CN101068342A - Video frequency motion target close-up trace monitoring method based on double-camera head linkage structure - Google Patents

Video frequency motion target close-up trace monitoring method based on double-camera head linkage structure Download PDF

Info

Publication number
CN101068342A
CN101068342A CN 200710017992 CN200710017992A CN101068342A CN 101068342 A CN101068342 A CN 101068342A CN 200710017992 CN200710017992 CN 200710017992 CN 200710017992 A CN200710017992 A CN 200710017992A CN 101068342 A CN101068342 A CN 101068342A
Authority
CN
China
Prior art keywords
target
camera
close
tracking
object
Prior art date
Application number
CN 200710017992
Other languages
Chinese (zh)
Other versions
CN100531373C (en
Inventor
王栋
张云峰
杨杰
朱虹
马展峰
涂善彬
于岩军
王昌军
吴卓林
Original Assignee
西安理工大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西安理工大学 filed Critical 西安理工大学
Priority to CN 200710017992 priority Critical patent/CN100531373C/en
Publication of CN101068342A publication Critical patent/CN101068342A/en
Application granted granted Critical
Publication of CN100531373C publication Critical patent/CN100531373C/en

Links

Abstract

A close up tracking and monitoring method of video movement object based on linkage structure of double camera shooting heads includes using a overall view monitor camera shooting head to carry out identification on object in monitoring region to confirm position and direction as well as movement speed of object, sending object information to close up tracking camera shooting head with direction being controlled by rotary table after tracking object is locked out, obtaining more information of locked object by utilizing close up tracking camera shooting head to display close up picture of locked object.

Description

基于双摄像头联动结构的视频运动目标特写跟踪监视方法 The method of tracking and monitoring the moving object based on the video camera close-up dual linkage structure

技术领域 FIELD

本发明属视频监控技术领域,涉及一种视频运动目标的监控方法,具体涉及一种利用双摄像头联动结构对视频运动目标进行监视的方法。 The present invention belongs to the technical field of video surveillance, it relates to a method for video monitoring of the moving object, particularly relates to a method for monitoring moving objects using a video camera linked double structure.

背景技术 Background technique

随着计算机视频图像处理技术的不断发展,以及安防、反恐等领域的迫切需要,智能化的监控系统要求对入侵目标的行为分析做到一定程度的自动化。 With the urgent need in the field of development of computer video image processing technology, as well as security, anti-terrorism, intelligent monitoring system requirements for behavior analysis intrusion target to achieve a certain degree of automation.

为了保证能够检测出运动目标,视频监视系统的摄像头设定大多为固定模式。 To ensure that detect moving targets, the camera video surveillance system is set mostly fixed mode. 固定模式下的目标检测存在的问题是:如果需要增大目标的分辨率,则监视的场景只能集中在某个局部,例如,机场的登机入口处等。 Problems of Target Detection mode is fixed: If you need to increase the resolution of the target, the monitoring scenario can only focus on a local, for example, such as airport boarding entrance. 如果需要增大监视范围,则无法保证监视画面中的入侵目标细节的刻画。 If you need to increase the monitoring range, there is no guarantee portray target details of the monitor screen invasion. 例如,在小区门口逃逸摩托车的车牌信息等。 For example, the escape door in the cell motorcycle license plate information.

如果既要对全景进行监控,又要对可疑目标进行自动判断,并进行特写跟踪,就需要以“一静一动”的模式来完成。 If we want to monitor the panorama, but also to automatically determine suspicious targets, and close-up track, it needs to be "a static one move" mode to complete.

发明内容 SUMMARY

本发明的目的在于提供一种基于双摄像头联动结构的视频运动目标特写跟踪监视方法,对视频目标即可全景监控,又可进行特写跟踪。 Object of the present invention is to provide a moving object close-up video based tracking dual camera monitoring method linkage structure of a video object to view monitoring, but also for close-up track.

本发明所采用的技术方案是,基于双摄像头联动结构的视频运动目标特写跟踪监视方法,通过一个全景监控摄像头对监视区域的目标进行识别,确定出目标的位置、运行速度和方向,锁定跟踪目标之后,将目标信息传给由云台控制转向的特写跟踪摄像头,由特写跟踪摄像头对锁定目标进行特写、放大后跟踪,显示目标的特写画面,从而获取目标更多信息,该方法按以下步骤进行,a.采用一全景监控摄像头和一设置在云台上的特写跟踪摄像头,采用一计算机实现全景监控摄像头和特写跟踪摄像头之间的联动,即建立场景画面中的每个点分别在两个摄像头所拍摄的视频画面中的位置对应关系,根据全景摄像头中目标的位置来确定特写摄像头的朝向,使之能够正对需要跟踪的目标;b.利用全景监控摄像头对监视区域的目标进行行为检测,即运用图像及视频 Technical proposal of the present invention is based on the video moving object close-up dual camera linkage structure tracking and monitoring methods, to identify the target surveillance area by a panoramic surveillance camera, determining the position of the object, speed and direction, the locking track targets Thereafter, the target information to the steering control by the PTZ camera to track a close-up, close-up by the tracking of the camera for close-targeted, amplified tracking, the close-up picture object, thereby acquiring more information about the target, the method proceeds according to the following steps , a. using Close up of a panoramic surveillance cameras and one disposed on the head tracking camera with a computer-implemented panoramic surveillance cameras and feature tracking linkage between the camera, i.e., to establish each point of scenes in the two cameras, respectively the captured video picture corresponding to the position relation, to determine the orientation close-up camera according to the position in the panoramic camera in the object, to enable them to target facing need to be tracked;. b of the target monitor area behavioral detected using a panoramic surveillance cameras, that is, the use of images and video 处理的方法将运动目标与场景的背景区别分离,获得入侵目标的个数、行走速度、行走方向以及所在位置的信息;c.对上述检测到的目标设置运动方向、目标位置、运动速度、目标颜色四个特征参数,并记录下来,在视频的帧与帧之间运用模板匹配的方法对运动目标进行跟踪,获得每一个目标的运动轨迹信息;d.全景监控摄像头将获得的目标运动轨迹信息通过步骤a建立的联动关系传输给特写跟踪摄像头,通过控制云台的转动带动特写跟踪摄像头转动,进行目标跟踪,实时准确地将锁定的目标经过放大之后显示在监控画面中;e.控制特写跟踪摄像头焦距的缩放来拉近目标,对上述得到的锁定在监控画面中的目标进行特写拍摄。 The background difference method for processing and moving object scene separation, the number of information obtained invasion target, walking speed, location and direction of travel; C above the detected moving direction of the target set, the target position, velocity, target. four color characteristic parameters, and recorded, the method using template matching between frames in the video track of the moving target, a target is obtained for each movement trajectory information;. d target trajectory panoramic surveillance cameras obtain information linkage relationship transmission established by the steps a to close-up the tracking camera, the rotation tracking camera by the rotation drive closeup PTZ control, the target track, in real time accurately locked target after the enlarged display in the monitor screen; E control close-up track. camera focal length of the zoom closer to the target, resulting in the lock above the target monitor screen close-up shots.

本发明方法采用双摄像头联动结构,实现对入侵监视视野的运动目标进行特写跟踪。 The method of the present invention employs dual cameras linkage structure, to achieve moving object intrusion status field of view close-up track. 全景监控摄像头完成的是一个视角比较广的,对某个区域的全景监视。 Panoramic complete surveillance cameras is a relatively wide viewing angle, panoramic surveillance of an area. 虽然全景中的每个目标的分辨率都比较小,很难详细辨认,但是计算机对获得的全景监视视频信息的自动分析,能够获得入侵目标在监视区域中的运行轨迹,动作的过程等等。 Although the resolution panorama of each target are relatively small, it is difficult to identify in detail, but the panorama computer monitor video information obtained by automatic analysis can be obtained attacking the target trajectory in the surveillance area, the operation of the process and so on. 为此,本发明利用全景监控摄像头完成对进入监视区域的目标进行识别,确定出目标的位置、运行速度和方向。 To this end, the present invention is completed using a panoramic camera to monitor the monitoring area into the target identification, to determine the position of the object, speed and direction. 全景监控摄像头在锁定了跟踪目标之后,将检测到的可疑目标的运动速度及方向信息经串口通信传给特写跟踪摄像头。 After the panoramic surveillance cameras locked tracking object, the moving speed of the detected suspicious object and the direction information transmitted via the serial communication Close-tracking camera. 特写跟踪摄像头负责对可疑目标进行特写放大后跟踪,显示目标的特写画面,从而可获取可疑目标的更多信息,弥补了当所得到的视频提供的可疑目标较小或模糊不清无法辨认的不足。 Close-up camera is responsible for tracking suspicious target tracking after close-ups, close-up picture of the target, which can get more information about suspicious target, make up the target suspicious when smaller the resulting video provided or blurred unrecognizable deficiencies.

附图说明 BRIEF DESCRIPTION

图1是本发明方法采用的双摄像头夹角与照射位置关系示意图;图2摄像头夹角与云台控制角的关系示意图;图3视频帧中某像素点的像素值变化曲线,横坐标为时间轴,单位为帧序号;纵坐标为某时刻该点的归一化像素值;图4云台六种基本运动的变化轨迹;图5摄像头焦距样本拟合效果图,横坐标为指令间隔时间,纵坐标为放大倍数。 1 is a schematic dual camera angle and the positional relationship between the irradiation method of the present invention is employed; FIG. 2 showing the relationship between the camera and PTZ control angle angle; change in pixel value of a pixel of a video frame in FIG. 3 curve, the abscissa is time axis, the unit is a frame index; time for a pixel ordinate values ​​were normalized to the point; six basic trajectory of movement of the head 4; Figure 5 the focal length of the camera fitting renderings sample, the abscissa is time command interval, ordinate magnification.

具体实施方式 Detailed ways

下面结合附图和具体实施方式对本发明进行详细说明。 The present invention will be described in detail in conjunction with accompanying drawings and specific embodiments.

本发明基于双摄像头联动结构的视频运动目标特写跟踪监视方法,采用的是两个摄像头联动的结构。 The method of the present invention to track and monitor moving object based video close-up of the dual-camera link configuration, uses two cameras linked structure. 所谓的双摄像头联动结构,是指设置一个固定不动的全景监控摄像头,完成对整个监视区域的全景监视。 A so-called dual camera linkage structure refers to a set of stationary panoramic surveillance camera, complete panoramic monitoring the whole monitoring area. 另一个架设在有两个自由度的可旋转云台上的摄像头,完成对锁定目标的特写跟踪。 Another camera set up in two degrees of freedom of the rotatable head, close-up completion of the targeted track. 当全景监视摄像头发现目标后,对其运动的速度与方向进行判断,并驱动特写跟踪摄像头随着目标的运动而转动跟踪。 When finding the target panoramic surveillance camera, the determination of its speed and direction of movement, and driving a close-up with the movement of the camera to track a target track is rotated. 所谓的特写跟踪,是指跟踪目标摄像头的监视画面只有目标,这样,能够因增大目标的分辨率而使得目标物的正确辨认成为可能。 Close-called tracking means to track the target camera monitoring screen only the target so that the resolution can be increased by proper identification of the target so that the target object becomes possible.

