CN105050097A - Optimization deployment method for image sensor network - Google Patents

Abstract

The present invention puts forward an optimization deployment method for an image sensor network, comprising two basic steps, i.e., establishment of an observation model of an image sensor, and optimization deployment of an image sensor network. In particular, the optimization deployment method comprises: a first step of establishing the observation model of the image sensor based on characteristics of the image sensor observing an object; and a second step of establishing an optimization deployment function of the network by establishing an observation area model, a network coverage model, and an observation model of the network, and obtaining the optimal deployment of the network by optimizing the function. According to the optimization deployment method of the present invention, an observation performance function of the image sensor to the object is established, and the more perfect image sensor observation model is introduced into the optimization deployment of the image sensor network, thereby effectively solving the problems of low deployment and operation flexibility and low observation efficiency in the existing image sensor network.

Description

A kind of image sensor network Optimization deployment method

Technical field

The present invention relates to computer vision and image sensor network field, particularly relate to a kind of research of image sensor network Optimization deployment method.

Background technology

Image sensor network is the core content of multimedia sensor network technology.It is except the feature such as distributed perception, self organization ability, the dense deployment of sensor node, limited communication bandwidth with traditional sensors network, also has that image, the information data amount that can obtain scene are huge, an observed direction and observation area is limited, Intelligent treatment and the feature such as storage capacity is stronger.Meanwhile, in order to adapt to more complicated application, image sensor network it is also proposed higher requirement to the collaborative perception between node and associated treatment ability.And the deployment of imageing sensor node determines collaborative perception and the disposal ability of network, be realize the key issue that image sensor network is efficient, low power operation must solve.

In traditional sensors network, usually adopt perceived distance description node to the perception of event, and on this basis the deployment of network analyzed and optimize.But the perception of imageing sensor to scene has certain directivity and regionality, the three-dimensional structure of observation scene also can have an impact to the perception of node camera.Therefore, traditional retrains analysis optimization method based on perceived distance, is no longer applicable to image sensor network.

In the last few years, also there is researcher to carry out the research of image sensor network covering performance for the feature of imageing sensor, proposed virtual visual field, the concepts such as space covering.But in research in the past, substantially with the overall coverage rate of network and coverage for index, consider the deployment issue of network, do not consider that imageing sensor is to the observation quality of target.Thus, need the more effective image sensor network Optimization deployment method of research, the Effec-tive Function for image sensor network provides substantial theoretical foundation.

Summary of the invention

The object of the invention is to propose a kind of image sensor network Optimization deployment method, improve the flexibility of wireless image sensor network disposition and operation, may be used for the fields such as military security, public safety, intelligent transportation, intelligent building, environmental monitoring.

For achieving the above object, the present invention proposes a kind of image sensor network Optimization deployment method, specifically comprises the foundation of imageing sensor observation model and Optimization deployment two basic steps of image sensor network.

Step one, in one embodiment of the invention, the foundation of described imageing sensor observation model comprises further: at certain moment t, image sensor model be one fan-shaped, by a four-tuple represent, with the continuous adjustment of its sensing direction, imageing sensor is had the ability the whole border circular areas covered in its distance sensing; Set up the observation performance function of imageing sensor wherein ω is the subtended angle of target to imageing sensor, E=2.7+0.7 × (ω/α _c), α _cfor the critical view angle of imageing sensor, relevant with its spatial resolving power; Observed object is equivalent to a circle, with the position of impact point X for central coordinate of circle, radius of a circle r is determined by target sizes; Impact point to the central angle on the subtended angle ω corresponding circle of imageing sensor is the observation model of described imageing sensor is:

Step 2, in one embodiment of the invention, the Optimization deployment of described image sensor network comprises further: set up the observation area model that size is M × N, to observation area sliding-model control, take density as gs (choosing of gs is relevant to system requirements precision) grid division, the grid point in described observation area is represented by matrix Ω; Setting up network coverage model is:

$P\left(\mathrm{\Ω}\right)=\frac{\underset{i=1}{\overset{m\×n}{\mathrm{\Σ}}}p\left(X_i\right)}{m\×n},$ Wherein $p\left(X_i\right)=\left\{\begin{array}{cc}1,& X\∈R\\ 0,& X\∉R\end{array}\right.,$ R represents the overlay area of any one imageing sensor, and Ω is observation area, and m × n represents all Grid dimensions in guarded region; The observation model setting up region Ω ' is: $G\left({\mathrm{\Ω}}^{\′}\right)=\frac{\underset{i=1}{\overset{p_1\×q_1}{\mathrm{\Σ}}}G\left(X_i\right)}{p_1\×q_1},$ Wherein, $G\left(X\right)=\underset{i=1}{\overset{T}{\mathrm{\Σ}}}{g}_i\left(X\right)-\mathrm{\Σ}{g}^{\′}\left(X\right),$ T is can all imageing sensor interstitial contents of coverage goal point X, Σ g ' (X) for different images transducer is relative to the measured value of target overlapped fov part, p ₁× q ₁for the number of Ω ' grid point in region; Setting up network coverage deployment majorized function is: wherein, the disposition optimization function that F (Ω) is network, the coverage rate that P (Ω) is network, G (Ω ' _k) be emphasis observation area Ω ' in network _kmeasured value, the number of described emphasis observation area Ω ' is that t, k are larger, and described emphasis observation area Ω ' is more important, λ _kfor weight factor, corresponding region is more important, λ _kvalue is larger, described λ _kvalue meets k=1,2 ... t, wherein, λ ₁=10, when not considering the observed reliability of emphasis observation area in network, λ=0; Utilize the deployment function of binary integer programming method to network to be optimized, obtain the network design when network coverage and key area measured value maximum case.

