CN201063764Y - Intelligentized object detail capture device in video monitoring system - Google Patents

Abstract

The utility model relates to an intelligent target detail capturing device of video monitoring system. The device adopts one or a plurality of fixed cameras to monitor the overall situation of a monitored region, and adopts a tripod head camera to capture target details in the monitored region. The utility model firstly extracts a moving target in the region from the images captured by the fixed cameras through a processing module, and then tracks the segmented moving target and judges motion states; if a target enters a set region, the utility model can adjust the direction and focal length of the tripod head camera according to the preset azimuth parameters and focal length parameters of the region, so as to capture the detailed images of the target in the region; meanwhile, the utility model can store and remotely transmit the images according to needs. The utility model has the advantages that: the utility model is a video monitoring system which gives consideration to large scene monitoring and can automatically realize the detail capturing of concrete targets, solving the contradiction between monitoring range and monitoring target detail in the prior art.

Description

Intelligent target detail capturing device in a kind of video monitoring system

Technical field

The utility model belongs to technical field of video monitoring, relates to the intelligent target detail capturing device in a kind of video monitoring system.

Background technology

At present, in technical field of video monitoring, most surveillance videos can only write down the motion state of target in the monitoring scene, and target detail information clearly but can not be provided.Like this, make the use value of video record reduce greatly.The method that improves the target detail definition at present mainly contains two classes: class methods are to adopt high-quality imaging device to improve imaging resolution, be the cost that has increased video monitoring system perhaps by a plurality of video cameras are set in scene the image information of target detail, the shortcoming of these class methods being provided; Another kind of method is to adopt the video monitoring system that has monopod video camera, but these class methods all are to carry out work by the mode of manual setting at present, and shortcoming is to realize intellectuality, and target detail IMAQ efficient is lower.

Therefore, the contradiction between video monitoring range in the existing supervisory control system of needs solution and objectives details are gathered provides a kind of contradiction that can effectively solve between the two, does not increase the video monitoring system of cost again.

Summary of the invention

The purpose of this utility model provides the intelligent target detail capturing device in a kind of video monitoring system, can either carry out monitoring video to large scene, can catch the target detail information that enters setting regions intelligent, efficiently again, solve the contradiction between large scene monitoring and the target detail information capture, improved the use value of monitoring video.

In order to achieve the above object, the technical solution of the utility model is as follows:

Intelligent target detail capturing device in a kind of video monitoring system, be used to catch the panoramic picture of guarded region and the target detail image of setting regions, this device is made up of image-forming module 1, input module 2, processing module 3, control module 4, memory module 5 and transport module 6; Image-forming module 1 is used for the imaging of panorama and target detail, and be output into the picture after panorama and the video image of target detail; Input module 2 is connected between image-forming module 1 and the processing module 3, is used to catch the video image of image-forming module 1 output; Processing module 3 is connected between input module 2 and the control module 4, is used to handle the full-view video image of being caught by input module 2, and detects whether the target appearance is arranged in the setting regions; Control module 4 is connected between processing module 3 and the image-forming module 1, is used to control the image that image-forming module 1 is caught target detail; Memory module 5 and transport module 6 are connected to input module 2, and memory module 5 is used to store all video captured images, and transport module 6 is used for transmitting video image to remote terminal.

Described image-forming module 1 is made up of one or more fixed cameras 7 and a The Cloud Terrace camera 8, fixed camera 7 is connected with input module 2, be used for the imaging of panorama, The Cloud Terrace camera 8 is connected with control module 4 with input module 2, is used for the imaging of target detail.

