CN103414870B - A kind of multi-mode warning analytical method - Google Patents

Abstract

The invention discloses a kind of multi-mode warning analytical method, the steps include: to obtain monitor video in region, large-scale map and high definition remote sensing image, realize the mutual mapping of monitor video and 2D geographical spatial data; According to warning demand, in GIS, Plotted line and face warning figure, form warning region set, and set permission direction, speed, time parameter, thereby form warning rule base, mapped in each associated monitoring video; Obtain the actual motion parameter under position, direction and the GIS mapping of dynamic object by foreground detection, target following; Whether real-time judge dynamic object motion state violates multi-mode warning rule base; In the time having the regular event of the warning of violating to occur, in generalized information system, carry out Realtime Alerts with figure or the alarm audio frequency given prominence to. The method overcome traditional warning arrange single, lack geographical georeferencing, be not easy to the shortcoming that area outlook is analyzed, greatly improved flexibility, the globality of area outlook.

Description

A kind of multi-mode warning analytical method

Technical field

The present invention relates to a kind of multi-mode warning analytical method, specifically, is that one can be set flexibly in single video cameraMultiple warning rule, realizes the multi-mode warning analytical method of warning as required, has overcome unicity that current warning sets and without geographyThe shortcoming of georeferencing.

Background technology

Along with socioeconomic fast development, mankind's activity is day by day frequent, because security protection disappearance causes any property loss, casualtiesAccident is of common occurrence. Therefore, lay video monitoring equipment at important place, warning rule is set, and to carry out in real time warning be region securityThe important means of management, real-time, various, warning method flexibly has important practical significance.

Current, warning analytical method is relatively single, lacks flexibility. The corresponding a kind of warning analytical method of a common video camera,Just need to set up another monitoring probe if carry out another kind warning analysis simultaneously. Each monitoring probe division of labor is clear and definite, as invasion inspectionSurvey, abandon detect, article are moved detection, detection 4 modules of hovering are independent separately. And in actual video monitoring, warning needsAsk and there is flexibility, need can set multiple warning rule in single monitoring scene, and As time goes on change flexibly, fromAnd reach " a tractor serves several purposes ". So, existing all vigilance performance modules are carried out abstract, unified rule flexibly are directly providedSet interface and will meet user flexibility, various warning demand.

Meanwhile, monitor video lacks geographical georeferencing, and under large scene, warning rule is set relatively loaded down with trivial details. When facing several monitoringWhen probe warning rule arranges, security protection personnel can complete fast. But along with socioeconomic development, monitoring probe presentsThe feature of densification. Warning rule is still set in a traditional way one by one will be very loaded down with trivial details, if lacked whole regional geography environmentUnderstanding be just easy to make mistakes. GIS gives spatial data clear and definite georeferencing, can be by geomorphological features exhibition detailed in large sceneNow in map. Therefore, GIS is incorporated in warning analysis, realizes the fusion of monitor video and 2D geographical spatial data, willMake to guard against regular setting facilitation.

Summary of the invention

Key issue to be solved by this invention is to overcome the unicity that in video monitoring, warning rule is set, and builds geographical space numberAccording to the mutual mapping model of monitor video, realize multi-mode warning and analyze, from traditional single warning, rank are set thereby warning is analyzedSection is moved towards flexible, multimodal warning the stage is set, from without geographical space with reference to the stage move towards GIS leading be suitable for large scene,The warning rule of densification video monitoring arranges the stage. Therefore, the present invention proposes multi-mode warning analytical method.

Basic ideas of the present invention: realize the mutual mapping of geographical spatial data and monitor video, make monitor video have clear and definite groundReason georeferencing; Taking geographical spatial data as main, set warning rule, and mapped in each associated monitoring video; Each relevantVideo camera performance objective detects, target following, according to the warning in real time of warning rule.