A kind of image sensor network Optimization deployment method that the present invention proposes, by setting up image sensor network observation model, further established more effective image sensor network Optimization deployment model.Instant invention overcomes conventional images sensor network disposition and operational flexibility low, the problem that observed efficiency is not high, provides strong Research foundation for image sensor network realizes efficient monitoring.

Accompanying drawing explanation

Fig. 1 is the imageing sensor observed reliability method for establishing model flow chart of the embodiment of the present invention;

Fig. 2 is the image sensor model schematic diagram of the embodiment of the present invention;

When Fig. 3 a is the closely observation of the embodiment of the present invention, target is to the subtended angle schematic diagram of imageing sensor;

Fig. 3 b be the embodiment of the present invention compared with the subtended angle schematic diagram of target during remote observation to imageing sensor;

Fig. 4 is the reliability schematic diagram of imageing sensor to target observation of the embodiment of the present invention;

Fig. 5 is that the different target size hypograph of the embodiment of the present invention is to the observed reliability curve of target;

Fig. 6 is the Optimization deployment model of network;

Embodiment

Be described below in detail embodiments of the invention, the example of described embodiment is shown in the drawings, and wherein same or similar label represents same or similar meaning from start to finish.The embodiments described below are exemplary, only for explaining the present invention, and can not be interpreted as limitation of the present invention.

The present invention be directed in image sensor network, dispose and operational flexibility low, the problem that observed efficiency is not high, a kind of image sensor network Optimization deployment method of proposition.

In order to clearer understanding can be had to the present invention, be briefly described at this.The present invention includes two basic steps: step one, the foundation of imageing sensor observation model; Step 2, the Optimization deployment of image sensor network.

Concrete, Figure 1 shows that the flow chart of a kind of image sensor network Optimization deployment method of the embodiment of the present invention, comprise the following steps:

Step S101, sets up image sensor model.

In one embodiment of the invention, setting up image sensor model is one fan-shaped (as shown in Figure 2), an available four-tuple represent this model; The wherein position coordinates of C presentation video transducer, the sensing radius of R presentation video transducer, vector for the sensing direction of imageing sensor, the view angle of α representative image transducer; At certain moment t, the sensitive zones of imageing sensor be one fan-shaped, but with the continuous adjustment of its sensing direction, imageing sensor is had the ability the whole border circular areas covered in its distance sensing; Further, at certain moment t, set up if impact point X is covered by imageing sensor, and if only if meets the following conditions

$\left|\right|\stackrel{\→}{\mathrm{CX}}\left|\right|\≤R---\left(1\right)$

Wherein represent the Euclidean distance of impact point X to this imageing sensor.

Step S102, sets up imageing sensor observation performance function.

In one embodiment of the invention, the observation of imageing sensor to target meets such criterion, imageing sensor C from target X distance more away from, target X is relative to less (as shown in Figure 3, the ω of the subtended angle ω of imageing sensor C ₂< ω ₁), the resolving power of imageing sensor to target is poorer, and thus, according to Johnson criterion, the observation performance function of camera to target meets following formula

$P\left(\mathrm{\&omega;}\right)=\frac{{(\mathrm{\&omega;}/{\mathrm{\&alpha;}}_C)}^E}{1+{(\mathrm{\&omega;}/{\mathrm{\&alpha;}}_C)}^E}---\left(3\right)$

Wherein E=2.7+0.7 × (ω/α _c), α _cfor the critical view angle of imageing sensor, relevant with its spatial resolving power;

Step S103, sets up imageing sensor observation model.

In one embodiment of the invention, carry out equivalent target point X with a circle, namely with described impact point X for the center of circle do circle Φ, radius is r, then imageing sensor C to the observation performance of target X by Φ to the central angle corresponding to the subtended angle ω (now, ω=α) of C characterize, as shown in Figure 4; When the observation scope of imageing sensor C is outside circle Φ, its observation performance is characterized by central angle during tangent situation, as shown in Fig. 3 (a), 3 (b) when target X meets formula (1) and formula (2), the observation performance function of imageing sensor C to target X is

Wherein,

Now, imageing sensor C characterizes the numerical value of the observation performance of target X between [0,0.5], and along with the change of observed object size r, imageing sensor C to the observation performance curve of target X as shown in Figure 5.