Intelligent target detail capturing device in a kind of video monitoring system of the utility model, its operation principle is as follows: the panoramic imagery of 7 pairs of guarded regions of fixed camera; Input module 2 is caught the full-view video image of guarded region, and image is exported to processing module 3; 6 pairs of images of catching of memory module 5 and transport module are stored and are transmitted; Processing module 3 utilizes moving average method to set up background model, and the setting regions in the guarded region is carried out zone number, and record The Cloud Terrace camera 8 is caught the location parameter and the focal length parameter in this reference numeral zone; Processing module 3 utilizes background subtraction to extract moving target; Processing module 3 is followed the tracks of and is detected moving target, and judges whether this moving target enters setting regions; If this moving target does not enter setting regions, then continue to follow the tracks of this moving target; If this moving target enters setting regions, then export the numbering and the Control Parameter of this setting regions and give control module 4; Control module 4 is according to the numbering and the Control Parameter of the setting regions that receives, control The Cloud Terrace camera 8 turn to that processing module 3 write down in advance to position that should the zone, and control the focusing parameter of its adjustment self; The Cloud Terrace camera 8 is caught the image of target detail; 6 pairs of images of catching of memory module 5 and transport module are stored and are transmitted.

The beneficial effects of the utility model are: utilize fixed camera to catch the full-view video image of guarded region, utilize the intellectuality of control module control The Cloud Terrace camera, catch the target detail video image of setting regions efficiently, contradiction between the target detail that has solved the overall view monitoring of guarded region and setting regions is caught has improved the use value of monitoring video.

Description of drawings

Fig. 1 is the structural representation of the intelligent target detail capturing device in a kind of video monitoring system of the utility model.

Among the figure: 1, image-forming module, 2, input module, 3, processing module, 4, control module, 5, memory module, 6, transport module, 7, fixed camera, 8, the The Cloud Terrace camera.

Embodiment

Below in conjunction with accompanying drawing the utility model is done description in further detail:

As shown in Figure 1, intelligent target detail capturing device in a kind of video monitoring system of the present utility model is made up of image-forming module 1, input module 2, processing module 3, control module 4, memory module 5 and transport module 6, and image-forming module 1 is made up of one or more fixed cameras 7 and a The Cloud Terrace camera 8.Be responsible for the imaging of panorama by one or more fixed cameras 7, The Cloud Terrace camera 8 is responsible for the imaging of target detail, video image by 2 pairs of inputs of input module is caught, handle the image that each fixing camera obtains by processing module 3, whether detect has target to occur in each setting regions, carry out catching of target detail image by control module 4 control The Cloud Terrace cameras 8, the storage and the transmission of memory module 5 and transport module 6 all images acquired of control.

Intelligent target detail capturing device in a kind of video monitoring system of the present utility model, the function of its each module is as follows:

One, image-forming module 1 is by the imaging of one or more fixed cameras 7 responsible panoramas, and camera direction, the focal length of The Cloud Terrace camera 8 can be adjusted, and it is responsible for the imaging of target detail.

Two, input module 2 is made up of video frequency collection card and capture program, is responsible for the input of image and the conversion of picture format.Initialization function by call driver is provided with contents such as the resolution of images acquired, picture format, sampling gaps; Call and gather function and finish collection all images.

Three, processing module 3 is achieved as follows function:

1, by mouse the setting regions in each camera collection scene is set and carries out regional numbering, simultaneously, note location parameter and focal length parameter that The Cloud Terrace camera 8 is caught the corresponding region.

Set up background model by moving average method, upgrade because scene content changes or illumination condition changes the change of background of bringing.

Obtain moving region in each camera collection image by the method for utilizing present frame subtracting background model.

Processing method is as follows:

Extract moving target by background subtraction, its treatment step is as follows:

1) extraction of moving region: adopt background subtraction to extract the moving region:

$V(x,y,t)=\left\{\begin{array}{cc}1& |I(x,y,t)-\mathrm{\μ}(x,y,t)|>T\\ 0& \mathrm{else}\end{array}\right.$

Wherein, (x, y t) are the region of variation bianry image to V, and (x, y t) are t input picture constantly to I, and (x, y are t background model constantly t) to μ, and T is a threshold value.

2) background model update algorithm: adopt improved moving average method to carry out context update, processing method is as follows:

μ _t＝Mμ _t-1+(1-M)(αI _t+(1-α)μ _t-1)

Wherein, M is the moving region template, μ _tAnd μ _T-1Be the t moment and t-1 background model constantly, α is a turnover rate.

