于深度学习的目标识别辅助的跟踪算法,实验结果证明了该方法对遮挡重现后的目标能够有效的恢复跟踪轨迹,适用在监控视频中对多个目标进行持续跟踪。     

A real-time object detecting and tracking system for outdoor night surveillance

Autonomous video surveillance and monitoring has a rich history. Many deployed systems are able to reliably track human motion in indoor and controlled outdoor environments. However, object detection and tracking at night remain very important problems for visual surveillance. The objects are often distant, small and their signatures have low contrast against the background. Traditional methods based on the analysis of the difference between successive frames and a background frame will do not work. In this paper, a novel real time object detection algorithm is proposed for night-time visual surveillance. The algorithm is based on contrast analysis. In the first stage, the contrast in local change over time is used to detect potential moving objects. Then motion prediction and spatial nearest neighbor data association are used to suppress false alarms. Experiments on real scenes show that the algorithm is effective for night-time object detection and tracking.     

Feature Point Tracking for Incomplete Trajectories

A new algorithm is presented for feature point based motion tracking in long image sequences. Dynamic scenes with multiple, independently moving objects are considered in which feature points may temporarily disappear, enter and leave the view field. This situation is typical for surveillance and scene monitoring applications. Most of the existing approaches to feature point tracking have limited capabilities in handling incomplete trajectories, especially when the number of points and their speeds are large, and trajectory ambiguities are frequent. The proposed algorithm was designed to efficiently resolve these ambiguities. Correspondences between moving points are established in a competitive linking process that develops as the trajectories grow. Appearing and disappearing points are treated in a natural way as the points that do not link. The proposed algorithm compares favorably to efficient alternative algorithms selected and tested in a performance evaluation study. Received: June 8, 1998; revised November 18, 1998     

SCOOP: A Real-Time Sparsity Driven People Localization Algorithm

Detecting and tracking people in scenes monitored by cameras is an important step in many application scenarios such as surveillance, urban planning or behavioral studies to name a few. The amount of data produced by camera feeds is so large that it is also vital that these steps be performed with the utmost computational efficiency and often even real-time. We propose SCOOP, a novel algorithm that reliably localizes people in camera feeds, using only the output of a simple background removal technique. SCOOP can handle a single or many video feeds. At the heart of our technique there is a sparse model for binary motion detection maps that we solve with a novel greedy algorithm based on set covering. We study the convergence and performance of the algorithm under various degradation models such as noisy observations and crowded environments, and we provide mathematical and experimental evidence of both its efficiency and robustness using standard datasets. This clearly shows that SCOOP is a viable alternative to existing state-of-the-art people localization algorithms, with the marked advantage of real-time computations.     

多帧监督的相关滤波无人机目标跟踪

目标跟踪是无人机的关键技术之一。无人机目标跟踪容易因相机运动、尺度变化等场景的影响,导致跟踪漂移或丢失。提出一种多帧监督的相关滤波无人机目标跟踪算法,加入多帧信息,根据视图的像差监督响应图变化率,有效地提高跟踪器的识别能力。采用裁剪矩阵引入真实负样本,并加入多个历史帧信息提高滤波器的鲁棒性。采用欧几里德范数定义响应图的像差,通过监督像差的变化防止跟踪漂移,得到目标的准确位置。根据相似度进行目标模型更新。在UAV123和VisDrone2019数据集上与其他算法对比实验。结果显示该算法在相机运动、尺度变化等场景具有良好的跟踪鲁棒性和精度。     

On hierarchical modelling of motion for workflow analysis from overhead view

Understanding human behaviour is a high level perceptual problem, one which is often dominated by the contextual knowledge of the environment, and where concerns such as occlusion, scene clutter and high within-class variations are commonplace. Nonetheless, such understanding is highly desirable for automated visual surveillance. We consider this problem in a context of a workflow analysis within an industrial environment. The hierarchical nature of the workflow is exploited to split the problem into ‘activity’ and ‘task’ recognition. In this, sequences of low level activities are examined for instances of a task while the remainder are labelled as background. An initial prediction of activity is obtained using shape and motion based features of the moving blob of interest. A sequence of these activities is further adjusted by a probabilistic analysis of transitions between activities using hidden Markov models (HMMs). In task detection, HMMs are arranged to handle the activities within each task. Two separate HMMs for task and background compete for an incoming sequence of activities. Imagery derived from a camera mounted overhead the target scene has been chosen over the more conventional oblique views (from the side) as this view does not suffer from as much occlusion, and it poses a manageable detection and tracking problem while still retaining powerful cues as to the workflow patterns. We evaluate our approach both in activity and task detection on a challenging dataset of surveillance of human operators in a car manufacturing plant. The experimental results show that our hierarchical approach can automatically segment the timeline and spatially localize a series of predefined tasks that are performed to complete a workflow.     