Step S104, sets up observation area model.

In one embodiment of the invention, observation area size is M × N, comprising normal observation region and important observation area; To observation area sliding-model control, take density as gs grid division, the grid point in described observation area is represented by matrix Ω

Wherein Ω ' _k, k=1,2 ... t represents important observation area;

Step S105, sets up network coverage model.

In one embodiment of the invention, network coverage model is

$P\left(\mathrm{\&Omega;}\right)=\frac{\underset{i=1}{\overset{m\&times;n}{\mathrm{\&Sigma;}}}p\left(X_i\right)}{m\&times;n}---\left(6\right)$

Wherein

$p\left(X_i\right)=\left\{\begin{array}{cc}1,& X\&Element;R\\ 0,& X\&NotElement;R\end{array}\right.---\left(7\right)$

Wherein, R represents the overlay area of any one imageing sensor, and Ω is observation area, and m × n represents all Grid dimensions in guarded region;

Step S106, sets up network observations model.

Step 6.1 sets up an imageing sensor to the observation model of an impact point.In one embodiment of the invention, an imageing sensor C is defined by formula (4) the observation model of an impact point X

Step 6.2 sets up the observation model of multiple imageing sensor to an impact point.In one embodiment of the invention, multiple imageing sensor C _i, i=1,2 ... T is all imageing sensor C that can observe impact point X to the observation model of an impact point X _imeasured value sum

$G\left(X\right)=\underset{i=1}{\overset{T}{\mathrm{\&Sigma;}}}{g}_i\left(X\right)-\mathrm{\&Sigma;}{g}^{\&prime;}\left(X\right)---\left(8\right)$

Wherein, T is can the number of all imageing sensors of coverage goal point X, and Σ g ' (X) is for different images transducer is relative to the measured value of target X overlapped fov part.

Step 6.3 sets up the observation model of multiple imageing sensor to region a certain in network.In one embodiment of the invention, multiple imageing sensor C _i, i=1,2 ... T, to the observation model of region Ω ' a certain in network is

$G\left({\mathrm{\&Omega;}}^{\&prime;}\right)=\frac{\underset{i=1}{\overset{p_1\&times;q_1}{\mathrm{\&Sigma;}}}G\left(X_i\right)}{p_1\&times;q_1}---\left(9\right)$

Wherein, X _i∈ Ω ', i=1,2 ... p ₁× q ₁, p ₁× q ₁for the sum of the middle grid point of region Ω ';

Step S107, the Optimization deployment of image sensor network.

In one embodiment of the invention, network design majorized function is

$F\left(\mathrm{\&Omega;}\right)=P\left(\mathrm{\&Omega;}\right)+\underset{k=1}{\overset{t}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_k\&CenterDot;G\left({\mathrm{\&Omega;}}_k^{\&prime;}\right)---\left(10\right)$

Wherein, the coverage rate that P (Ω) is network, G (Ω ' _k) be emphasis observation area Ω ' in network _kmeasured value, the number of described emphasis observation area is that t, k are larger, and described emphasis observation area Ω ' is more important, λ _kfor weight factor, corresponding region is more important, λ _kvalue is larger, described λ _kvalue meets

$\frac{{\mathrm{\&lambda;}}_{k+1}}{{\mathrm{\&lambda;}}_k}=10,k=\mathrm{1,2},...t---\left(11\right)$

Wherein, λ ₁=10, when not considering to there is emphasis observation area in network, λ _k=0;

The network design Optimized model corresponding to majorized function (10) as shown in Figure 6, utilizes binary integer programming method to be optimized to above-mentioned Optimized model, obtains the Optimization deployment result of network.

By a kind of image sensor network Optimization deployment method that the present invention proposes, conventional images sensor network disposition can be overcome and operational flexibility low, the problem that observed efficiency is not high, provides strong Research foundation for image sensor network realizes efficient monitoring.

Last it is noted that above embodiment is only in order to illustrate technical scheme of the present invention, be not intended to limit.Those of ordinary skill in the art is to be understood that: it still can be modified to the technical scheme described in previous embodiment, or equivalent replacement is carried out to wherein portion of techniques feature, and these amendments or replacement, do not make the essence of appropriate technical solution depart from the spirit and scope of various embodiments of the present invention technical scheme, scope of the present invention is by claims and equivalency thereof.

