CN104463900A - Method for automatically tracking target among multiple cameras - Google Patents

Method for automatically tracking target among multiple cameras Download PDF

Info

Publication number
CN104463900A
CN104463900A CN201410855139.4A CN201410855139A CN104463900A CN 104463900 A CN104463900 A CN 104463900A CN 201410855139 A CN201410855139 A CN 201410855139A CN 104463900 A CN104463900 A CN 104463900A
Authority
CN
China
Prior art keywords
target
targets
interest
camera
detected
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN201410855139.4A
Other languages
Chinese (zh)
Inventor
朱爱民
聂凯
梁丽杰
吴昊
张丽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
TIANJIN HGXIANGYUN INFORMATION TECHNOLOGY Co Ltd
Original Assignee
TIANJIN HGXIANGYUN INFORMATION TECHNOLOGY Co Ltd
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 TIANJIN HGXIANGYUN INFORMATION TECHNOLOGY Co Ltd filed Critical TIANJIN HGXIANGYUN INFORMATION TECHNOLOGY Co Ltd
Priority to CN201410855139.4A priority Critical patent/CN104463900A/en
Publication of CN104463900A publication Critical patent/CN104463900A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/20Analysis of motion
    • G06T7/292Multi-camera tracking
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10016Video; Image sequence

Abstract

The invention provides a method for automatically tracking a target among multiple cameras. The method comprises the following steps that a camera group is initialized; one camera is selected at will, interested target selecting and marking are carried out through a video display window of the camera, the interested target is initialized, the characteristic information of the interested target is extracted, and an online target detection model is established and is shared; a current data frame of the camera group is detected, and whether the interested target is detected or not is judged; if yes, the target is marked at a display window of the corresponding camera, and the new characteristic information of the interested target is updated to the online target detection model to enrich the machine learning experience; the next frame of the camera group is read, and the third step is repeatedly carried out till the tracking process is externally interrupted and ended. Real-time, effective, continuous, stable and unique tracking of the interested target among the cameras can be achieved.

Description

Target method for automatic tracking between a kind of multiple-camera
Technical field
The invention belongs to safety monitoring technical field, relate to target method for automatic tracking between a kind of multiple-camera.
Background technology
In public safety strick precaution field, target tracking is the important precautionary measures of a class, has been widely used in national defence and civilian every field, as intelligent monitoring, man-machine interaction, medical diagnosis, navigational guidance etc.Existing target tracking can only utilize target to carry out at the characteristic information of current camera, and tracking effect is unstable, only can realize tenacious tracking in current camera, once target is blocked, disappears, then follow the tracks of failure; When target occurs in another video camera, to target again characteristic information extraction, tracking can only be re-started, original trace information can be lost, thus tracking target that cannot be effective, complete, also cause target information bulk deposition simultaneously, cannot effectively be combined.
Summary of the invention
Can only carrying out on single video camera for existing target tracking and can not follow the trail of and the problem such as target tracking is inaccurate, unstable in the indirect force of multiple-camera, the invention provides target method for automatic tracking between a kind of multiple-camera, to solving the problem.
Target method for automatic tracking between described a kind of multiple-camera, comprises the following steps:
Step 1: camera cluster initialization, and capture camera group code stream showing in respective display window;
Step 2: select a road video camera arbitrarily, targets of interest selection and mark is carried out at its display window, and initialization is carried out to this target, set up target on-line checkingi model, extract targets of interest feature, targets of interest characteristic information is recorded to target on-line checkingi model, and target on-line checkingi model is shared in camera cluster;
Step 3: by contrasting with target on-line checkingi model, camera cluster current data frame being detected, determines whether targets of interest to be detected;
As detected, then at corresponding video camera display window, this targets of interest is marked, and the new characteristic information of targets of interest is updated to target on-line checkingi model;
As do not detected, then corresponding video camera display window only shows current data two field picture, and target on-line checkingi model is not updated;
Step 4: read next Frame of camera cluster, repeats step 3, until external interrupt terminates tracing process.
Further, between this multiple-camera in target method for automatic tracking, describedly camera cluster current data frame is detected synchronously carried out by multithreading by contrasting with target on-line checkingi model, each thread process No. one video camera current data frame.
Further, between this multiple-camera in target method for automatic tracking, described by contrasting with target on-line checkingi model, a wherein road video camera current data frame being detected, as targets of interest detected, processing by the following method:
When having and a targets of interest only detected, at corresponding video camera display window, this targets of interest is marked, and characteristic information new for this targets of interest is updated to target on-line checkingi model;
When multiple doubtful targets of interest being detected, choose maximum one of similarity after contrasting with target on-line checkingi model as final targets of interest, at corresponding video camera display window, this targets of interest is marked, and characteristic information new for this targets of interest is updated to target on-line checkingi model.
The present invention passes through the various characteristic information real-time update of targets of interest in target on-line checkingi model, in the present frame of multiple camera video, carry out targets of interest detection (comparing with target on-line checkingi model) simultaneously, as long as it appears in camera cluster display window, namely mark it, thus achieve targets of interest between multiple video camera real-time, effective, continuous, stable, uniquely follow the trail of.
Accompanying drawing explanation
Fig. 1 is schematic flow sheet of the present invention
Embodiment
Elaborate below in conjunction with accompanying drawing 1 pair of specific embodiments of the invention.
Target method for automatic tracking between a kind of multiple-camera of the present invention, comprises the following steps:
Step 1: complete the initial work to camera cluster, and show in the video camera display window of capture camera group code Liu Bingge road;
Step 2: select a road video camera arbitrarily, targets of interest selection and mark is carried out at its display window, and initialization is carried out to this target, set up target on-line checkingi model, extract targets of interest feature, targets of interest characteristic information is recorded to target on-line checkingi model, and target on-line checkingi model is shared in camera cluster;
Specifically, namely after selecting targets of interest, the extraneous rectangle frame of targets of interest is drawn in display window, and extract the positive and negative samples of the various attitudes (rotation, convergent-divergent, translation etc.) of targets of interest, and it is recorded to target on-line checkingi model, and target on-line checkingi model is shared in camera cluster.
Step 3: by contrasting with target on-line checkingi model, camera cluster current data frame being detected, determines whether targets of interest to be detected;
As detected, then at corresponding video camera display window, this targets of interest is marked, and characteristic information new for targets of interest is updated to target on-line checkingi model;
As do not detected, then corresponding video camera display window only shows current data two field picture, and target on-line checkingi model is not updated;
Step 4: read next Frame of camera cluster, repeats step 3, until external interrupt terminates tracing process.
Testing process in step 3,4 is all the time along with the real-time update of targets of interest positive and negative samples information in target on-line checkingi model, thus enriched target on-line checkingi model content, by machine learning, further ensure targets of interest between multiple video camera real-time, effective, continuous, stable, uniquely follow the trail of.
Between this multiple-camera in target method for automatic tracking, described by contrasting with target on-line checkingi model, camera cluster current data frame is detected and is synchronously carried out by multithreading, each thread process No. one video camera current data frame, camera cluster current data frame synchronization processes, each road video camera current data frame of targets of interest detected, all need to carry out targets of interest mark, and characteristic information new for targets of interest is updated to target on-line checkingi model respectively.
Specifically, between this multiple-camera in target method for automatic tracking, described by contrasting with target on-line checkingi model, a wherein road video camera current data frame being detected, as targets of interest detected, processing by the following method:
When having and a targets of interest only detected, at corresponding video camera display window, this targets of interest is marked, and characteristic information new for this targets of interest is updated to target on-line checkingi model;
When multiple doubtful targets of interest being detected, choose maximum one of similarity after contrasting with target on-line checkingi model as final targets of interest, at corresponding video camera display window, this targets of interest is marked, and characteristic information new for this targets of interest is updated to target on-line checkingi model.
Above section Example of the present invention has been described in detail, but described content being only preferred embodiment of the present invention, can not being considered to for limiting practical range of the present invention.All equalizations done according to the present patent application scope change and improve, and all should still belong within patent covering scope of the present invention.