The basic step of a kind of multi-mode warning of the present invention analytical method is:

The first step, obtain monitor video in region, large-scale map and high definition remote sensing image, according to mapping matrix, realize monitoringThe mutual mapping of video and 2D geographical spatial data;

Under second step, GIS, warning region is set: according to warning demand, draw alert based on map and remote sensing image in GIS platformGuard against line and face, form warning region set;

Under the 3rd step, GIS, warning rule is set: the warning region set based on drawing in second step, set flexibly the direction of motion,Speed and time parameter, thus structure multi-mode is built warning rule base; Wherein, if user specifies one to allow direction, in the time of warning, needSet the direction that forms broken line and polygonal each selections, according to allowing direction and warning region to draw all line segments of direction calculatingDirection; After warning rule sets, based on the mapping matrix of the first step, these warning rules are mapped to each relevant monitoringIn video;

The 4th step, warning in real time: obtain under the position, direction of dynamic object and GIS mapping by foreground detection, target followingActual motion parameter; Whether real-time judge dynamic object motion state violates multi-mode warning rule base; Violate warning rule when havingWhen event occurs, in generalized information system, carry out Realtime Alerts with figure or the alarm audio frequency given prominence to.

In the described first step, realize the mutual mapping of monitor video and 2D geographical spatial data, its concrete steps are:

(a) obtain height, inclination angle, horizontal view angle and the azimuth of each video camera, based on these 4 each monitoring of calculation of parameterThe ground monitoring region of probe, i.e. trapezoidal guarded region corresponding to each monitoring probe;

(b) according to the position of video camera, by the Coordinate Conversion of each trapezoidal guarded region to 2D map of living in or remote sensing imageCoordinate, and form guarded region figure layer;

(c) according to trapezoidal geographical coordinate and the corresponding image coordinate of trapezoidal each point, resolve homography matrix, look thereby realize monitoringFrequently the mutual mapping of image and 2D geographical spatial data.

A kind of multi-mode warning analytical method of the present invention is the mutual mapping method based on monitor video and geographical spatial data, userDraw warning line and face and possess clear and definite georeferencing, warning arranges and can under unified interface, complete, and has overcome traditional warning settingSingle, lack geographical georeferencing, be not easy to the shortcoming that area outlook is analyzed, greatly improved flexibility, the entirety of area outlookProperty.

Brief description of the drawings

Fig. 1 is camera supervised area schematic, is (a) camera supervised area schematic in local coordinate system; (b) be to throwCamera supervised area schematic in shadow coordinate system;

Fig. 2 is trapezoidal guarded region figure layer schematic diagram;

Fig. 3 monitors the ken and image coordinate corresponding relation schematic diagram;

Fig. 4 (a) draws warning line and allows direction schematic diagram on remote sensing image; (b) be on remote sensing image, to draw warningPolygon schematic diagram;

Fig. 5 (a) is that warning allows direction setting procedure figure; (b) be that warning line allows direction setting method and each line segment permission sideTo schematic diagram;

In Fig. 6 (a) map, ray mode warning rule maps to video schematic diagram; (b) in map, polygon Mode warning rule is reflectedBe incident upon video schematic diagram;

Fig. 7 guards against flow chart in real time.

Detailed description of the invention

Below in conjunction with drawings and Examples, the present invention is described in further detail.

The first step, monitor video and geographical spatial data shine upon mutually concrete steps and are:

(1) obtain photo centre's geographical coordinate of each video monitoring probe, highly, inclination angle, horizontal view angle, azimuth, andObtain the detailed 2D geographical spatial data in region, comprise map and remote sensing image; According to the relevant parameter of video camera, calculate its groundFace guarded region, i.e. a trapezoid area, its coordinate is projection coordinate's system, as Fig. 1;

(2) in GIS, create guarded region figure layer, corresponding each monitoring probe ground monitoring region polygon is added toIn this figure layer, as Fig. 2.

(3) calculate homography matrix according to the coordinate of trapezoidal each point and corresponding image coordinate thereof, this matrix is monitor video and 2DSpatial data is the core of mapping mutually.