Claims (1)

1. an image sensor network Optimization deployment method, is characterized in that, comprises the foundation of imageing sensor observation model and Optimization deployment two steps of image sensor network,

The foundation of described imageing sensor observation model comprises:

(1) set up image sensor model: image sensor model be one fan-shaped, by a four-tuple represent; The wherein position coordinates of C presentation video transducer, the sensing radius of R presentation video transducer, vector for the sensing direction of imageing sensor, the view angle of α representative image transducer; At certain moment t, the sensitive zones of imageing sensor be one fan-shaped, but with the continuous adjustment of its sensing direction, imageing sensor is had the ability the whole border circular areas covered in its distance sensing; Further, at certain moment t, set up if impact point X is covered by imageing sensor, and if only if meets the following conditions:

$\left|\right|\stackrel{\&RightArrow;}{\mathrm{CX}}\left|\right|\&le;R---\left(1\right)$

Wherein represent the Euclidean distance of impact point X to this imageing sensor, representation vector with between angle;

(2) set up imageing sensor observation performance function: along with imageing sensor wide, the subtended angle ω of target to it can reduce gradually, and then the observation performance of imageing sensor can progressively decline, set up observation performance function:

$P\left(\mathrm{\&omega;}\right)=\frac{{(\mathrm{\&omega;}/{\mathrm{\&alpha;}}_C)}^E}{1+{(\mathrm{\&omega;}/{\mathrm{\&alpha;}}_C)}^E}---\left(3\right)$

Wherein E=2.7+0.7 × (ω/α _c), α _cfor the critical view angle of imageing sensor, relevant with its spatial resolving power;

(3) imageing sensor observation model is set up: utilize the observation performance function of imageing sensor can construct its observation model to target

Wherein for target visibility region is to the subtended angle of target's center;

The Optimization deployment of described image sensor network comprises:

(1) observation area model is set up: described observation area size is M × N, comprising normal observation region and important observation area; To observation area sliding-model control, take density as gs (choosing of gs is relevant to system requirements precision) grid division, the grid point in described observation area is represented by matrix Ω:

Wherein Ω ' _k, k=1,2 ... t represents important observation area;

(2) network coverage model is set up: the described network coverage is:

$P\left(\mathrm{\&Omega;}\right)=\frac{\underset{i=1}{\overset{m\&times;n}{\mathrm{\&Sigma;}}}p\left(X_i\right)}{m\&times;n}---\left(6\right)$

Wherein

$p\left(X_i\right)=\left\{\begin{array}{cc}1,& X\&Element;R\\ 0,& X\&NotElement;R\end{array}\right.---\left(7\right)$

Wherein, R represents the overlay area of any one imageing sensor, and Ω is observation area, and m × n represents all Grid dimensions in guarded region;

(3) set up network observations model: described network observations model of setting up comprises: set up an imageing sensor to the observation model of an impact point, set up multiple imageing sensor to the observation model of an impact point and set up the observation model of multiple imageing sensor to region a certain in network;

A described imageing sensor is defined by formula (4) the observation model of an impact point:

Described multiple imageing sensor is the measured value sum of all imageing sensors that can observe impact point to the observation model of an impact point:

$G\left(X\right)=\underset{i=1}{\overset{T}{\mathrm{\&Sigma;}}}{g}_i\left(X\right)-\mathrm{\&Sigma;}{g}^{\&prime;}\left(X\right)---\left(8\right)$

Wherein, T is can the number of all imageing sensors of coverage goal point X, and Σ g ' (X) is for different images transducer is relative to the measured value of target overlapped fov part;

The observation model of described multiple imageing sensor to region Ω ' a certain in network is:

$G\left({\mathrm{\&Omega;}}^{\&prime;}\right)=\frac{\underset{i=1}{\overset{p_1\&times;q_1}{\mathrm{\&Sigma;}}}G\left(X_i\right)}{p_1\&times;q_1}---\left(9\right)$

Wherein, p ₁× q ₁for the number of grid point in region;

(4) network design Optimized model is set up: described network design majorized function is

$F\left(\mathrm{\&Omega;}\right)=P\left(\mathrm{\&Omega;}\right)+\underset{k=1}{\overset{t}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_k\&CenterDot;G\left({\mathrm{\&Omega;}}_k^{\&prime;}\right)---\left(10\right)$

Wherein, the coverage rate that P (Ω) is network, G (Ω ' _k) be emphasis observation area Ω ' in network _kmeasured value, the number of described emphasis observation area is that t, k are larger, and described emphasis observation area Ω ' is more important, λ _kfor weight factor, corresponding region is more important, λ _kvalue is larger, described λ _kvalue meets

$\frac{{\mathrm{\&lambda;}}_{k+1}}{{\mathrm{\&lambda;}}_k}=10,k=\mathrm{1,2},...t---\left(11\right)$

Wherein, λ ₁=10, when not considering the observed reliability of emphasis observation area in network, λ=0;

Utilize binary integer programming method to be optimized to formula (10), obtain the Optimization deployment result of network.