两个摄像头通过一个计算机实现联动。 Two cameras achieved through a computer linkage. 全景摄像头固定在某个位置上,摄像头连接到计算机上的视频卡上,将视频信号传输到计算机中,计算机对输入的视频信号进行处理,并判断出何时有可疑目标出现。 Panoramic camera at a fixed position, the video camera connected to a computer card, the video signal is transmitted to a computer, the computer processes the input video signal, and when it is judged that a suspicious target appears. 一旦判断出有可疑目标出现,就计算其所在位置,以及其运动速度和方向,并将这些参数通过运动控制卡传输出去,控制另一个架设在两自由度的转动云台上的摄像头,转动到目标所在的位置,之后摄像头调整焦距,使该摄像头能够对目标进行放大特写跟踪。 Once it is determined that the target appears suspicious, calculates its location, as well as its velocity and direction, and these parameters are transmitted out through the motion control card, to control another set up in the two degrees of freedom rotation head camera, rotated to where the location of the target, then the camera to adjust the focus, so that the camera can be enlarged close-up of the target track. 当这个特写跟踪摄像头获得目标之后,也将视频信号通过视频卡传输到计算机中,计算机根据目标的运动速度及方向,控制该特写跟踪摄像头随着目标的运动而运动,完成对目标的特写跟踪。 When the close-up tracking camera to obtain the target also transmits the video signal to the computer through the video card, the computer according to the movement speed and direction of the target, and controlling the close-up tracking camera as moving object moves to complete the close-up target tracking.

双摄像头联动方法当全景监控摄像头锁定了入侵目标之后,要使另一个特写跟踪摄像头能够找到该目标物并对其进行放大跟踪,则需要建立两个摄像头之间的联动关系,即建立场景画面中的每个点分别在两个摄像头所拍摄的视频画面中的位置对应关系。 The method of linkage dual camera view monitoring camera when locked after attacking the target, to make another close-up camera to track the object was to find and amplifies the tracking, it is necessary to establish the linkage relationship between the two cameras, i.e. the establishment of scenes each point in the two camera positions are photographed video picture corresponding relationship. 即根据全景摄像头中目标的位置来确定特写摄像头的朝向,使之能够正对需要跟踪的目标。 I.e., toward the close-up camera is determined according to the target position in the panoramic camera to make it capable of facing the target to track. 下面来介绍一下该联动关系建立的过程。 Here to tell us about the process of establishing the linkage relationship.

如图1所示,假定两部摄像头的成像点重合(如图1中的O点),据摄像头的成像原理可知,特写摄像头所对准的位置与它相对于全景摄像头的夹角有关。 1, assume that two imaging camera coincides with a point (O point in FIG. 1), according to the principle of the imaging camera seen, close-up camera is aligned with respect to the position to which it is related to the angle of the panoramic camera.

设全景监视摄像头的方向向量为Φ,是一个固定值,设特写跟踪摄像头的初始方向向量为φ,二者的夹角为θ(Φ,φ)。 Provided panoramic surveillance camera in the direction of the vector [Phi], is a fixed value, the initial tracking direction vector is provided a close-up camera is [Phi], the angle of the both of θ (Φ, φ). 以全景摄像头的方向向量为基准建立水平投影面X和竖直投影面Y。 In the panoramic camera direction vector as a reference to establish a vertical projection and horizontal projection plane X Y. 将夹角投影到两个面上,分解为X方向夹角θX(Φ,φ)和Y方向夹角θY(Φ,φ)。 The angle between the two projected onto the surface, the decomposition of the X-direction angle θX (Φ, φ) and the Y direction angle θY (Φ, φ). 在这里,特写摄像头对准全景摄像头中的(x,y)点,从图中可知如下关系:θX(Φ,φ)=kxx---(1)]]>θY(Φ,φ)=kyy---(2)]]>位置坐标和摄像头夹角成线性关系。 Here, a close-up cameras in the panoramic camera is (x, y) point, it is understood the following relationship from the figure: & theta; X (& Phi;, & phi;) = kxx --- (1)]]> & theta; Y ( & Phi;, & phi;) = kyy --- (2)]]> camera angle and position coordinates of a linear relationship.

有了这样的关系,当需要让特写摄像头指向全景摄像头中的某一点的时候,就可以通过控制两部摄像头的夹角来实现。 With such a relationship, when you need to make a close-up camera pointing at a point when the panoramic camera, can be achieved by controlling the angle between two cameras. 换句话说,如果需要让特写摄像头指向(x,y)点,就控制摄像头使之与全景摄像头的夹角为θ(Ф,φ)。 In other words, if the need to point to a close-up camera (x, y) point, so as to control the camera with a panoramic camera angle is θ (Ф, φ).

这一控制过程是由云台的转动来实现的。 This process is controlled by the rotation of the head to achieve. 下面就是要将对两部摄像头的夹角控制转换为对云台的两个夹角的控制。 Here is the angle between two cameras will be controlled to control the conversion of the two angles of the head. 如图2所示,全景摄像头的方向向量为Φ,它与水平面的夹角设为β,是一个固定值,设特写摄像头的方向向量为φ。 2, the direction of the vector [Phi] is a panoramic camera, which is defined as the angle between the horizontal beta], is a fixed value, the close-up camera is provided for the direction vector φ. 以全景摄像头的照射方向为基准,建立X投影面和Y投影面,分别对应于图中的OAX平面和OAY平面。 In the irradiation direction of the panoramic camera as a reference, the establishment of X and Y projection plane projection plane, corresponding to OAX OAY plane and the plane of FIG.

对特写跟踪摄像头方向向量φ向两个平面上投影,从向量φ上一点B分别向两个投影面做垂线,得到X面的垂点为B',Y面的垂点为Y,Y'是B在向量Φ上的投影点。 Close-up camera tracking direction vector φ projected on two planes, are made from the point B to the two projection plane perpendicular vector φ, to obtain the vertical plane is the point B X ', Y plane is perpendicular to the point Y, Y' B is the point on the projection of the vector Φ. 这样就分别得到两部摄像头X方向夹角θX(Ф,φ)和Y方向夹角θY(Ф,φ),对应于图2中∠AOB′,∠AOY。 Thus were obtained two X-direction camera angle θX (Ф, φ) and the Y direction angle θY (Ф, φ), corresponding to FIG. 2 ∠AOB ', ∠AOY. 另一方面,根据云台的结构可知,云台水平方向的转动夹角对应于图中αX,竖直方向的转动夹角对应于图中的αY。 On the other hand, according to the structure of the head can be seen, the rotation angle corresponding to the head in the horizontal direction in FIG aX, the rotation angle corresponding to the vertical direction in FIG αY.

接下来就是要建立坐标(x,y)与(αX,αY)之间的关系。 The next step is to establish a relationship with (αX, αY) coordinates (x, y). 首先,计算云台的水平转动角αX。 First, a horizontal head rotation angle calculated αX. 根据图中关系,可知:tanθX(Φ,φ)=|Y′B′||Y′O|---(3)]]>tanαX=|YB||YO′|---(4)]]>根据投影关系可知:|YB|=|Y′B′|则由式(3)(4)可得:tanαX=|Y′O||YO′|tanθX(Φ,φ)---(5)]]>图2中 According to the diagram, it is understood: tan & theta; X (& Phi;, & phi;) = | Y & prime; B & prime; || Y & prime; O | --- (3)]]> tan & alpha; X = | YB || YO & prime; | - - (4)]]> The projection relationship found: | YB | = | Y'B '| from equation (3) (4) can be obtained: tan & alpha; X = | Y & prime; O || YO & prime; | tan & theta; X (& Phi;, & phi;) --- (5)]]> 2 in FIG. 则有: There are: |YO|=|Y′O|cosθY(Φ,φ)---(7)]]>由此可得: | YO | = | Y & prime; O | cos & theta; Y (& Phi;, & phi;) --- (7)]]> thus be obtained: 将式(1)(2)式代入(10)得到αX的最终计算公式: The formula (1) (2) into (10) to give a final αX of formula: 接下来再计算云台的竖直转动角αY。 Next recalculated vertical head rotation angle αY. 根据图中关系,可知:|BY|=|BO′|sinαX(12)|YO′|=|BO′|cosαX(13) According to FIG relationship, found: | BY | = | BO '| sinαX (12) | YO' | = | BO '| cosαX (13)

将式(13)代入(6)可得: The formula (13) into (6) yields: 在直角三角形⊥BYO中有如下关系: We have the following relationship in the right triangle ⊥BYO: 在直角三角形⊥BO′O中有如下关系: We have the following relationship in the right triangle ⊥BO'O: 再将式(1),(2)代入(16),得到αY的最终计算公式: Then formula (1), (2) into (16), to give the final αY formula: 根据上面得到的双摄像头的联动关系,首先计算出全景监视摄像头中锁定目标在特写跟踪摄像头中的位置,将特写跟踪摄像头转动至相应的位置,使锁定的目标物在特写跟踪摄像头的监视视野中,之后,特写跟踪摄像头根据判断出的目标运动方向与速度进行相应的转动,完成对目标的特写跟踪。 The linkage relationship between two-camera obtained above, calculated first monitoring View panoramic surveillance camera in targeted close-up track the position of the camera in the rotational Close-tracking camera to the appropriate position, the lock of the object tracking camera close-up of Thereafter, the tracking camera for close-up corresponding to the target rotational speed in accordance with the movement direction judged complete close-up of the target track. 在完成了一个目标的特写跟踪之后,该摄像头复位,并等待下一个目标的出现。 After completion of the close-up of a target tracking, the camera is reset, and wait for the next target.

全景监控摄像头的信息处理全景监视摄像头是固定不动的,该摄像头的作用是,对固定的监视场景中出现的运动目标进行检测,并对其行为进行简单的分析,获得入侵目标的个数、入侵目标的行走速度、行走方向,以及所在位置等信息,可对运动目标的可疑行为进行自动识别并及时报警。 Information panoramic surveillance camera processing panoramic surveillance camera is stationary, the action of the camera is moving targets stationary surveillance scene appearing detected, and simple analysis of their behavior, the number obtained invasion target, attacking the target travel speed information, running direction, location and the like, can automatically identify suspicious behavior of the moving object and timely warning.

根据实际需要可自行设置警戒方式如:入侵警戒区域、警戒线,也可以是整个监视区域,以进入监视视野的先后顺序进行跟踪。 It may be provided according to actual needs, such as self-alert mode: intrusion warning area, cordon, or may be the whole monitoring area, to monitor the field of view of the order in which they enter the track. 当标识完成之后,系统就对进入警戒区域内的目标进行检测,在检测出目标后,还可对监视场景中的人数进行统计,并可对在监视视野中的人物聚众进行自动判断并报警。 When the identification is completed, the system enters the target within the surveillance area is detected, after detecting the target, but also on the number of surveillance scene statistics, and the characters in the monitoring field of view of the gathering and automatically determine alarm.

