Adaptive Contour Model for Real-Time Foreground Detection

A multiscale foreground detection method was developed to segment moving objects from a stationary background. The algorithm is based on a fixed-mesh-based contour model, which starts at the bounding box of the difference map between an input image and its background and ends at a final contour.An adaptive algorithm was developed to calculate an appropriate energy threshold to control the contours to identify the foreground silhouettes. Experiments show that this method more successfully ignores the negative influence of image noise to obtain an accurate foreground map than other foreground detection algorithms. Most shadow pixels are also eliminated by this method.     

Real-time vehicle tracking for traffic monitoring systems

A real-time vehicle tracking method is proposed for traffic monitoring system at road intersec-tions, and the vehicle tracking module consists of an initialization stage and a tracking stage .Li-cense plate location based on edge density and color analysis is used to detect the license plate re -gion for tracking initialization .In the tracking stage , covariance matching is employed to track the license plate .Genetic algorithm is used to reduce the computational cost .Real-time image tracking of multi-lane vehicles is achieved .In the experiment , test videos are recorded in advance by record-ers of actual E-police systems at several different city intersections .In the tracking module , the av-erage false detection rate and missed plates rate are 1.19%, and 1.72%, respectively.     

Vehicle recognition and tracking based on simulated annealing chaotic particle swarm optimization-Gauss particle filter algorithm

Target recognition and tracking is an important research filed in the surveillance industry.Traditional target recognition and tracking is to track moving objects, however, for the detected moving objects the specific content can not be determined.In this paper, a multi-target vehicle recognition and tracking algorithm based on YOLO v5 network architecture is proposed.The specific content of moving objects are identified by the network architecture, furthermore, the simulated annealing chaotic m...     

A New Method of Color Edge Detection Based on Local Structure Analysis

Human＇s real life is within a colorful world. Compared to the gray images, color images contain more information and have better visual effects. In today＇s digital image processing, image segmentation is an important section for computers to ＂understand＂ images and edge detection is always one of the most important methods in the field of image segmentation. Edges in color images are considered as local discontinuities both in color and spatial domains. Despite the intensive study based on integration of single-channel edge detection results, and on vector space analysis, edge detection in color images remains as a challenging issue.     

Spatio-temporal Compensation Based Object Detection for Video Surveillance Systems

Moving object detection in video surveillance is an important step. This paper addresses an automatic object detection algorithm based on spatio-temporal compensation for video surveillance. Temporal difference of the pairs of two frames with a k-frame distance is utilized to obtain coarse object masks. Usually, object regions in these coarse masks have discontinuous boundaries and some holes. Region growing with the distance constraint is proposed to compensate these coarse object regions in spatial domain, followed by filling holes. The added distance constraint can prevent object regions from growing infinitely. The proposed filling holes method is simple and effective. To solve the temporarily stopping problem of moving objects, temporal compensation is proposed to compensate the object mask by utilizing temporal coherence of moving objects in temporal domain. The proposed detection algorithm can extract moving objects as completely as possible. Experimental results have successfully demonstrated the validity of the proposed algorithm.     

Improved hand tracking algorithm in video sequences for intelligent rehabilitation

Intelligent rehabilitation system is an active research topic. It is motivated by the increased number of limb disabled patients. Human motion tracking is the key technology of intelligent rehabilitation system, because the movement of limb disabled patients needs to be localized and learned so that any undesired motion behavior can be corrected in order to reach an expectation. This paper introduces a realtime tracking system of human hand motion, specifically intent to be used for home rehabilitation. Vision sensor (camera) is employed in this system to track the hand movement, and the improved Camshift algorithm and Kalman filter are used to implement dynamic hand tracking in the video. CAMSHIFT algorithm is able to track any kind of target colors by building a histogram distribution of the H channel in HSV color space from the region of interests selected by users at the initial stage. Kalman filter is able to predict hand location in one image frame based on its location data detected in the previous frame. The experimental results show that this system can track 2D hand motion and has acceptable accuracy by using the two algorithms properly. The new algorithm proposed in this paper can not only deal with the skin color interference problems, but also deal well with the track of complex background.     

Scale-based spatial data model for GIS

Being the primary media of geographical information and the elementary objects manipulated, almost all of maps adopt the layer-based model to represent geographic information in the existent GIS. However,it is difficult to extend the map represented in layer-based model. Furthermore,in Web-Based GIS,It is slow to transmit the spatial data for map viewing. In this paper,for solving the questions above,we have proposed a new method for representing the spatial data. That is scale-based model. In this model we represent maps in three levels：scale-view,maps in a set of map layers,named Scale-Lastly,a prototype Web-Based GIS using described briefly.     