Second step, under GIS, warning region setting concrete steps are:

(1) in GIS platform, create the figure layer of line and Noodles type;

(2) draw warning line and polygon with remote sensing image according to the map. According to warning demand, in map, draw warning line andFace, forms warning region set, as Fig. 4. System provides warning line and face that interface is set, can setpoint color, line style, fillingType.

The 3rd step, under GIS, warning rule setting concrete steps are:

(1), according to warning demand, taking warning region set as basis, specify and allow direction, speed, time parameter. For oftenThe warning analytical method of seeing, concrete setting means is as follows:

Intrusion detection: specify the figure in warning region set; Direction initialization in GIS (unidirectional, two-way).

The detection of hovering: specify the region of hovering in warning region set; Setting is hovered the time.

Abandon detects: specify the surveyed area in warning region set; Specify the abandon time of staying.

Other warning rules are set, and the rest may be inferred.

Wherein, if user specifies one to allow direction, in the time of warning, need to set the direction that forms broken line and polygonal each selections,Detailed process is as Fig. 5. The concrete steps of each line segment direction assignment are as follows:

First, broken line and the judgement of polygonal drafting direction. Mainly form vector according to two initial line segments herein, pitch by vectorTake advantage of and can judge broken line and polygonal drafting direction.

Then, judge that user draws the position relationship of direction and broken line. Set user and draw vector for V₁, crossing line segment institute with itCorresponding vector is V₂, judge V₁With V₂Relation, makes V₂To V₁The direction rotation 90 of vector is spent, and forms vectorial V₃，V₃BeV₂The permission vector of vector. According to said method, obtain the permission vector that each line segment vector is corresponding.

(2), according to the homography matrix resolving in the first step, warning rule is mapped in the monitor video of each video camera, as figure6；

The 4th step, guard against in real time concrete steps and be:

(1) in each monitor video, carry out foreground detection, fresh target detection, target following operation, the position that obtains dynamic objectPut and direction. Foreground detection has adopted mixed Gauss model background subtraction, and tracking adopts Meanshift algorithm.

(2) according to the warning rule of the kinematic parameter of dynamic object and the mapping of the 3rd step, guard against in real time.

(3) dynamic object is presented in generalized information system in real time, when there being target to violate warning rule, generalized information system highlighting andAuditory tone cues security protection personnel, simultaneously the time to event, place, break the rules and carry out record.

Embodiment:

The first step: relevant device is prepared. Prepare a monitor video server, some video cameras, high accuracy map and remote sensing image.

Second step: monitor video and 2D geographical spatial data shine upon mutually.

(1) in GIS software, open high accuracy map and remote sensing image, these data have clear and definite geographical space reference, buildView adopts WGS84UTM projected coordinate system. The figure layer of newly-built vertex type is used for managing each video camera, and positional information comprises shootingX under projected coordinate system and the Y of machine, attribute field comprises height, inclination angle, azimuth and the horizontal field of view angle of video camera.

(2) according to camera height, inclination angle, azimuth and horizontal field of view angle, calculate the corresponding guarded region of ground level, asFig. 1. Specific formula for calculation, as follows:

In Fig. 1, the coordinate of c point in local coordinate system calculates, as formula 1, (x_c,y_c) be the seat in local coordinate system corresponding to c pointMark, H is camera height, Pitch is inclination angle.

$\left\{\begin{array}{c}{x}_c=H*\tan (\mathrm{Pitch}*\frac{\mathrm{pi}}{180})\\ y_c=0\end{array}\right.---\left(1\right)$

The coordinate of c point in projected coordinate system, as formula 2, (X_c,Y_c) be the coordinate of c point under projected coordinate system, Azimuth isAzimuth.