1)运动目标的检测下面以人物作为运动目标来进行说明。 1) following the detection of the moving object character to be described as a moving target. 所谓的运动目标检测就是指运用图像及视频处理的手段将运动的目标(人)与场景的背景相区别分离。 The so-called target moving object detection refers to the use of image and video processing means of the movement (of people) and distinguished background scene separation. 本发明采用了对背景进行建模的方法,获得一个背景画面(指自动生成的无目标物的监视场景),之后不断对该背景进行更新,以保证在目标检测中,能够适应户外不同时间,不同天气下的光照环境的变化导致的背景变化。 The present invention uses a background modeling method to obtain a background screen (scene monitoring means automatically generated without the target object), then the background and constantly updated, in order to ensure the detection target, it is possible to adapt to different time outdoors, change background change under different lighting conditions due to the weather.

本发明采用了基于像素灰度归类的方法来实现对背景进行建模,在获得不包含运动目标的背景帧之后,采用简单的帧间差方法,就可检测出运动目标。 The present invention uses a method based on the pixel intensity classification to achieve the background model, after obtaining the background frame does not contain the moving target, using a simple inter-frame difference method, can detect moving targets.

像素灰度归类算法是建立在假设背景像素灰度以最大概率出现在图像序列中的前提下的。 Pixel intensity classification algorithm is built on the assumption background pixel grayscale image with the greatest probability in the sequence. 这种假设在实际应用中是可行的,因为只有背景是固定不变的,而运动目标只是短时间内遮挡了背景,而大部分时间内背景是未被遮挡的。 This assumption is feasible in practical applications, because only the background are fixed, and moving objects obscured background only a short time, but most of the time the background is not obscured. 所以,本发明利用灰度差对相应像素点灰度进行归类,选择出现频率最高的灰度值作为该点的背景像素值。 Therefore, the present invention utilizes the gray-scale difference of a corresponding pixel gray level classified, select the highest frequency as the gray value of the background pixel value occurrence point. 具体的算法如下。 The specific algorithm is as follows.

首先,输入一段包含有N帧图像的视频,之后,对该N帧图像中的每个像素的灰度值分布进行统计,在本专利中,灰度值是经过归一化处理后的结果。 First, the input section comprising N video frame image, then, the gray scale value of each pixel in the frame image N statistical distribution, in this patent, gray scale value is the result of the normalization process. 即,设当前点的像素值为f(i,j),则归一化后的像素值x(i,j)为:x(i,j)=f(i,j)/255 (18)这里,255是目前8位位图的最大像素值。 That is, the current set point value of the pixel f (i, j), the normalized value of a pixel of x (i, j) as: x (i, j) = f (i, j) / 255 (18) here, 255 is the maximum pixel value of the 8-bit in FIG.

之后,判断该N帧图像中的灰度分布进行统计,设定波动范围为0.1,则将归一化后的灰度值划分为10个数值等级,对每个像素值给定一个10维的数组Nx(i,j)(k),k=1,2,...,10,则有:Nx(i,j)(k)={N帧图像中,灰度值落在(k-0.05,k+0.05)范围内的像素个数}统计结束之后,找出Nx(i,j)(k),k=1,2,...,10为最大时,对应的k。 Thereafter, the gradation of the N frame image is determined in a statistical distribution, the range of fluctuation is set to 0.1, then the gray value normalized values ​​is divided into 10 levels, each pixel value is given a 10-dimensional an array of Nx (i, j) (k), k = 1,2, ..., 10, there are: Nx (i, j) (k) = {N frame image, the gray values ​​fall (- K- 0.05, k + 0.05)} of the number of pixels in a range after the end of the statistics, find Nx (i, j) (k), k = 1,2, ..., 10 is maximum, the corresponding k.

即:Nmax=max{Nx(i,j)(k),k=1,2,...,10}之后,计算其均值和方差,获得候补背景点灰度分布的置信区间。 That is: Nmax = max {Nx (i, j) (k), k = 1,2, ..., 10} then calculate the mean and variance, confidence intervals obtained candidate background gray level distribution. 这样,实际上就得到了该监视画面的背景模型。 In this way, in fact, get a background model of the monitor screen.

如图3所示,是一段具有155帧视频中的某一像素点的像素值归一化后的变化曲线,横轴表示时间轴,单位为帧序号。 3, a pixel section having a pixel point in the video 155 values ​​were normalized after a curve of the horizontal axis represents the time unit is a frame sequence number. 纵轴为经过归一化后的像素值。 After the longitudinal axis of the normalized pixel value. 从该曲线可知,在该视频段中,该点会在某些时刻被目标物遮挡,统计出现频度较多的部分是归一化后的0.55附近,将其作为背景值,通过聚类处理,获得其分布均值与方差,根据统计数学的置信区间的定义,将落在背景置信区间内的点当作是背景点,否则就判断为候补目标点。 Seen from this graph, in the video segment, the target point is blocked at some point, near 0.55 after statistical occurrence frequency of large portions is normalized, as the background value, clustering process to obtain the distribution mean and variance, according to the definition of statistical confidence interval mathematics, it will fall within the context of confidence-point range as the background point, otherwise it is determined that the candidate target point.

之后,将判断为候补目标的点再进行连通域的特性分析,将判断为非目标的点删除,留下的就是判断为目标的点。 Thereafter, the point is determined as the candidate target domain further communication characteristics analysis, the determination of the non-target point deletion, left it is determined that the target point.

对输入的新一帧的视频画面,判断相应的像素点的灰度值是否落在了置信区间内。 A video picture of the new input, determines the corresponding pixel gray value falls within the confidence interval. 如果不是的话,该点为目标的像素点,这时,背景模型不变。 If not, the pixel point as the goal, then, change the background model. 如果是的话,则表明该点为未被目标遮挡的背景点,则按照下面的公式计算均值与方差,修正背景模型。 If so, it indicates that the target point is not occluded background points, then calculate the mean and variance in accordance with the following formula, background correction model.

均值更新:xt+1(i,j)=( xt(i,j)+xt+1(i,j))/2 (19)其中, xt+1(i,j)表示当前时刻的像素均值, xt(i,j)为前一时刻的像素均值,xt+1(i,j)为当前时刻的像素值。 Mean Update: xt + 1 (i, j) = (xt (i, j) + xt + 1 (i, j)) / 2 (19) wherein, xt + 1 (i, j) represents the current time pixel mean , xt (i, j) is the average of pixels of the previous time, xt + 1 (i, j) is the pixel value of the current time.

显然,按照式(19),可以实现渐消记忆,使得对背景的建模,始终反映当前时刻的光照环境。 Obviously, according to formula (19) can be realized fading memory, so that the modeling of the background, always reflects the light conditions at the current time.

方差更新:σ2t+1(i,j)=Σk=1Nmax(x(i,j)-x‾t+1(i,j))2---(20)]]>这时,用作检测运动目标的背景帧图像为[ xt+1(i,j)]m*n。 Variance updater: & sigma; 2t + 1 (i, j) = & Sigma; k = 1Nmax (x (i, j) -x & OverBar; t + 1 (i, j)) 2 --- (20)]]> time background frame image of the moving object detection as [xt + 1 (i, j)] m * n.

系统在执行监控任务时,对输入的当前帧的像素值x(i,j),采用帧间差的方法按照下式计算:e(i,j)=|x(i,j)- xt+1(i,j)| (21)如果帧间差e(i,j)大于事先设定的阈值,则表明该点为目标点,否则,表明为背景点,将其用式(19),(20)进行更新。 When system monitoring tasks, the pixel value x (i, j) of the current frame input, the interframe difference method is calculated according to the following formula: e (i, j) = | x (i, j) - xt + 1 (i, j) | (21) if the inter-frame difference e (i, j) is greater than the preset threshold value, it indicates that the point of target point, otherwise, indicating the background point, which is represented by the formula (19), (20) to be updated.

2)运动目标的跟踪计算机对全景监控摄像头拍摄到的视频信号按照上面的方法进行检测,一旦发现了目标之后,就要记录所检测到的目标的特性参数。 2) moving target tracking computer panoramic surveillance camera captured the video signal is detected according to the above method, once the target discovery, the detected characteristic parameter of the target will record. 考虑到在监视区域中,有可能出现多个目标,而目标在区域中的运动轨迹是不断发生变化的,为了防止错误跟踪,对一个检测到的目标设置“运动方向”、“目标位置”、“运动速度”、“目标颜色”四个特征参数。 Taking into account in the surveillance area, there may be a plurality of targets, in the region of the target trajectory is constantly changing, and to prevent erroneous tracking of a detected target set "movement direction", "target position", "velocity", "target color" four characteristic parameters. 将这四个特征参数记录下来,在视频的帧与帧之间运用模板匹配的方法对运动目标进行跟踪,并且记录每一个目标的运动轨迹,通过目标的运动轨迹以及其当前位置驱动特写跟踪摄像头,让特写跟踪摄像头找到锁定的目标,并进行特写跟踪。 These four characteristic parameters recorded, from frame to frame used in the method of template matching video moving target tracking, and recording motion trajectory for each target driving trajectory by close-up object and its current location tracking camera so close-up tracking camera to find the target lock, and close-up track.

因为数据传送至特写跟踪摄像头会有一定的时间滞后,因此还需要考虑前后视频帧之间的关系。 Because the tracking data to the close-up camera there will be some time lag, it is also necessary to consider the relationship between the front and rear video frames.

当有多个运动目标进入监视视野,如果运动目标之间没有交叉,本系统通过判断连续两帧上检测出的目标之间的距离来区分目标,并对目标进行编号。 When there are a plurality of moving object enters the monitoring field of view, if there is no intersection between the moving object, the system to distinguish the target is determined by the distance between the target detected on two consecutive frames, and the target number. 同时计算出每个目标的运动速度及方向。 While the calculated moving speed and direction of each object.

在多个目标之间的运动轨迹出现交叉的情况下,则通过对目标运动速度、颜色特征、方向特征进行比较来区分目标,从而获得每个目标的运动轨迹并对其进行记录。 In the case of cross trajectory occurs between a plurality of targets, the target through speed movement, color feature, the direction comparing to distinguish the target features to obtain a trajectory of each object and recorded.

对运动目标速度的统计,采用的是估计目标中心点在不同帧中移动的像素数的方法来实现。 Statistics on the speed of moving targets, using the estimation method of moving the center point of the target number of frames in different pixels to achieve. 目标颜色特征的分析,则是通过对人的体型及人们的日常习惯把颜色特征分成三部分即:头,上身,下身,经过一定比例的分割并统计出每一部分的三个颜色通道的一个均值。 Analysis of target color characteristics, it is by the people and the size of people's daily habits of the color feature that is divided into three parts: head, upper body, lower body, split after a certain proportion of the three color channels and the statistics of each part of a mean . 对目标运动方向的估计,主要是由目标在当前几帧内坐标的变化求差获取目标的运动方向。 Estimation of target direction of motion, mainly by the difference between the acquisition target seeking movement in the direction of the target coordinates of the current change of a few frames. 然后根据这三个特征来区分我们所检测的目标并对其进行编号,实时获取每个运动目标的运动轨迹信息。 Then based on these three features to distinguish our target detected and its serial number, get real-time movement of each moving target track information.