To the moving object detection line trace of going forward side by side, target appears in finding setting regions, and numbering that then should the zone is exported to control module.

Video tracking is handled, and moving target is followed the tracks of and the judgement of motion state, and treatment step is as follows:

A) eigenvalue calculation, for step 1 1) moving target that extracts after the operation, calculate its characteristic value, comprise barycenter, follow the tracks of window.

Select barycenter and tracking window size to come tracking target as characteristic value.

At first be the tracking window of setting moving target, the boundary rectangle of just using target is as following the tracks of window.

L＝x _max-x _min

W＝y _max-y _min

Wherein, x _Max, x _MinBe respectively the maximum coordinates and the min coordinates of target level direction, y _Max, y _MinBe respectively the maximum coordinates and the min coordinates of target vertical direction.

Each follow the tracks of window mark good after, respectively the target in this window is asked its barycenter, establish input picture and be f (x y), is shown below:

(x y) for following the tracks of movement destination image in the window, can calculate the barycenter of window to f, and (then the center-of-mass coordinate of window is for x, y) ∈ S to establish target pixel

$\stackrel{\‾}{x}=\frac{\underset{S}{\mathrm{\Σ}}\mathrm{xf}(x,y)}{\underset{S}{\mathrm{\Σ}}f(x,y)},$ $\stackrel{\‾}{y}=\frac{\underset{S}{\mathrm{\Σ}}\mathrm{yf}(x,y)}{\underset{S}{\mathrm{\Σ}}f(x,y)}$

B) utilize Kalman filter to set up the motion model of system, the definition status vector is predicted the position that moving target may occur in the next frame.

It is that system sets up the estimation model that this treatment step adopts Kalman filtering.Utilize Kalman filtering to carry out estimation, can reduce noise jamming, dwindle the hunting zone of feature extraction, only need to detect current tracking window, reduced amount of calculation.

If k+1 state vector S constantly in the model _K+1, by k vectorial S constantly _kTransfer function and noise form.And observation vector is by k+1 vectorial S constantly _K+1Observation function and noise decision.

State equation is as follows

s _k+1＝As _k+w _k

Measurement equation

z _k+1＝Cs _k+1+v _k+1

In the formula, w _k, v _K+1For average is zero normal white noise.

s _kBe that state vector is made of an octuple vector:

$s_k=\left[\begin{array}{c}{x}_k\\ y_k\\ {\stackrel{\·}{x}}_k\\ {\stackrel{\·}{y}}_k\\ L_{\mathrm{xk}}\\ L_{\mathrm{yk}}\\ {\stackrel{\·}{L}}_{\mathrm{xk}}\\ {\stackrel{\·}{L}}_{\mathrm{yk}}\end{array}\right]$

In the formula, x _k, y _kBe respectively the target center-of-mass coordinate,

Be respectively center-of-mass coordinate at x, the unit displacement on the y direction, L _Xk, L _YkBe respectively and follow the tracks of window at x, the width on the y direction,

Be respectively and follow the tracks of window width at x, the unit displacement on the y direction.

z _K+1Be observation vector, constitute by four-dimensional vector.

$z_{k+1}=\left[\begin{array}{c}{x}_{k+1}\\ y_{k+1}\\ L_{\mathrm{xk}+1}\\ L_{\mathrm{yk}+1}\end{array}\right]$

Because the sampling interval is very short, therefore, can be similar to and think that the movement velocity of moving target is constant, and the size variation of tracking window is little, then state-transition matrix A is:

$A=\begin{array}{cccccccc}1& 0& t& 0& 0& 0& 0& 0\\ 0& 1& 0& t& 0& 0& 0& 0\\ 0& 0& 1& 0& 0& 0& 0& 0\\ 0& 0& 0& 1& 0& 0& 0& 0\\ 0& 0& 0& 0& 1& 0& t& 0\\ 0& 0& 0& 0& 0& 1& 0& t\\ 0& 0& 0& 0& 0& 0& 1& 0\\ 0& 0& 0& 0& 0& 0& 0& 1\end{array}$

Observing matrix C is:

$C=\begin{array}{cccccccc}1& 0& 0& 0& 0& 0& 0& 0\\ 0& 1& 0& 0& 0& 0& 0& 0\\ 0& 0& 1& 0& 0& 0& 0& 0\\ 0& 0& 0& 1& 0& 0& 0& 0\end{array}$

C) characteristic matching, the similar function of objective definition utilizes the variation of relative interframe target, utilizes eigenvalue calculation similar function value, judges whether to be same tracking target.