$\left\{\begin{array}{c}{X}_c=X_s+x_c*\cos (\mathrm{Azimuth}*\frac{\mathrm{pi}}{180})\\ Y_c=Y_s+x_c*\sin (\mathrm{Azimuth}*\frac{\mathrm{pi}}{180})\end{array}\right.---\left(2\right)$

Trapezoidal other each points, according to the method that is similar to c and orders, calculate respectively local coordinate, and then calculate geographical coordinate, therebyGet the geographical coordinate of those trapezoidal and middle points, amount to 9 points.

(3) calculate homography matrix. Trapezoid area is corresponding with whole image-region, therefore just can calculate list according to respective pointAnswer matrix. In application, choosing of trapezoidal point can be selected flexibly according to the distribution of ground level in image. Homograph can be based onOpenCV open source software is realized.

The 3rd step: under GIS, warning region is drawn and the setting of warning rule.

Draw warning line and face, specify and allow direction, speed, time parameter, and be mapped in each monitor video. MultimodeFormula guards against regular setting and computer expression can be realized according to described in Fig. 5, and core is vectorial related operation. Warning rule is on groundMapping in reason spatial data and monitor video, the homography matrix mainly calculating based on second step.

The 4th step: warning in real time.

(1) dynamic object is carried out to real-time tracking, and based on it mapping target in map, obtain its actual kinematic parameter.The algorithm of following the tracks of can call OpenCV correlation function, and the mapping of target and track thereof realizes by homography matrix, the obtaining of actual parameterGet mainly the mapping target based in map and time acquisition of information. Dynamic object is expressed in GIS, newly-built dynamic layer, in real timeReceive relevant information.

(2) warning: according to the kinematic parameter of dynamic object, comprise position, speed, track, in real-time judge itself and rule baseThe dependency rule of defined, comprises position, direction, time.

Claims (2)

1. a multi-mode warning analytical method, the steps include:

The first step, obtain monitor video in region, large-scale map and high definition remote sensing image, according to mapping matrix, realize monitoringThe mutual mapping of video and 2D geographical spatial data;

Under second step, GIS, warning region is set: according to warning demand, draw alert based on map and remote sensing image in GIS platformGuard against line and face, form warning region set;

Under the 3rd step, GIS, warning rule is set: the warning region set based on drawing in second step, set flexibly the direction of motion,Speed and time parameter, thus build multi-mode warning rule base; Wherein, if user specifies one to allow direction, in the time of warning, needSet the direction that forms broken line and polygonal each selections, according to allowing direction and warning region to draw all line segments of direction calculatingDirection, be specially: set user and draw vector for V₁, the corresponding vector of crossing line segment is V with it₂, judge V₁With V₂Relation, makes V₂To V₁The direction rotation 90 of vector is spent, and forms vectorial V₃，V₃Be V₂The permission vector of vector, according to said method,Obtain the permission vector that each line segment vector is corresponding; After warning rule sets, based on the mapping matrix of the first step, by these warning ruleBe mapped in each relevant monitor video;

The 4th step, warning in real time: obtain under the position, direction of dynamic object and GIS mapping by foreground detection, target followingActual motion parameter; Whether real-time judge dynamic object motion state violates multi-mode warning rule base; Violate warning rule when havingWhen event occurs, in generalized information system, carry out Realtime Alerts with figure or the alarm audio frequency given prominence to.

2. a kind of multi-mode warning analytical method according to claim 1, is characterized in that, realizes prison in the described first stepThe mutual mapping of control video and 2D geographical spatial data, its concrete steps are:

(a) obtain height, inclination angle, horizontal view angle and the azimuth of each video camera, based on these 4 each monitoring of calculation of parameterThe ground monitoring region of probe, i.e. trapezoidal guarded region corresponding to each monitoring probe;

(b) according to the position of video camera, by the Coordinate Conversion of each trapezoidal guarded region to 2D map of living in or remote sensing imageCoordinate, and form guarded region figure layer;

(c) according to trapezoidal geographical coordinate and the corresponding image coordinate of trapezoidal each point, resolve homography matrix, look thereby realize monitoringFrequently the mutual mapping of image and 2D geographical spatial data.