3)可疑行为的报警有了对运动目标轨迹信息的检测与自动分析的方法,就可实现在用户设置的警戒方式下,对目标进入警戒区域、目标穿越警戒线、目标在热点区域聚众等全局性的可疑行为的识别,并在识别出可疑行为后进行报警。 3) With the alarm suspicious behavior detection method and the motion information of the target trajectory of the automatic analyzer, the alert mode can be achieved at a user's setting, the target enters the warning area, the target through the cordon, at certain hot spots like the global mob to the recognition of suspicious behavior, and alarm after recognizing suspicious behavior.

特写跟踪摄像头的控制当全景监控摄像头获得了入侵目标之后,通过前面介绍的双摄像头的联动方式,要求特写跟踪摄像头完成“特写拍摄”及“目标跟踪”两个动作。 After the tracking control closeup view monitoring camera when the camera is obtained attacking the target, by two-camera linked manner described earlier, to complete the required close-up camera tracking "closeup photographing" and "object tracking" two actions. 为此,特写跟踪摄像头的控制,实际上需要完成对架设摄像头的云台转动进行控制,以及对摄像头焦距进行控制。 To this end, the tracking control close-up camera, in fact, need to complete erection of the PTZ camera is controlled, and the focal length of the camera control.

“特写拍摄”就是通过控制摄像头焦距的大小来实现目标的清晰显示,以便于后续的人脸三维重建、目标身份识别。 "Close-up shooting" is to achieve the goal by controlling the size of the focal length of the camera clearly shows, for subsequent three-dimensional reconstruction of the face, the target identification.

“目标跟踪”就是根据传输过来的目标特征参数,控制云台带动摄像头转动,实时准确地将锁定的目标经过放大之后显示在监控画面中。 "Tracking" is transmitted by the target feature parameter, the drive control PTZ camera is rotated, the locking accurate real-time elapsed after amplification of the target displayed on the monitor screen.

1)云台控制云台控制,是为了完成“目标跟踪”的任务。 1) PTZ control PTZ control, in order to complete the "tracking" task. 可采用一个由水平和竖直方向上两个高精度步进电机来驱动,两部电机由MPC07运动控制卡来发送指令控制。 It can be driven by one of the two horizontal and vertical directions with high accuracy stepper motors, two motors are controlled by commands sent to the motion control card MPC07. MPC07控制卡是基于PC机PCI总线的步进电机的上位控制单元,它与PC机构成主从式控制结构。 MPC07 control card is based on an upper stepping motor control unit PC, PCI bus, to which the PC means master-slave control structure. PC机负责人机交互界面的管理和控制系统的实时监控等方面的工作(例如键盘和鼠标的管理、系统状态的显示、控制指令的发送、外部信号的监控等等)。 Aspects of real-time monitoring and management and control system of the PC is responsible for man-machine interface and other work (such as display keyboard and mouse management, system status, send control commands to monitor external signals, etc.). MPC07卡完成运动控制的所有细节(包括脉冲和方向信号的输出、自动升降速的处理、原点和限位等信号的检测等等)。 MPC07 motion control card to complete all the details (including the output pulse signal and direction, like automatic lifting speed detecting process, the origin and the limit signals, etc.).

MPC07控制卡的运动控制功能主要取决于运动函数库。 Motion control MPC07 motion control card depends on the library. 运动函数库为单轴及多轴的步进或伺服控制提供了许多运动函数:单轴运动、多轴独立运动、多轴插补运动等等。 Motion function library provides a number of single-axis and multi-axis motion functions as a stepper or servo control: single-axis motion, multi-axis independent movement, multi-axis interpolation motion and so on. 另外,为了配合运动控制系统的开发,还提供了间隙补偿功能。 Further, in order to meet the development of the motion control system, also provides clearance compensation. 下面简单介绍一下这些功能对应的函数和运动方式。 Following is a brief look at these functions corresponding to the function and movement.

该控制卡提供了六种基本运动类型,列在表1中。 The control card provides six basic types of movement, listed in Table 1. 图4给出了在该控制卡控制下,云台的六种基本运动轨迹。 Figure 4 shows the control under the control card, six basic trajectory of the head.

表1MPC07控制卡的六种基本运动类型 Table 1MPC07 six basic types of motion control card

带有升/降速控制的运动函数称之为快速(fast)运动函数,例如:fast_pmove,fast_vmove和fast_hmove。 With l / down movement control function called fast (FAST) motion functions, for example: fast_pmove, fast_vmove and fast_hmove. 而常速运动函数则称之为常速(con)运动函数,如con_pmove,con_vmove,con_hmove。 The constant velocity motion is called normal speed function (CON) motion functions, such as con_pmove, con_vmove, con_hmove. 此外该卡还提供了多轴独立运动、多轴插补运动等多种运动方式。 In addition, the card also offers a variety of multi-axis motion independence movement, multi-axis interpolation motion and so on.

在跟踪控制上,本系统采用此卡做硬件支持,在VC开发环境下,系统检测到指定目标位置后,位置信息与控制卡的脉冲信号经过一定的转换关系,控制MPC07发送指定的脉冲信号,从而带动特写跟踪摄像头运动来完成对指定目标的跟踪。 In the tracking control, the system uses the card to make hardware support, in the VC development environment, the system detects the specified target position, the pulse signal position information and control cards after a certain conversion relation, the control MPC07 sends the specified pulse signal, thus boosting the close-up camera motion tracking to complete the trace of the specified target.

2)摄像头焦距控制特写跟踪摄像头的焦距可根据实际场景进行事先设置,以保证在监视区域范围内,对人物细节的清楚观察。 2) Close-tracking camera focus control camera focal length can be set in advance in accordance with the actual scene, in order to ensure that the range of the monitoring area, a clear view of the details of the figures.

摄像头焦距控制,是为了完成对目标的“特写拍摄”任务。 Camera focus control, in order to complete "Taking Close" task objectives. 为了能够更清楚的记录目标,为后面的人物正脸检测和面部三维重建打下基础,必须进行特写的拍摄,这就必须调节特写跟踪摄像头的焦距。 In order to more clearly record target, lay back positive for the face detection and facial reconstruction character basis, it must be close-up shot, close-up track which must adjust the focal length of the camera. 本系统采用串口通信技术控制摄像头焦距的缩放来拉近目标给其以特写拍摄。 The system uses serial communication technology to control the focal length of the camera to zoom in closer to their target close-up shots. 在VC开发环境下,采用MSCOMM控件实现串口通信,焦距的控制指令有固定的格式,发送信号的时间长短直接决定了焦距变化的大小。 In the VC development environment implemented using a control instruction MSCOMM control serial communication, a fixed focal length format, the time length of the transmission signal directly determines the magnitude of change in focal length.

表2中,列出了焦距控制指令表。 Table 2 lists the focus control instruction list.

表2摄像头焦距控制指令表 Table 2 a focal length of the camera control instruction table

由于系统采用的摄像头无法直接获得焦距的大小参数,因此,本发明采用了间接的方法来获得焦距的大小。 Because the system uses camera parameters can not be obtained directly focus size, therefore, the present invention uses an indirect method to obtain the size of the focal length. 这里,选取目标的面积与标准视频(假设为320×240像素)大小的比值作为反映焦距大小的标志。 Here, the selected target area of ​​the standard video (assumed to be 320 × 240 pixels) size as a ratio of focal length to reflect the size of the sign. 实际上这两者的比例并不是实际焦距的精确反映,因为即使在相同的焦距下目标人物的大小不仅与焦距有关还与其它客观因素有关,比如大人和小孩在相同焦距下所占的视频比例是不一样的,但对于特写显示而言,目的是将目标放大(或缩小)到一定比例只要能够辨识就达到要求了,并不需要精确反映焦距,所以这样的间接反映关系完全满足系统要求。 In fact the ratio of the two is not precisely reflect the actual focal length, because the size of the target person even in the same focal length and focal length not only related but also with other objective factors, such as adults and children occupied at the same focal length ratio video it is not the same, but for close-up display, the target object is enlarged (or reduced) can be recognized as long as a certain proportion of claims reached, does not need to accurately reflect the focal length, so this indirectly reflect the relationship fully meet the system requirements.

下面再介绍一下本发明中摄像头的数学模型建立过程。 Here again introduce a mathematical model of the invention in the camera setup process. 本摄像头一个重要特点是,在放大过程中,起始大小与终止大小的比值,与放大指令的间隔时间是对应的。 An important feature of this camera is that in the amplification process, the size ratio of the starting and termination of the size, and the time interval corresponding to the instruction is enlarged. 这样就可以根据这一特点对摄像机进行建模。 This camera can be modeled based on this characteristic.

设摄像机初始放大倍数为α0,Tn时刻的摄像机放大倍数为αnΔT=Tk-Tk-1k∈n (22) The initial set magnification camera α0, the camera timing Tn magnification αnΔT = Tk-Tk-1k∈n (22)

假设ΔT保持恒定并且足够小,则可以近似认为在ΔT时间内起始焦距大小和终止焦距大小的比值与ΔT成线性关系,则有:αkαk-1=AΔT---(23)]]>其中,αk表示Tk时刻的摄像机焦距大小,A为常数。 Suppose ΔT remains constant and sufficiently small, it can be approximately considered to be within ΔT time starting focal size and the termination ratio of the focal length of the size of the linear relationship between ΔT so, there are: & alpha; k & alpha; k-1 = A & Delta; T --- ( 23)]]> where, αk represents the size of the camera focal length of the time Tk, a is a constant.

依此类推则有:αnα0=αnαn-1·αn-1αn-2Λα1α0=(AΔT)n---(24)]]>放缩指令的间隔时间t可以被表示为:t=nΔT (25)n=tΔT---(26)]]>将(26)代入(24)得:Δα(t)=αtα0=(AΔT)tΔT---(27)]]>由于A,ΔT都是常数,则(27)可以被转换为:Δα(t)=αtα0=Tt---(28)]]>上式中T=(AΔT)1ΔT]]>为常数。 And so on, there are: & alpha; n & alpha; 0 = & alpha; n & alpha; n-1 & CenterDot; & alpha; n-1 & alpha; n-2 & Lambda; & alpha; 1 & alpha; 0 = (A & Delta; T) n --- (24)]] > interval scaling instruction t may be expressed as: t = nΔT (25) n = t & Delta; T --- (26)]]> of (26) into (24) to give: & Delta; & alpha; (t) = & alpha; t & alpha; 0 = (A & Delta; T) t & Delta; T --- (27)]]> Since A, ΔT is a constant, then (27) can be converted to: & Delta; & alpha; (t) = & alpha ; t & alpha; 0 = Tt --- (28)]]> above formula T = (A & Delta; T) 1 & Delta; T]]> is a constant.

另一方面,考虑到系统放缩焦距时,是通过解码器中继电器的导通与断开来实现的,因此焦距放缩时会存在机械响应时间τ,则对(28)进行调整得:Δα(t)=αtα0=Tt-τ---(29)]]>这样摄像机焦距放大的数学模型就被建立了。 On the other hand, taking into account the focal length of the zoom system, by the decoder is turned on and off to achieve the relay, and thus there is a mechanical response time τ zoom focal length, the pair (28) is adjusted to be: & Delta ; & alpha; (t) = & alpha; t & alpha; 0 = Tt- & tau; --- (29)]]> focal length of the zoom camera such mathematical model was established.