Multi-features Based Approach for Moving Shadow Detection

IntroductionAs the core of smart surveillance system, image analysistechnology has been used to many fields including home andcommunity surveillance, and transportation monitoring. Themain aim is to detect special moving objects, such as vehicle,walking person, etc. By computing object features as motion,shape, color, texture, centroid, and area, the events in thescene can be evaluated. The requirements of moving objectdetection is the tracking application should be robus…     

Active Tracking with Pan—Tilt—Zoom Camera

This paper intends to introduce the active vision system with autonomous robot competition in the background.It presents some characteristics of the human oculomotor system in active machine vision system:smooth pursuit and saccade.Besides,zoom tracking is used to continuous adjustment of a camera′s focal length to keep a constant sized image of an object moving along the camera′s optical axis.Experiments indicate the technology to be efficient for tracking the ball in the robot competition.     

Clustering approach based on hierarchical expansion for community detection of scientific collaboration network①

This study presents a clustering algorithm based on hierarchical expansion to solve the problem of community detection in scientific collaboration network .The characteristics of achievements infor-mation related to scientific and technological domains are analyzed , and then an ontology that repre-sents their latent collaborative relations is built to detect clusters from the collaboration network .A case study is conducted to collect a data set of research achievements in the electric vehicle field and better clustering results are obtained .A hierarchical recommendation framework that enriches the domain ontologies and retrieves more relevant information resources is proposed in the last part of this paper .This work also lays out a novel insight into the exploitation of scientific collaboration net-work to better classify achievements information .     

联合熵流与边缘算子的红外运动目标检测

针对复杂背景下红外目标检测问题,提出一种联合熵流与边缘算子的红外运动目标检测算法.该算法分4个步骤检测目标:首先将熵流作为一种图像运动描绘子,通过熵流确定含目标运动的区域;其次运用Canny算子检测含背景和目标的边缘;然后联合两者捕获背景成分急剧降低、近似精确的目标边缘图像;最后计算边缘图像聚类中心,依据亮度、熵流约束进行区域生长,识别红外目标.对于海空背景下的红外目标图像,实验结果表明该方法能准确地检测、定位目标.     

非高斯噪声背景下计算机视觉目标跟踪方法

针对杂波背景下计算机视觉目标跟踪问题,提出一种非高斯噪声背景下计算机视觉目标跟踪方法.在视频目标运动模型和观测模型的基础上引入了柯西混合噪声模型,对非高斯噪声运动目标的状态进行建模;然后,在传统高斯噪声粒子滤波的框架内给出文中方法的具体实现步骤.针对大面积遮挡和夜晚光照改变的极端情况下对路上行驶的车辆进行实时跟踪实验,结果表明:文中方法明显提升极端杂波环境下的目标运动过程的建模精度,有效提升目标跟踪精度.     

基于融合邻域信息的海面运动目标检测

运动视目标检测是视频信息处理的重要研究课题之一.本文提出了一种基于高斯混合模型邻域信息融合的海面运动目标检测算法.该算法融合了背景差分和背景邻域信息差分,充分利用同一幅图像的像素邻域信息得到运动目标的种子点,认为高斯背景差分图像中包含种子点的连通区域为真实前景目标.实验表明,该方法可以避免背景模型在构建或更新阶段对场景的表征不足或错误而造成的误检,对强光下的海杂波也有良好的抑制作用,且对不同的气候环境有较好的鲁棒性.     

基于水平集方法的多运动目标分割

给出了一种对光照变化等实际环境中干扰不敏感的基于水平集和背景减的运动目标检测与分割方法。首先建立了当前帧图像与背景参考帧图像间的纹理差，并将其与灰度差相结合构成差图像；然后，分析了纹理特征，并采用适当措施减小由场景中杂乱运动产生的不利影响，从而实现运动目标的鲁棒检测；最后采用了基于区域的水平集方法，将多个信息有机地结合在一起，实现对多个刚体或非刚体运动目标的分割。实际采集序列图的仿真实验验证了该方法的有效性。     

一种间断观测下的无人机地面目标跟踪方法

提出了一种基于贝叶斯滤波器和适航地图的跟踪算法(BF-Hmap),解决间断观测下的无人机地面目标跟踪问题.采用基于中心线和梯度图的快速道路提取算法生成目标适航地图,该图描述了地面目标与当前地形相关的运动能力.采用贝叶斯滤波器分两步在线更新目标状态的后验概率分布,即在预测阶段考虑适航地图对目标运动的约束,以及在估计阶段充分利用目标的所有观测信息.仿真结果表明,在间断观测下该算法对地面机动目标有着良好的跟踪效果.     