People-flow counting in complex environments by combining depth and color information

People-flow counting is one of the key techniques of intelligence video surveillance systems and the information of people-flow obtained from this technique is an very important evidence for many applications, such as business analysis, staff planning, security, etc. Traditionally, the color image information based methods encounter kinds of challenges, such as shadows, illumination changing, cloth color, etc., while the depth information based methods suffer from lack of texture. In this paper, we propose an effective approach of people-flow counting by combining color and depth information. First, we adopt a background subtraction technique to fast obtain the moving regions on depth images. Second, the water filling algorithm is used to effectively detect head candidates on the moving regions. Then we use the SVM to recognize the real heads from the candidates. Finally, we adopt a weighted K Nearest Neighbor based multi-target tracking method to track each confirmed head and count the people through the surveillance region. Four datasets constructed from two surveillance scenes are used to evaluate the proposed method. Experimental results show that our method outperform the state-of-the-art methods. Our method can work stably on condition of kinds of interruptions and can not only obtain high precisions, but also high recalls on four datasets.     

复杂环境下高效物体跟踪级联分类器

目的 传统跟踪算法在复杂环境下容易发生漂移（drift）现象,通过改进TLD（tracking learning detection）跟踪技术算法,提出了基于Sliding-window的局部搜索和全局搜索策略、积分直方图过滤器和随机Haar-like块特征过滤器。方法 首先,采用积分直方图过滤器可以有效地过滤大量非目标子窗口块,从而减少后续过滤器特征匹配数;其次,利用随机Haar-like块特征过滤器能够解决跟踪算法在复杂环境（多物体、部分或较大区域遮挡、快速运动等）跟踪过程易发生漂移而导致跟踪精度的不足。结果 结合TLD原始过滤器与新提出的两个过滤器组合而成的级联分类器,通过与主流的跟踪算法实验进行对比表明,级联分类器在稳定的背景或复杂环境的跟踪鲁棒性强、跟踪精度高,并且采用了局部和全局搜索策略提高了计算速度。结论 提出的方法在诸多背景环境变化,跟踪物体形变等情况下,能够精确地多尺度跟踪待测目标;结合全局和局部搜索跟踪策略能够有效地克服级联分类器所带来的时间复杂度过高的问题,从而实现实时目标跟踪。     

融合时空信息的运动目标检测算法

传统运动目标检测算法在处理诸如树叶晃动、水面波纹等动态场景时效果不理想。为此,针对动态场景下所存在的背景扰动问题,提出一种融合时间和空间信息的运动目标检测算法。该算法通过增量式主成分分析提取空间上图像的背景信息,结合三帧差分法所提取的时域信息进行融合决策以提取运动目标。实验结果表明,该算法能够在动态场景中有效提取运动目标,且检测结果优于混合高斯模型算法。     

一种有效的实时人群计数方法

对固定镜头下视频序列中运动人体的检测和跟踪方法进行研究,利用灰度图像差分双向投影信息检测人体目标,提出一种基于统计运动区域几何特征固定比例的分割算法,使用最近邻匹配方法对人体进行跟踪。完整地实现了一个有效的实时人群计数系统。大量室内和室外场景实验结果表明,该算法具有很好的实时性(每秒处理25帧~30帧且可并行处理4路视频)、对光照变化的鲁棒性以及对稀疏人群检测精度高等特点。     

基于视觉背景提取的自适应运动目标提取算法

在复杂场景下的视频运动目标提取是视频分析技术的首要工作。为了解决前景运动目标提取的精确度不高的问题,提出一种基于视觉背景提取(ViBE)的改进视频运动目标提取算法(ViBE⁺)。首先,在背景模型初始化阶段采用像素的菱形邻域来简化样本信息;其次,在前景运动目标提取阶段引入自适应分割阈值来适应场景的动态变化;最后,在更新阶段提出背景重建和调整更新因子方法来处理光照变化的情形。实验结果表明,对于复杂视频场景LightSwitch的运动目标提取结果在相似度指标上,改进后的算法与混合高斯模型(GMM)算法、码本模型算法以及原始ViBE算法相比,分别提高了1.3倍、1.9倍以及3.8倍。所提算法能够在有效时间内对复杂场景具有较好的自适应性,且性能明显优于对比算法。     

基于颜色-纹理特征的目标跟踪

针对传统的基于颜色特征目标跟踪算法在一些复杂场景中存在的跟踪不稳定性,提出一种基于颜色 纹理特征的目标跟踪算法;在传统的基于颜色Mean shift的目标跟踪算法中加入纹理特征,在提取目标颜色特征的同时提取目标的纹理特征,并且采取串接原则,在搜索目标新位置时仍然沿用传统的基于颜色的均值漂移跟踪算法,但在每一次迭代过程搜寻目标最佳的位置点即特征相似最大的区域时,利用纹理特征来实现,并且采用八邻域搜索法（候选区域周围扩大八个大小相等的区域）来解决部分遮挡的问题。通过对比实验表明,该算法在复杂场景中表现出的实时性和鲁棒性较好。关键词：     

颜色和纹理特征的运动目标检测

针对复杂场景中运动目标检测这一难题,提出利用RGB颜色特征和尺度不变局部三元模式的运动目标检测算法。利用时域中值法得到估算背景图像并快速初始化背景模型。通过颜色特征、纹理特征相似性度量,融合得出背景概率网络,通过侧抑制滤波提高对比度分类出前景与背景像素,前景像素进一步进行阴影检测,将阴影点归为背景点,但不用于模型更新。将算法与GMM、SC-SOBS、SUBSENS算法在变化检测数据库中进行对比验证。实验表明,新算法在满足实时性的基础上,对动态背景,阴影和相机抖动等有一定的鲁棒性。