同理:由于焦距的放大与缩小是对称的,则可以得到摄像机焦距缩小的数学模型:Δα(t)=α0αt=Tt-τ---(30)]]>接下来就是要确定参数,在这里,对摄像机的放大倍数和指令间隔时间做了如下的采样测试,实验结果见表3。 Similarly: since amplification focal length and the reduction is symmetric, it is possible to obtain the mathematical model of the camera focal length of the reduced: & Delta; & alpha; (t) = & alpha; 0 & alpha; t = Tt- & tau; --- (30)]]> the next step is to determine the parameters, here, the magnification of the camera interval and instructions made the following test samples, the experimental results shown in Table 3.

表3摄像机放大时间t与放大比例采样表 Table 3 time t camera zoom enlarged scale sample table

根据以上的采样测试实验数据,对以上数学模型进行了拟合,如图11所示,是对采样点进行拟合后得到的拟合曲线。 According to the above experimental data of test samples, the above mathematical model was fitted, as shown in FIG. 11, curve fitting is performed on samples obtained by fitting. 线条1为样本曲线,线条2为拟合出的摄像机焦距放大的数学模型曲线。 Line 1 is a curve of the sample, line 2 is a focal length of the zoom camera fitted mathematical model curve. 这样,就可得到,焦距放大的数学模型为:Δα(t)=αtα0=6.119t-0.5217---(31)]]>焦距缩小的数学模型为:Δα(t)=α0αt=6.119t-0.5217---(32)]]>对摄像机焦距的控制主要目的是为了调整摄像机焦距,使关心的目标物在画面中保持一个理想的大小,以便于观察。 Thus, it is possible to obtain the mathematical model focal amplification is: & Delta; & alpha; (t) = & alpha; t & alpha; 0 = 6.119t-0.5217 --- (31)]]> mathematical model of focal length reduced to: & Delta; & alpha ; (t) = & alpha; 0 & alpha; t = 6.119t-0.5217 --- (32)]]> control main purpose of the camera focal length is to adjust the camera focal length of the object of interest to maintain a desired size in the picture , in order to observe. 根据这一要求,对摄像机焦距的控制可以不需要像云台位置控制那样,对精度有较高的要求。 According to this requirement, the focal length of the camera control may not require that the image head position control, higher precision requirements. 因此不需要采用较复杂的控制方法。 In which case no more complex control method.

在这里,根据上面确定的数学模型,对焦距的控制采用如下算法:视频画面中目标物的大小为m,设给定的目标期望大小为M,则误差e表示为:e=mM---(33)]]>本系统设定允许误差范围为(0.9,1.1),当e∈(0.9,1.1),可以不对摄像机焦距进行调解,否则利用如下公式来确定调节时间:t=lneln6.119+0.5217e>1.1ln1/eln6.119+0.5217e<0.9---(34)]]>通过这种算法,就可以实现根据大小误差调节摄像机焦距的目的,其调节效果能够满足人观察的需要即可。 Here, according to the mathematical model identified above, the focal length of the control algorithm as follows: size of the video picture of the object was m, setting a desired size of a given target is M, the error e expressed as: e = mM --- (33)]]> this system is set tolerance range (0.9,1.1), when e∈ (0.9,1.1), the focal length of the camera can not mediate, or conditioning time is determined using the following equation: t = lneln6.119 + 0.5217e> 1.1ln1 / eln6.119 + 0.5217e & lt; 0.9 --- (34)]]> With this algorithm, the size of the object can be achieved according to adjust the camera focus error, which regulates the effect to meet the needs of people observe It can be. 如果需要提高控制效果,只需要调整误差允许范围即可。 If the need to improve the control effect, only need to adjust the allowable error range.

Claims (5)

1.基于双摄像头联动结构的视频运动目标特写跟踪监视方法,通过一个全景监控摄像头对监视区域的目标进行识别,确定出目标的位置、运行速度和方向,锁定跟踪目标之后,将目标信息传给由云台控制转向的特写跟踪摄像头,由特写跟踪摄像头对锁定目标进行特写、放大后跟踪,显示目标的特写画面,从而获取目标更多信息,其特征在于,该方法按以下步骤进行,a.采用一全景监控摄像头和一设置在云台上的特写跟踪摄像头,采用一计算机实现全景监控摄像头和特写跟踪摄像头之间的联动,即建立场景画面中的每个点分别在两个摄像头所拍摄的视频画面中的位置对应关系,根据全景摄像头中目标的位置来确定特写摄像头的朝向,使之能够正对需要跟踪的目标;b.利用全景监控摄像头对监视区域的目标进行行为检测,即运用图像及视频处理的方法将 1. After based video moving object close-up dual camera linkage structure tracking and monitoring methods, to identify the target surveillance area by a panoramic surveillance camera, determining the position of the object, speed and direction, the locking track the target, the target information to controlled by PTZ steering close-tracking camera, the close-up tracking camera to targeted close-up, enlarged track, close-up picture object, thereby obtaining the target additional information, characterized in that, the method proceeds according to the following steps, a. using a panoramic surveillance camera and a setting close-up on a head tracking camera with a computer-implemented panoramic surveillance cameras and feature tracking linkage between the camera, i.e. the establishment of each point in the scene in the picture are captured at two cameras video picture corresponding to the position relation, to determine the orientation close-up camera according to the position in the panoramic camera in the object, to enable them to target facing need to be tracked; b. a target monitoring area behavioral using panoramic surveillance camera detection, i.e. using the image. and the video processing method 动目标与场景的背景区别分离,获得入侵目标的个数、行走速度、行走方向以及所在位置的信息;c.对上述检测到的目标设置运动方向、目标位置、运动速度、目标颜色四个特征参数,并记录下来,在视频的帧与帧之间运用模板匹配的方法对运动目标进行跟踪,获得每一个目标的运动轨迹信息;d.全景监控摄像头将获得的目标运动轨迹信息通过步骤a建立的联动关系传输给特写跟踪摄像头,通过控制云台的转动带动特写跟踪摄像头转动,进行目标跟踪,实时准确地将锁定的目标经过放大之后显示在监控画面中;e.控制特写跟踪摄像头焦距的缩放来拉近目标,对上述得到的锁定在监控画面中的目标进行特写拍摄。 BACKGROUND difference moving target and scene separation, the number of information obtained invasion target, walking speed, location and direction of travel; C above the detected moving direction of the target set, the target position, velocity, four target color characteristics. parameters, and recorded, using the method of template matching from frame to frame of the video of the moving target tracking, obtaining a motion for each target trajectory information;. d target trajectory information panoramic surveillance cameras obtained by the step a build linkage between transmitted to close-up the tracking camera, the rotation tracking camera by the rotation drive closeup PTZ control, the target track, in real time accurately locked target after the enlarged display in the monitor screen;. e control closeup tracking zoom camera focal length to narrow the target, resulting in the lock above the target monitor screen close-up shots.
2.按照权利要求1所述的方法,其特征在于,所述步骤b中对运动目标进行检测,是采用,首先利用像素灰度归类算法对背景进行建模,获得一个背景帧,不断对该背景进行更新,再采用帧间差方法,将运动目标检测出,具体步骤如下,首先,输入一段包含有N帧图像的视频,设当前点的像素值为f(i,j),归一化后的像素值x(i,j)为:x(i,j)=f(i,j)/255对该N帧图像中的每个像素的灰度值分布进行统计,设定波动范围为0.1,则将归一化后的灰度值划分为10个数值等级,对每个像素值给定一个10维的数组Nx(i,j)(k),k=1,2,...,10,则有:Nx(i,j)(k)={N帧图像中,灰度值落在(k-0.05,k+0.05)范围内的像素个数}统计结束之后,找出Nx(i,j)(k),k=1,2,...,10为最大时,对应的k,即:Nmax=max{Nx(i,j)(k),k=1,2,...,10}计算其均值和方差,获得候补背景点灰度分布的置 2. The method according to claim 1, wherein said step (b) detection of moving objects, using a first algorithm using the pixel intensity classification background model to obtain a background frame, constantly the background update, then inter-frame difference method, the moving object is detected, the following steps, firstly, the input section includes a video image frame N, a current set point value of the pixel f (i, j), normalized pixel values ​​of x (i, j) as: gradation value of each pixel x (i, j) = f (i, j) / N frame image 255 to the statistical distribution, setting the fluctuation range 0.1, then the normalized gray value of a value divided into 10 levels, each pixel value is given a 10-dimensional array of Nx (i, j) (k), k = 1,2, .. ., 10, there are: Nx (i, j) (k) = {N frame image, the gray scale value falls within the number of pixels (k-0.05, k + 0.05)} after the reference end of the range, to find Nx (i, j) (k), k = 1,2, ..., 10 is maximum, the corresponding k, namely: Nmax = max {Nx (i, j) (k), k = 1,2 , ..., 10} calculated mean and variance is set to give the candidate background gray level distribution 信区间,即得到了该监视画面的背景模型;之后,将判断为候补目标的点再进行连通域的特性分析,将判断为非目标的点删除,留下的就是判断为目标的点;对输入的新一帧的视频画面,判断相应的像素点的灰度值是否落在了置信区间内,如果没有落在置信区间内,该点为目标的像素点,这时,背景模型不变;如果落在置信区间内,则表明该点为未被目标遮挡的背景点,则按照下面的公式计算均值与方差,修正背景模型;均值更新:xt+1(i,j)=( xt(i,j)+xt+1(i,j))/2其中, xt+1(i,j)表示当前时刻的像素均值, xt(i,j)为前一时刻的像素均值,xt+1(i,j)为当前时刻的像素值,方差更新:σ2t+1(i,j)=Σk=1Nmax(x(i,j)-x‾t+1(i,j))2]]>这时,用作检测运动目标的背景帧图像为[ xt+1(i,j)]m*n,在执行监控任务时,对输入的当前帧的像素值x(i,j),采用帧间差的方 Confidence interval, i.e., the background model to obtain a monitoring screen; then, it is determined that the candidate of the target characteristic point then communicates domain analysis, the non-deleted target point is determined, left is determined that the target point; for entered a new video picture, it is determined the corresponding pixel gray value falls within the confidence interval, if not fall within the confidence interval, the pixels of the target point, then, the same background model; If falls within the confidence interval, it indicates that the target point is not occluded background points, in accordance with the following formula to calculate the mean and variance, the modified background model; mean update: xt + 1 (i, j) = (xt (i , j) + xt + 1 (i, j)) / 2 where, xt + 1 (i, j) represents the current pixel mean time, xt (i, j) is a pixel before a moment mean, xt + 1 ( i, j) is the pixel value of the current time, variance updater: & sigma; 2t + 1 (i, j) = & Sigma; k = 1Nmax (x (i, j) -x & OverBar; t + 1 (i, j)) 2 ]]> in this case, the background frame image of the moving object detection as [xt + 1 (i, j)] m * n, while monitoring tasks, the current frame input pixel value x (i, j) , an inter-frame difference square 法按照下式计算:e(i,j)=|x(i,j)- xt+1(i,j)|如果帧间差e(i,j)大于事先设定的阈值,则表明该点为目标点,否则,表明为背景点,将其用均值更新公式或方差更新公式进行更新。 Method calculated as follows: e (i, j) = | x (i, j) - xt + 1 (i, j) | If the inter-frame difference e (i, j) is greater than the preset threshold value, it indicates that the point to the target point, otherwise, show the background point, it will be updated with the mean variance update update formula or formula.
3.按照权利要求1所述的方法,其特征在于,所述步骤c中,运用模板匹配的方法对运动目标进行跟踪具体采用,当运动目标之间没有交叉时,通过判断连续两帧上检测出的目标之间的距离来区分目标,并对目标进行编号,同时得出每个目标的运动速度及方向;当多个目标之间的运动轨迹出现交叉时,则通过对目标运动速度、颜色特征、方向特征进行比较来区分目标,从而获得每个目标的运动轨迹并对其进行记录;对运动目标速度的统计,采用的是估计目标中心点在不同帧中移动的像素数的方法来实现;目标颜色特征的分析,则是通过对人的体型及人们的日常习惯把颜色特征分成三部分即:头,上身,下身,经过一定比例的分割并统计出每一部分的三个颜色通道的一个均值;对目标运动方向的估计,由目标在当前几帧内坐标的变化求差获取目标的运动 3. The method according to claim 1, wherein said step (c), a method using template matching to track moving targets the specific use, when there is no intersection between the moving object, is determined by detecting the two consecutive frames a distance between a target to distinguish targets, and the target number obtained while moving speed and direction of each object; when the plurality of trajectory between the target appears intersect, then by target movement speed, color wherein, the direction comparing to distinguish the target features to obtain a trajectory of each moving object and recorded; statistics moving target velocity, using the estimation method center point of the target number of pixels in different frames to achieve; analysis of target color characteristics, it is by the people and the size of people's daily habits of the color feature that is divided into three parts: head, upper body, lower body, split after a certain proportion of the three color channels and the statistics of each part of a mean ; estimating the direction of motion of the target, the target acquired by the target differencing several changes in current moving coordinate frame 方向;然后根据这三个特征来区分检测的目标并对其进行编号,实时获取每个运动目标的运动轨迹信息。 Direction; then detected according to distinguish the target features and subjected to three numbers acquired in real time the movement information of each moving object locus.
4.按照权利要求1所述的方法,其特征在于,所述步骤d中,采用MPC07运动控制卡来发送指令控制云台的转动,MPC07运动控制卡中预存入运动函数库来控制运动方式,当检测到指定目标位置后,位置信息与控制卡的脉冲信号经过一定的转换关系,控制MPC07发送指定的脉冲信号,从而带动特写跟踪摄像头运动来完成对指定目标的跟踪。 4. The method according to claim 1, wherein said step (d) by a rotary motion control card MPC07 send the instructions PTZ control, motion control card MPC07 pre-stored library to control the motion of motion, when the target position is detected, the control pulse signal and the position information of the card after a certain conversion relation, the control MPC07 designated transmission pulse signal, so as to drive the tracking camera motion close-up complete tracking of targeting.
5.按照权利要求1所述的方法,其特征在于,所述步骤e中,对焦距的控制采用如下算法:视频画面中目标物的大小为m,设给定的目标期望大小为M,则误差e表示为:e=mM]]>若误差e在设定的误差允许范围内,不对摄像机焦距进行调解,否则利用如下公式来确定调节时间:t=lneln6.119+0.5217e>1.1ln1/eln6.119+0.5217e<0.9]]>其调节效果能够满足人观察的需要即可。 5. The method according to claim 1, wherein said step e, the focus control algorithm is as follows: the size of the video picture of the object was m, setting a desired size of a given target is M, error e expressed as: e = mM]]> e of the error if the error allowable range is set, a focal length of the camera does not mediate, or conditioning time is determined using the following equation: t = lneln6.119 + 0.5217e> 1.1ln1 / eln6.119 + 0.5217e & lt; 0.9]]> regulating effects which can satisfy the needs of people can observe.
CN 200710017992 2007-06-05 2007-06-05 Video frequency motion target close-up trace monitoring method based on double-camera head linkage structure CN100531373C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200710017992 CN100531373C (en) 2007-06-05 2007-06-05 Video frequency motion target close-up trace monitoring method based on double-camera head linkage structure