Claims (3)

1. a target method for automatic tracking between multiple-camera, is characterized in that: comprise the following steps:
Step 1: camera cluster initialization, and capture camera group code stream showing in respective display window;
Step 2: select a road video camera arbitrarily, targets of interest selection and mark is carried out at its display window, and initialization is carried out to this target, set up target on-line checkingi model, extract targets of interest feature, targets of interest characteristic information is recorded to target on-line checkingi model, and target on-line checkingi model is shared in camera cluster;
Step 3: by contrasting with target on-line checkingi model, camera cluster current data frame being detected, determines whether targets of interest to be detected;
As detected, then at corresponding video camera display window, this targets of interest is marked, and the new characteristic information of targets of interest is updated to target on-line checkingi model;
As do not detected, then corresponding video camera display window only shows current data two field picture, and target on-line checkingi model is not updated;
Step 4: read next Frame of camera cluster, repeats step 3, until external interrupt terminates tracing process.
2. target method for automatic tracking between a kind of multiple-camera according to claim 1, it is characterized in that: described by contrasting with target on-line checkingi model, camera cluster current data frame is detected and is synchronously carried out by multithreading, each thread process No. one video camera current data frame.
3. target method for automatic tracking between a kind of multiple-camera according to claim 2, it is characterized in that: described by contrasting with target on-line checkingi model, a wherein road video camera current data frame being detected, as targets of interest detected, processing by the following method:
When having and a targets of interest only detected, at corresponding video camera display window, this targets of interest is marked, and characteristic information new for this targets of interest is updated to target on-line checkingi model;
When multiple doubtful targets of interest being detected, choose maximum one of similarity after contrasting with target on-line checkingi model as final targets of interest, at corresponding video camera display window, this targets of interest is marked, and characteristic information new for this targets of interest is updated to target on-line checkingi model.
CN201410855139.4A 2014-12-31 2014-12-31 Method for automatically tracking target among multiple cameras Pending CN104463900A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410855139.4A CN104463900A (en) 2014-12-31 2014-12-31 Method for automatically tracking target among multiple cameras