At first define the centroid distance function of j target of the barycenter of i target of k frame and k+1 frame:

$D(i,j)=\frac{\left|c_k^i{c}_{k+1}^j\right|}{\max \left|c_k^i{c}_{k+1}^j\right|}$

In the formula: $\left|c_k^i{c}_{k+1}^j\right|=\sqrt{{(x_k^i-x_{k+1}^j)}^2+{(y_k^i-y_{k+1}^j)}^2}$ The area discrepancy function, promptly compare with the window area of j target of the window area of i target of k frame and k+1 frame:

$A(i,j)=\frac{|a_k^i-a_{k+1}^j|}{\max |a_k^i-a_{k+1}^j|}$

In the formula: $|a_k^i-a_{k+1}^j|=|L_{\mathrm{xk}}^i\×L_{\mathrm{yk}}^i-L_{\mathrm{xk}+1}^j\×L_{\mathrm{yk}+1}^j|$

The definition similarity function

Δ(i，j)＝γD(i，j)+ξA(i，j)

In the formula, γ, ξ is weights, and satisfies γ＞ξ, γ+ξ=1, Δ (i, j)≤1.

If D (i, j) more little, illustrate that target is approaching more, and also A (i, j) more little, illustrate that target shape is close more, and also Δ (i, j) more little, the possibility maximum that these two targets are similar is described.Set the threshold value T of similar function _ΔAs the foundation that is same target.

D) model modification upgrades motion model, as the input of the Kalman filtering of next motion model.

When searching out the similar function minimum value, found the follow-up of same target, that is to say that j target of k+1 frame can be thought i target of k frame, promptly both are same targets.At this moment,,, so analogize, finish the tracking of model as the input of motion model estimation next frame with j clarification of objective value of k+1 frame.

2) after detecting moving target and entering setting regions, then the numbering and the Control Parameter of setting regions are exported.

Four, the treatment step of control module 4 is as follows:

According to the zone number that image processing module produces, search corresponding location parameter and focal length parameter, the control module of giving the The Cloud Terrace camera by control port outgoing position parameter and focal length parameter.

Its treatment step is as follows:

1, the zone number that produces according to image processing module is searched corresponding location parameter and focal length parameter, the control module of giving the The Cloud Terrace camera by control port outgoing position parameter and focal length parameter.

2, by PORT COM, as RS-232 or USB, the output control command, control The Cloud Terrace camera turns to desired location and the adjustment of camera focusing parameter.

Five, memory module 5 is stored in hard disk with the image of each fixing camera 7 and 8 collections of The Cloud Terrace camera after overcompression, and the image of being gathered can adopt MPEG-1, MPEG-4, and H.263 WAVELET waits form to compress storage

Six, transport module 6 can be according to actual needs, adopts Ethernet to be transmitted through the network to remote terminal the image of each camera collection.

Claims (2)

1. the intelligent target detail capturing device in the video monitoring system, it is characterized in that, this device is by image-forming module (1), input module (2), processing module (3), control module (4), memory module (5) and transport module (6) are formed, input module (2) is connected between image-forming module (1) and the processing module (3), processing module (3) is connected between input module (2) and the control module (4), control module (4) is connected between processing module (3) and the image-forming module (1), and memory module (5) is connected with input module (2) respectively with transport module (6).

2. the intelligent target detail capturing device in a kind of video monitoring system as claimed in claim 1, it is characterized in that, described image-forming module (1) is made up of one or more fixed cameras (7) and a The Cloud Terrace camera (8), fixed camera (7) is connected with input module (2), and The Cloud Terrace camera (8) is connected with control module (4) with input module (2).

Images