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200710017992 CN100531373C (en) 2007-06-05 2007-06-05 Video frequency motion target close-up trace monitoring method based on double-camera head linkage structure

Publications (2)

Publication Number Publication Date
CN101068342A true CN101068342A (en) 2007-11-07
CN100531373C CN100531373C (en) 2009-08-19

Family

ID=38880752

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200710017992 CN100531373C (en) 2007-06-05 2007-06-05 Video frequency motion target close-up trace monitoring method based on double-camera head linkage structure

Country Status (1)

Country Link
CN (1) CN100531373C (en)

Cited By (82)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009079809A1 (en) * 2007-12-07 2009-07-02 Multi Base Limited Video surveillance system with object tracking and retrieval
CN100568266C (en) * 2008-02-25 2009-12-09 北京理工大学 Exception action detecting method based on athletic ground partial statistics characteristic analysis
CN101790041A (en) * 2010-03-31 2010-07-28 天津三星光电子有限公司 Method for realizing specific target prompting function by digital video camera
CN101547344B (en) * 2009-04-24 2010-09-01 清华大学深圳研究生院 Video monitoring device and tracking and recording method based on linkage camera
CN101458169B (en) 2007-12-14 2010-09-15 中国科学院沈阳自动化研究所 Tracking speed measuring method for real time image tracking system
CN101969548A (en) * 2010-10-15 2011-02-09 中国人民解放军国防科学技术大学 Active video acquiring method and device based on binocular camera shooting
CN102072702A (en) * 2010-12-31 2011-05-25 天津市亚安科技电子有限公司 Method for calculating corresponding positions of images of two cameras
CN102096928A (en) * 2011-01-27 2011-06-15 浙江工业大学 Angle-based wandering track detection method in video surveillance
CN102137251A (en) * 2010-01-22 2011-07-27 鸿富锦精密工业(深圳)有限公司 Image monitoring system and method
CN102196240A (en) * 2010-03-18 2011-09-21 鸿富锦精密工业(深圳)有限公司 Pick-up device and method for dynamically sensing monitored object by utilizing same
CN102222338A (en) * 2010-06-04 2011-10-19 微软公司 Automatic depth camera alignment
CN102222342A (en) * 2010-04-16 2011-10-19 上海摩比源软件技术有限公司 Tracking method of human body motions and identification method thereof
CN102238366A (en) * 2010-04-26 2011-11-09 鸿富锦精密工业(深圳)有限公司 Camera of realizing image tracking and monitoring and method thereof
CN102256109A (en) * 2011-06-07 2011-11-23 上海芯启电子科技有限公司 Automatic tracking camera system for multiple targets and focusing method for system
CN102263932A (en) * 2010-05-27 2011-11-30 中兴通讯股份有限公司 A method of video surveillance array control device and system
CN102300039A (en) * 2010-06-28 2011-12-28 佳能株式会社 And a control method for the imaging apparatus
CN102438122A (en) * 2010-09-29 2012-05-02 鸿富锦精密工业(深圳)有限公司 Camera device and method for dynamically detecting monitoring object by using same
CN102509306A (en) * 2011-10-08 2012-06-20 西安理工大学 Specific target tracking method based on video
CN101640788B (en) 2008-07-29 2012-06-20 深圳市朗驰欣创科技有限公司 Method and device for controlling monitoring and monitoring system
CN102567999A (en) * 2010-12-29 2012-07-11 新奥特(北京)视频技术有限公司 Method for applying track data
CN102592127A (en) * 2010-11-24 2012-07-18 索尼公司 Information processing device, information processing method and computer program product
CN102592288A (en) * 2012-01-04 2012-07-18 西安理工大学 Method for matching pursuit of pedestrian target under illumination environment change condition
CN102647559A (en) * 2012-04-26 2012-08-22 广州盈可视电子科技有限公司 Pan-tilt tracing and recording method and device
CN102724386A (en) * 2012-05-18 2012-10-10 浙江捷尚视觉科技有限公司 Real-time video intelligent system and processing method thereof
WO2012151777A1 (en) * 2011-05-09 2012-11-15 上海芯启电子科技有限公司 Multi-target tracking close-up shooting video monitoring system
CN102118607B (en) 2009-12-31 2012-11-21 浙江宇视科技有限公司 Method for monitoring video and encoder
CN101753999B (en) 2008-12-04 2013-08-14 索尼株式会社 Image processing device and method, image processing system, and image processing program
CN103310422A (en) * 2013-06-28 2013-09-18 新晨易捷(北京)科技有限公司 Image acquiring method and device
CN103501423A (en) * 2013-09-18 2014-01-08 苏州景昱医疗器械有限公司 Video monitoring method and device adopting remote program control
CN103561213A (en) * 2013-11-14 2014-02-05 深圳市视晶无线技术有限公司 Wireless shooting management system with auxiliary camera
CN103686080A (en) * 2013-12-03 2014-03-26 深圳如果技术有限公司 Omni-directional intelligent monitoring system
CN101436348B (en) 2008-12-22 2014-04-23 北京中星微电子有限公司 Electronic policeman system and information identification method applied to the same
CN103763519A (en) * 2013-12-31 2014-04-30 南宁市公安局 Automatic monitoring method based on multiple cameras
CN103959764A (en) * 2011-11-29 2014-07-30 科奥比株式会社 Security system for tracking and surveilling object determined as unrecognizable by using surveillance camera and method for providing security service by using the system
CN103957360A (en) * 2014-05-20 2014-07-30 合肥君达高科信息技术有限公司 Control method and device with high-speed particle follow shooting function
CN103985257A (en) * 2014-05-14 2014-08-13 南通大学 Intelligent traffic video analysis method
CN104125433A (en) * 2014-07-30 2014-10-29 西安冉科信息技术有限公司 Moving object video surveillance method based on multi-PTZ (pan-tilt-zoom)-camera linkage structure
CN104125419A (en) * 2014-06-24 2014-10-29 北京理工大学 Self-adaptive resolution ratio implementing method based on CMOS image sensor
CN104202578A (en) * 2014-09-27 2014-12-10 江阴延利汽车饰件股份有限公司 Method for automatically identifying criminal suspects
CN104199425A (en) * 2014-09-15 2014-12-10 中国农业科学院农业信息研究所 Intelligent agricultural monitoring pre-warning system and method
CN104253981A (en) * 2014-09-28 2014-12-31 武汉烽火众智数字技术有限责任公司 Method for sequencing movement objects for video detection according to colors
CN104270570A (en) * 2014-10-17 2015-01-07 北京英泰智软件技术发展有限公司 Binocular video camera and image processing method thereof
CN104301669A (en) * 2014-09-12 2015-01-21 重庆大学 Suspicious target detection tracking and recognition method based on dual-camera cooperation
CN104506776A (en) * 2015-01-09 2015-04-08 成都新舟锐视科技有限公司 Automatic focusing system for real-time balling machine tracking
CN104618650A (en) * 2015-01-30 2015-05-13 苏州阔地网络科技有限公司 Movement detection method and device through combination between front image acquisition device and side image acquisition device
CN104751490A (en) * 2015-04-09 2015-07-01 苏州阔地网络科技有限公司 Target locking method and device based on online classroom
CN105023248A (en) * 2015-06-25 2015-11-04 西安理工大学 Low-SNR (signal to noise ratio) video motion target extraction method
CN105075248A (en) * 2013-03-29 2015-11-18 日本电气株式会社 Target object identifying device, target object identifying method and target object identifying program
CN105491342A (en) * 2015-12-30 2016-04-13 广州励丰文化科技股份有限公司 Intelligent monitoring method and system based on fingerprint and pressure under multi-camera condition
CN105516678A (en) * 2015-12-30 2016-04-20 广州励丰文化科技股份有限公司 Method and system for intelligent monitoring according to contact surface and fingerprints in multi-camera scene
CN105530489A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for intelligently monitoring according to touch pressure in multi-camera scene
CN105530482A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for controlling monitoring under multi-camera scene with pressure change and fingerprint as reference
CN105530488A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for controlling and monitoring reference fingerprint and duration in multi-camera scene
CN105530483A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for controlling monitoring under multi-camera scene with fingerprint and touch surface as reference
CN105530486A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for carrying out intelligent monitoring under multi-camera scene according to touch control area
CN105530487A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for carrying out intelligent monitoring under multi-camera scene according to pressure change and fingerprint
CN105554467A (en) * 2015-12-30 2016-05-04 广州励丰文化科技股份有限公司 Intelligent monitoring method and system according to touch duration in multi-camera scene
CN105554468A (en) * 2015-12-30 2016-05-04 广州励丰文化科技股份有限公司 Touch duration based method and system for controlling monitoring in scene with multiple cameras
CN105657359A (en) * 2015-12-30 2016-06-08 广州励丰文化科技股份有限公司 Intelligent monitoring method and system in multi-camera scene according to pressure change
CN105657363A (en) * 2015-12-30 2016-06-08 广州励丰文化科技股份有限公司 Method and system for controlling monitoring through referring to pressure variation under scene of multiple cameras
CN105657349A (en) * 2015-12-30 2016-06-08 广州励丰文化科技股份有限公司 Method and system for controlling monitoring by referring to touch pressure in multi-camera scene
CN105681739A (en) * 2015-12-30 2016-06-15 广州励丰文化科技股份有限公司 Reference fingerprint and pressure control monitoring method and system in multi-camera scene
CN105681742A (en) * 2015-12-30 2016-06-15 广州励丰文化科技股份有限公司 Method and system for performing intelligent monitoring according to fingerprints and duration in multi-camera scene
CN105844634A (en) * 2016-03-18 2016-08-10 阜阳师范学院 Multi-motion-object video monitoring system and tracking monitoring method thereof
CN105898107A (en) * 2016-04-21 2016-08-24 北京格灵深瞳信息技术有限公司 Target object snapping method and system
CN105939445A (en) * 2016-05-23 2016-09-14 武汉市公安局公共交通分局 Fog penetration shooting method based on binocular camera
WO2016141641A1 (en) * 2015-03-11 2016-09-15 中兴通讯股份有限公司 Method and device for recognizing suspicious person
CN105955311A (en) * 2016-05-11 2016-09-21 阔地教育科技有限公司 Tracking control method, tracking control device and tracking control system
CN105954163A (en) * 2016-04-19 2016-09-21 长沙理工大学 On-line measuring method for sedimentation rate and denudation rate of dirt in condenser
CN105959655A (en) * 2016-07-18 2016-09-21 四川君逸数码科技股份有限公司 Alarm method and device for identifying region invasion by intelligent eye
CN105979209A (en) * 2016-05-31 2016-09-28 浙江大华技术股份有限公司 Monitoring video display method and monitoring video display device
WO2016149938A1 (en) * 2015-03-26 2016-09-29 北京旷视科技有限公司 Video monitoring method, video monitoring system and computer program product
CN106340035A (en) * 2015-07-06 2017-01-18 无锡天脉聚源传媒科技有限公司 Method and device for tracking character in scene
WO2017028047A1 (en) * 2015-08-14 2017-02-23 富士通株式会社 Background model extracting method and apparatus and image processing device
CN106488124A (en) * 2016-10-20 2017-03-08 努比亚技术有限公司 Tracking-locking micro-distance shooting method and terminal
CN106791706A (en) * 2017-01-24 2017-05-31 上海木爷机器人技术有限公司 Method and system for locking objects
CN106899829A (en) * 2017-02-07 2017-06-27 深圳云天励飞技术有限公司 Video processing method and device
CN107026959A (en) * 2016-02-01 2017-08-08 杭州海康威视数字技术股份有限公司 Image collection method and device
CN107566724A (en) * 2017-09-13 2018-01-09 维沃移动通信有限公司 Panoramic image shooting method and mobile terminal
WO2018019135A1 (en) * 2016-07-29 2018-02-01 华为技术有限公司 Target monitoring method, camera, controller and target monitoring system
CN107749950A (en) * 2017-11-09 2018-03-02 睿魔智能科技(东莞)有限公司 Shooting method and system based on deep learning
WO2018171234A1 (en) * 2017-03-24 2018-09-27 上海交通大学 Video processing method and apparatus