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410855139.4A CN104463900A (en) 2014-12-31 2014-12-31 Method for automatically tracking target among multiple cameras

Publications (1)

Publication Number Publication Date
CN104463900A true CN104463900A (en) 2015-03-25

Family

ID=52909882

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410855139.4A Pending CN104463900A (en) 2014-12-31 2014-12-31 Method for automatically tracking target among multiple cameras

Country Status (1)

Country Link
CN (1) CN104463900A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106023252A (en) * 2016-05-16 2016-10-12 浙江理工大学 Multi-camera human body tracking method based on OAB algorithm
CN106373143A (en) * 2015-07-22 2017-02-01 中兴通讯股份有限公司 Adaptive method and system
CN106559645A (en) * 2015-09-25 2017-04-05 杭州海康威视数字技术股份有限公司 Based on the monitoring method of video camera, system and device
WO2021114702A1 (en) * 2019-12-10 2021-06-17 中国银联股份有限公司 Target tracking method, apparatus and system, and computer-readable storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101751677A (en) * 2008-12-17 2010-06-23 中国科学院自动化研究所 Target continuous tracking method based on multi-camera
CN101883261A (en) * 2010-05-26 2010-11-10 中国科学院自动化研究所 Method and system for abnormal target detection and relay tracking under large-range monitoring scene
US20120078097A1 (en) * 2010-09-27 2012-03-29 Siemens Medical Solutions Usa, Inc. Computerized characterization of cardiac motion in medical diagnostic ultrasound
CN103679155A (en) * 2013-12-30 2014-03-26 深圳锐取信息技术股份有限公司 Multi-target automatic tracking system and method
CN103679687A (en) * 2012-09-18 2014-03-26 杭州海康威视数字技术股份有限公司 Target tracking method of intelligent tracking high-speed dome camera

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101751677A (en) * 2008-12-17 2010-06-23 中国科学院自动化研究所 Target continuous tracking method based on multi-camera
CN101883261A (en) * 2010-05-26 2010-11-10 中国科学院自动化研究所 Method and system for abnormal target detection and relay tracking under large-range monitoring scene
US20120078097A1 (en) * 2010-09-27 2012-03-29 Siemens Medical Solutions Usa, Inc. Computerized characterization of cardiac motion in medical diagnostic ultrasound
CN103679687A (en) * 2012-09-18 2014-03-26 杭州海康威视数字技术股份有限公司 Target tracking method of intelligent tracking high-speed dome camera
CN103679155A (en) * 2013-12-30 2014-03-26 深圳锐取信息技术股份有限公司 Multi-target automatic tracking system and method

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106373143A (en) * 2015-07-22 2017-02-01 中兴通讯股份有限公司 Adaptive method and system
CN106559645A (en) * 2015-09-25 2017-04-05 杭州海康威视数字技术股份有限公司 Based on the monitoring method of video camera, system and device
CN106559645B (en) * 2015-09-25 2020-01-17 杭州海康威视数字技术股份有限公司 Monitoring method, system and device based on camera
CN106023252A (en) * 2016-05-16 2016-10-12 浙江理工大学 Multi-camera human body tracking method based on OAB algorithm
WO2021114702A1 (en) * 2019-12-10 2021-06-17 中国银联股份有限公司 Target tracking method, apparatus and system, and computer-readable storage medium

Similar Documents

Publication Publication Date Title
CN104463900A (en) Method for automatically tracking target among multiple cameras
CN105069408A (en) Video portrait tracking method based on human face identification in complex scenario
MX2018013364A (en) Systems and methods for presenting content.
CN103003844A (en) Monitoring system and method of monitoring
WO2012154989A3 (en) Point-of-view object selection
CN108040221B (en) Intelligent video analysis and monitoring system
CN105183758A (en) Content recognition method for continuously recorded video or image
CN103413395A (en) Intelligent smoke detecting and early warning method and device
CN105469043A (en) Gesture recognition system
CN102915755A (en) Method for extracting moving objects on time axis based on video display
CN104298748A (en) Device and method for face search in videos
PH12018501518A1 (en) Method for detecting frequent passer-passing in monitoring video
CN103646242B (en) Extended target tracking based on maximum stable extremal region feature
CN103150736A (en) Camera motion detecting method based on video monitoring
CN103473533B (en) Moving Objects in Video Sequences abnormal behaviour automatic testing method
CN110910665A (en) Signal lamp control method and device and computer equipment
CN204305213U (en) The interactive sighting device of multi-cam human body tracking
de Carvalho et al. Anomaly detection with a moving camera using multiscale video analysis
PH12017501392A1 (en) A system, method and apparatus for determining the disposition of structural features present in borehole cores
CN104092923A (en) Power transmission line intelligent video distance measurement method
CN103914850A (en) Automatic video labeling method and system based on motion matching
CN103020981A (en) Rapid key frame extraction algorithm based on video moving target
CN102984601A (en) Generation system for video abstract of camera
CN105513088A (en) Detection and tracking system of moving target
US10176695B2 (en) Real-time water safety analysis based on color-movement tracking

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20150325