视频序列中运动目标跟踪新方法

提出了一种跟踪视频图像序列中运动目标的新方法.该方法利用一种基于动态信息窗口的自适应背景更新方法解决背景的复杂性问题,结合一种新的计算颜色模型解决运动阴影问题,从而得到具有精确边缘的特定运动目标.计算了运动目标灰度质心,在坐标系中记录其位置,并采用最小二乘法拟合实现了对运动目标的跟踪.实验结果表明这种方法能有效地跟踪并预测视频序列中的运动目标.     

基于特征光流的多运动目标检测跟踪算法与评价

光流车辆检测算法其光流不仅携带了运动物体的运动信息,还包含丰富的三维结构信息,能够在未知场景信息的情况下对运动目标进行准确检测;但传统光流法计算方法复杂、抗噪性能差、处理速度缓慢,无法满足多目标实时检测的实际需求。为提高光流法实时检测效率,同时保持较好的检测精度,提出了一种基于Harris特征点光流及卡尔曼滤波模型的多运动目标跟踪算法;并提出新的视频目标检测算法性能评价指标。通过对不同实验场景下多个运动目标的检测与跟踪实验统计结果表明,对比主流Meanshift车辆跟踪算法,检测精度平均提高4.61%;且跟踪持续性提升41.5%,具有更好的鲁棒性及准确性。在时间效率上较比传统光流法平均提升42.9%,能够更好地满足目标跟踪实时性要求。     

动态场景中运动目标检测与跟踪

为了在静态和动态场景中均能实现对运动目标的检测与跟踪,提出了基于运动检测和视频跟踪相结合的视频监控方法. 建立四参数运动仿射模型来描述全局运动,采用块匹配法对其进行参数估计;采用基于全局运动补偿的Horn-Schunck算法检测出运动目标;使用卡尔曼滤波对运动目标的质心位置、宽度和高度进行跟踪. 实验结果表明,该方法能够有效地对静态和动态场景中运动目标进行检测与跟踪.     

基于生成对抗网络的多目标行人跟踪算法

多目标跟踪领域中,在背景复杂、目标遮挡、目标尺度和姿态变换等情况下,容易出现目标丢失、身份交换和跳变等问题.针对这些问题,提出了一种基于检测的多目标跟踪算法,使用改进的YOLO人体人脸关联算法,对当前帧待检目标进行分类和位置检测,使用生成对抗网络构建特征提取模型,学习目标的主要特征以及细微特征,再运用生成对抗网络生成多目标的运动轨迹,最终融和目标的运动信息和外观信息,得到跟踪目标的最优匹配.在MOT16数据集下的实验结果表明,提出的多目标跟踪算法具有较高的精确度和鲁棒性,对比目前身份交换和跳变最少的算法,跳变的次数少了65%,准确度提高了0.25%.     

Motion Tracking with Fast Adaptive Background Subtraction

0　ＩｎｔｒｏｄｕｃｔｉｏｎＩｎａｎｉｎｔｅｌｌｉｇｅｎｔｖｉｄｅｏｓｕｒｖｅｉｌｌａｎｃｅａｎｄｍｏｎｉｔｏｒｉｎｇｓｙｓｔｅｍ ,ｏｎｅｏｆｔｈｅｍｏｓｔｃｏｍｐｌｅｘａｎｄｉｍｐｏｒｔａｎｔｔａｓｋｓｉｓｒｅａｌｔｉｍｅｅｘｔｒａｃｔｉｎｇａｎｄｔｒａｃｋｉｎｇｍｏｔｉｏｎｏｂｊｅｃｔｓ .Ｉｎｇｅｎｅｒａｌ,ｔｈｅｒｅｅｘｉｓｔｔｗｏｍｅｔｈｏｄｓｆｏｒｅｘｔｒａｃｔｉｏｎｎａｍｅｄａｓｆｏｒｅｇｒｏｕｎｄｅｘｔｒａｃｔｉｏｎａｎｄｂａｃｋｇｒｏｕｎｄｓｕｂｔｒａｃｔｉｏｎ ,ｒｅｓｐｅｃｔｉｖ…