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1658670A (en) 2004-02-20 2005-08-24 上海银晨智能识别科技有限公司 Intelligent tracking monitoring system with multi-camera
CN1725230A (en) 2004-07-21 2006-01-25 上海高德威智能交通系统有限公司 Vehicle license tag identification device with lane all-round monitoring function and its identification method

Cited By (107)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101918989B (en) 2007-12-07 2013-02-13 常州环视高科电子科技有限公司 Video surveillance system with object tracking and retrieval
WO2009079809A1 (en) * 2007-12-07 2009-07-02 Multi Base Limited Video surveillance system with object tracking and retrieval
CN101458169B (en) 2007-12-14 2010-09-15 中国科学院沈阳自动化研究所 Tracking speed measuring method for real time image tracking system
CN100568266C (en) * 2008-02-25 2009-12-09 北京理工大学 Exception action detecting method based on athletic ground partial statistics characteristic analysis
CN101640788B (en) 2008-07-29 2012-06-20 深圳市朗驰欣创科技有限公司 Method and device for controlling monitoring and monitoring system
CN101753999B (en) 2008-12-04 2013-08-14 索尼株式会社 Image processing device and method, image processing system, and image processing program
CN101436348B (en) 2008-12-22 2014-04-23 北京中星微电子有限公司 Electronic policeman system and information identification method applied to the same
CN101547344B (en) * 2009-04-24 2010-09-01 清华大学深圳研究生院 Video monitoring device and tracking and recording method based on linkage camera
CN102118607B (en) 2009-12-31 2012-11-21 浙江宇视科技有限公司 Method for monitoring video and encoder
CN102137251A (en) * 2010-01-22 2011-07-27 鸿富锦精密工业(深圳)有限公司 Image monitoring system and method
CN102196240A (en) * 2010-03-18 2011-09-21 鸿富锦精密工业(深圳)有限公司 Pick-up device and method for dynamically sensing monitored object by utilizing same
CN102196240B (en) 2010-03-18 2014-08-20 鸿富锦精密工业(深圳)有限公司 Pick-up device and method for dynamically sensing monitored object by utilizing same
CN101790041A (en) * 2010-03-31 2010-07-28 天津三星光电子有限公司 Method for realizing specific target prompting function by digital video camera
CN102222342A (en) * 2010-04-16 2011-10-19 上海摩比源软件技术有限公司 Tracking method of human body motions and identification method thereof
CN102238366A (en) * 2010-04-26 2011-11-09 鸿富锦精密工业(深圳)有限公司 Camera of realizing image tracking and monitoring and method thereof
WO2011147214A1 (en) * 2010-05-27 2011-12-01 中兴通讯股份有限公司 Control method of video monitor array, device and system thereof
CN102263932A (en) * 2010-05-27 2011-11-30 中兴通讯股份有限公司 A method of video surveillance array control device and system
CN102222338A (en) * 2010-06-04 2011-10-19 微软公司 Automatic depth camera alignment
US9008355B2 (en) 2010-06-04 2015-04-14 Microsoft Technology Licensing, Llc Automatic depth camera aiming
US8648923B2 (en) 2010-06-28 2014-02-11 Canon Kabushiki Kaisha Image pickup apparatus
CN102300039B (en) 2010-06-28 2014-01-08 佳能株式会社 And a control method for the imaging apparatus
CN102300039A (en) * 2010-06-28 2011-12-28 佳能株式会社 And a control method for the imaging apparatus
CN102438122B (en) * 2010-09-29 2016-05-11 鸿富锦精密工业(深圳)有限公司 Camera motion detection means and using its object monitoring method
CN102438122A (en) * 2010-09-29 2012-05-02 鸿富锦精密工业(深圳)有限公司 Camera device and method for dynamically detecting monitoring object by using same
CN101969548A (en) * 2010-10-15 2011-02-09 中国人民解放军国防科学技术大学 Active video acquiring method and device based on binocular camera shooting
CN101969548B (en) 2010-10-15 2012-05-23 中国人民解放军国防科学技术大学 Active video acquiring method and device based on binocular camera shooting
CN102592127A (en) * 2010-11-24 2012-07-18 索尼公司 Information processing device, information processing method and computer program product
CN102567999B (en) * 2010-12-29 2014-11-05 新奥特(北京)视频技术有限公司 Method for applying track data
CN102567999A (en) * 2010-12-29 2012-07-11 新奥特(北京)视频技术有限公司 Method for applying track data
CN102072702B (en) * 2010-12-31 2015-11-18 天津市亚安科技股份有限公司 Calculation method of the image corresponding to the position of a double camera
CN102072702A (en) * 2010-12-31 2011-05-25 天津市亚安科技电子有限公司 Method for calculating corresponding positions of images of two cameras
CN102096928A (en) * 2011-01-27 2011-06-15 浙江工业大学 Angle-based wandering track detection method in video surveillance
WO2012151777A1 (en) * 2011-05-09 2012-11-15 上海芯启电子科技有限公司 Multi-target tracking close-up shooting video monitoring system
CN102256109A (en) * 2011-06-07 2011-11-23 上海芯启电子科技有限公司 Automatic tracking camera system for multiple targets and focusing method for system
CN102509306A (en) * 2011-10-08 2012-06-20 西安理工大学 Specific target tracking method based on video
CN102509306B (en) 2011-10-08 2014-02-19 西安理工大学 Specific target tracking method based on video
CN103959764A (en) * 2011-11-29 2014-07-30 科奥比株式会社 Security system for tracking and surveilling object determined as unrecognizable by using surveillance camera and method for providing security service by using the system
CN102592288A (en) * 2012-01-04 2012-07-18 西安理工大学 Method for matching pursuit of pedestrian target under illumination environment change condition
CN102592288B (en) 2012-01-04 2014-07-02 西安理工大学 Method for matching pursuit of pedestrian target under illumination environment change condition
CN102647559A (en) * 2012-04-26 2012-08-22 广州盈可视电子科技有限公司 Pan-tilt tracing and recording method and device
CN102647559B (en) * 2012-04-26 2016-04-13 广州盈可视电子科技有限公司 A method and apparatus for recording track head
CN102724386A (en) * 2012-05-18 2012-10-10 浙江捷尚视觉科技有限公司 Real-time video intelligent system and processing method thereof
CN105075248A (en) * 2013-03-29 2015-11-18 日本电气株式会社 Target object identifying device, target object identifying method and target object identifying program
CN103310422A (en) * 2013-06-28 2013-09-18 新晨易捷(北京)科技有限公司 Image acquiring method and device
CN103501423A (en) * 2013-09-18 2014-01-08 苏州景昱医疗器械有限公司 Video monitoring method and device adopting remote program control
CN103561213A (en) * 2013-11-14 2014-02-05 深圳市视晶无线技术有限公司 Wireless shooting management system with auxiliary camera
CN103686080A (en) * 2013-12-03 2014-03-26 深圳如果技术有限公司 Omni-directional intelligent monitoring system
CN103763519A (en) * 2013-12-31 2014-04-30 南宁市公安局 Automatic monitoring method based on multiple cameras
CN103985257A (en) * 2014-05-14 2014-08-13 南通大学 Intelligent traffic video analysis method
CN103957360A (en) * 2014-05-20 2014-07-30 合肥君达高科信息技术有限公司 Control method and device with high-speed particle follow shooting function
CN103957360B (en) * 2014-05-20 2017-06-20 合肥君达高科信息技术有限公司 A method and apparatus for controlling high-speed photographing function of particle tracking
CN104125419A (en) * 2014-06-24 2014-10-29 北京理工大学 Self-adaptive resolution ratio implementing method based on CMOS image sensor
CN104125419B (en) * 2014-06-24 2017-08-15 北京理工大学 One resolution implementation based adaptive image sensor cmos
CN104125433A (en) * 2014-07-30 2014-10-29 西安冉科信息技术有限公司 Moving object video surveillance method based on multi-PTZ (pan-tilt-zoom)-camera linkage structure
CN104301669A (en) * 2014-09-12 2015-01-21 重庆大学 Suspicious target detection tracking and recognition method based on dual-camera cooperation
CN104199425A (en) * 2014-09-15 2014-12-10 中国农业科学院农业信息研究所 Intelligent agricultural monitoring pre-warning system and method
CN104202578A (en) * 2014-09-27 2014-12-10 江阴延利汽车饰件股份有限公司 Method for automatically identifying criminal suspects
CN104253981A (en) * 2014-09-28 2014-12-31 武汉烽火众智数字技术有限责任公司 Method for sequencing movement objects for video detection according to colors
CN104253981B (en) * 2014-09-28 2017-11-28 武汉烽火众智数字技术有限责任公司 A method for sorting by color motion video targets for investigation
CN104270570B (en) * 2014-10-17 2018-05-11 北京明泰智软件技术发展有限公司 Binocular camera and image processing method
CN104270570A (en) * 2014-10-17 2015-01-07 北京英泰智软件技术发展有限公司 Binocular video camera and image processing method thereof
CN104506776A (en) * 2015-01-09 2015-04-08 成都新舟锐视科技有限公司 Automatic focusing system for real-time balling machine tracking
CN104618650A (en) * 2015-01-30 2015-05-13 苏州阔地网络科技有限公司 Movement detection method and device through combination between front image acquisition device and side image acquisition device
WO2016141641A1 (en) * 2015-03-11 2016-09-15 中兴通讯股份有限公司 Method and device for recognizing suspicious person
WO2016149938A1 (en) * 2015-03-26 2016-09-29 北京旷视科技有限公司 Video monitoring method, video monitoring system and computer program product
US9792505B2 (en) 2015-03-26 2017-10-17 Beijing Kuangshi Technology Co., Ltd. Video monitoring method, video monitoring system and computer program product
CN105519102B (en) * 2015-03-26 2018-12-28 北京旷视科技有限公司 video monitoring method and video monitoring system
CN104751490B (en) * 2015-04-09 2018-04-17 阔地教育科技有限公司 A method and device for locking the online classroom based on the target
CN104751490A (en) * 2015-04-09 2015-07-01 苏州阔地网络科技有限公司 Target locking method and device based on online classroom
CN105023248A (en) * 2015-06-25 2015-11-04 西安理工大学 Low-SNR (signal to noise ratio) video motion target extraction method
CN105023248B (en) * 2015-06-25 2017-11-03 西安理工大学 A low-SNR video moving object extraction method
CN106340035A (en) * 2015-07-06 2017-01-18 无锡天脉聚源传媒科技有限公司 Method and device for tracking character in scene
WO2017028047A1 (en) * 2015-08-14 2017-02-23 富士通株式会社 Background model extracting method and apparatus and image processing device
CN105657349A (en) * 2015-12-30 2016-06-08 广州励丰文化科技股份有限公司 Method and system for controlling monitoring by referring to touch pressure in multi-camera scene
CN105681739A (en) * 2015-12-30 2016-06-15 广州励丰文化科技股份有限公司 Reference fingerprint and pressure control monitoring method and system in multi-camera scene
CN105681742A (en) * 2015-12-30 2016-06-15 广州励丰文化科技股份有限公司 Method and system for performing intelligent monitoring according to fingerprints and duration in multi-camera scene
CN105657359A (en) * 2015-12-30 2016-06-08 广州励丰文化科技股份有限公司 Intelligent monitoring method and system in multi-camera scene according to pressure change
CN105554468A (en) * 2015-12-30 2016-05-04 广州励丰文化科技股份有限公司 Touch duration based method and system for controlling monitoring in scene with multiple cameras
CN105554467A (en) * 2015-12-30 2016-05-04 广州励丰文化科技股份有限公司 Intelligent monitoring method and system according to touch duration in multi-camera scene
CN105530487A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for carrying out intelligent monitoring under multi-camera scene according to pressure change and fingerprint
CN105530486A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for carrying out intelligent monitoring under multi-camera scene according to touch control area
CN105530483A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for controlling monitoring under multi-camera scene with fingerprint and touch surface as reference
CN105530488A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for controlling and monitoring reference fingerprint and duration in multi-camera scene
CN105530482A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for controlling monitoring under multi-camera scene with pressure change and fingerprint as reference
CN105516678A (en) * 2015-12-30 2016-04-20 广州励丰文化科技股份有限公司 Method and system for intelligent monitoring according to contact surface and fingerprints in multi-camera scene
CN105491342A (en) * 2015-12-30 2016-04-13 广州励丰文化科技股份有限公司 Intelligent monitoring method and system based on fingerprint and pressure under multi-camera condition
CN105530489A (en) * 2015-12-30 2016-04-27 广州励丰文化科技股份有限公司 Method and system for intelligently monitoring according to touch pressure in multi-camera scene
CN105657363A (en) * 2015-12-30 2016-06-08 广州励丰文化科技股份有限公司 Method and system for controlling monitoring through referring to pressure variation under scene of multiple cameras
CN107026959A (en) * 2016-02-01 2017-08-08 杭州海康威视数字技术股份有限公司 Image collection method and device
CN105844634A (en) * 2016-03-18 2016-08-10 阜阳师范学院 Multi-motion-object video monitoring system and tracking monitoring method thereof
CN105844634B (en) * 2016-03-18 2019-04-05 阜阳师范学院 A kind of multiple mobile object tracking monitor method
CN105954163A (en) * 2016-04-19 2016-09-21 长沙理工大学 On-line measuring method for sedimentation rate and denudation rate of dirt in condenser
CN105954163B (en) * 2016-04-19 2018-05-15 长沙理工大学 Method condenser line measurement scale deposition rate and the erosion rate
CN105898107A (en) * 2016-04-21 2016-08-24 北京格灵深瞳信息技术有限公司 Target object snapping method and system
CN105898107B (en) * 2016-04-21 2019-01-25 北京格灵深瞳信息技术有限公司 A kind of target object grasp shoot method and system
CN105955311A (en) * 2016-05-11 2016-09-21 阔地教育科技有限公司 Tracking control method, tracking control device and tracking control system
CN105939445A (en) * 2016-05-23 2016-09-14 武汉市公安局公共交通分局 Fog penetration shooting method based on binocular camera
CN105939445B (en) * 2016-05-23 2019-04-19 武汉市公安局公共交通分局 A kind of Penetrating Fog image capture method based on binocular camera
CN105979209A (en) * 2016-05-31 2016-09-28 浙江大华技术股份有限公司 Monitoring video display method and monitoring video display device
CN105959655A (en) * 2016-07-18 2016-09-21 四川君逸数码科技股份有限公司 Alarm method and device for identifying region invasion by intelligent eye
WO2018019135A1 (en) * 2016-07-29 2018-02-01 华为技术有限公司 Target monitoring method, camera, controller and target monitoring system
CN106488124A (en) * 2016-10-20 2017-03-08 努比亚技术有限公司 Tracking-locking micro-distance shooting method and terminal
CN106791706A (en) * 2017-01-24 2017-05-31 上海木爷机器人技术有限公司 Method and system for locking objects
CN106899829A (en) * 2017-02-07 2017-06-27 深圳云天励飞技术有限公司 Video processing method and device
WO2018171234A1 (en) * 2017-03-24 2018-09-27 上海交通大学 Video processing method and apparatus
CN107566724A (en) * 2017-09-13 2018-01-09 维沃移动通信有限公司 Panoramic image shooting method and mobile terminal
CN107749950A (en) * 2017-11-09 2018-03-02 睿魔智能科技(东莞)有限公司 Shooting method and system based on deep learning

Also Published As

Publication number Publication date
CN100531373C (en) 2009-08-19

Similar Documents

Publication Publication Date Title
Wang et al. Recent developments in human motion analysis
Haritaoglu et al. W^ 4: A real time system for detecting and tracking people
Bobick et al. The recognition of human movement using temporal templates
US7929017B2 (en) Method and apparatus for stereo, multi-camera tracking and RF and video track fusion
KR100911695B1 (en) Image capturing apparatus and image capturing method
Hampapur et al. Smart video surveillance: exploring the concept of multiscale spatiotemporal tracking
US20080075334A1 (en) Combined face and iris recognition system
EP2131328A2 (en) Method for automatic detection and tracking of multiple objects
US7321386B2 (en) Robust stereo-driven video-based surveillance
US20150015787A1 (en) Automatic extraction of secondary video streams
Ballard Animat Vision
Darrell et al. Integrated person tracking using stereo, color, and pattern detection
US7391907B1 (en) Spurious object detection in a video surveillance system
CN101918989B (en) Video surveillance system with object tracking and retrieval
EP1444667B1 (en) Face imaging system for recordal and automated identity confirmation
US7480414B2 (en) Method and apparatus for object normalization using object classification
Benedek et al. Bayesian foreground and shadow detection in uncertain frame rate surveillance videos
US8599266B2 (en) Digital processing of video images
Ren et al. Motion detection with nonstationary background
US20080259179A1 (en) Automatic Multiscale Image Acquisition from a Steerable Camera
EP0878965A2 (en) Method for tracking entering object and apparatus for tracking and monitoring entering object
CN1186923C (en) Abnormal object automatic finding and tracking video camera system
US20070058717A1 (en) Enhanced processing for scanning video
US7574021B2 (en) Iris recognition for a secure facility
Wang et al. Robust video-based surveillance by integrating target detection with tracking

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted
C17 Cessation of patent right