CN105513043A - Method for counting workers underground coal mine based on vision - Google Patents

Abstract

The invention discloses a kind of coal mine underground operators method of counting of view-based access control model, one piece of elliptic region characterization in personnel targets image in image, picture material is expressed with stochastic model, elliptical appearance is then a random process in image, describes the random process i.e. with Gibbs density Searching makes the maximum oval configuration of Gibbs density, i.e. solution optimization problem Statistics In oval number, as total number of persons's mesh. This method can count coal mine operation number automatically, cooperate on one's own initiative without personnel, realize noiseless real time monitoring.

Description

The coal mine underground operators method of counting of view-based access control model

Technical field

The present invention relates to coal mine underground operators method of counting, particularly relate to a kind of coal mine underground operators method of counting of view-based access control model.

Background technology

In order to stop colliery spy, major accident occurs, and colliery forbids overdetermination person to produce, and monitoring coal mine operation number is one of effective measures preventing overdetermination person from producing in real time.At present, coal mine operation number carries tag card to count by personnel in the pit, and identification card system None-identified many people one card phenomenon, occurs skip often, thus can not accurate counting coal mine underground operators.

Need a kind of coal mine underground operators method of counting solving or at least improve one or more problems intrinsic in prior art.

Summary of the invention

The object of the present invention is to provide a kind of coal mine underground operators method of counting of view-based access control model, the method to coal mine operation number Auto-counting, can be cooperated with on one's own initiative without the need to personnel, realizes noiseless monitoring.

According to a kind of embodiment form, a kind of coal mine underground operators method of counting of view-based access control model is provided, it is characterized in that: one piece of elliptic region in the personnel targets image in image characterizes, elliptic space χ=[0, the X in definition image _m] × [0, Y _m] × [a _m, a _m] × [b _m, b _m] × [0, π], wherein, X _mand Y _mrepresent width and the height of image respectively, (a _m, a _m) and (b _m, b _m) representing minimum value and the maximal value of transverse and minor axis respectively, θ ∈ [0, π] represents oval direction; Express picture material with probabilistic model, appearance oval in image is then a stochastic process, describes described stochastic process namely by Gibbs density wherein, U (o) represents the potential energy of oval stochastic process, o={o ₁=(x ₁, m ₁) ..., o _n=(x _n, m _n) ∈ χ represent oval stochastic process one configuration, for normaliztion constant; Find the ellipse configuration making Gibbs density maximum, namely separate optimization problem statistics in oval number, be total number of persons's order.

Accompanying drawing explanation

By following explanation, accompanying drawing embodiment becomes aobvious and sees, its only with at least one described by reference to the accompanying drawings preferably but the way of example of non-limiting example provide.

Fig. 1 is the method for the invention process flow diagram.

Fig. 2 is that the method for the invention personnel targets characterizes schematic diagram.

Fig. 3 is the method for the invention personnel targets candidate region and background schematic diagram.

Embodiment

Fig. 1 is the process flow diagram of the inventive method, and contrast Fig. 1 flow process, is described.

The necessary safe wearing cap of coal mine underground operators, safety helmet is subcircular, and is generally yellow, obvious with gray background contrast.When underground coal mine installs seizure operation area scene camera, because underground coal mine is tunnel, camera optical axis usually and ground have and be greater than 0 angle being less than 90, therefore, the safety helmet that operating personnel wears is in the picture then near oval, based on above 2 points, one piece of elliptic region in the personnel targets image in image characterizes.Personnel in the pit scatters operation within the specific limits, and the Distance geometry visual angle of they and video camera is different, therefore safety helmet position in the picture and form also different, for this reason, in the picture definition oval belonging to space χ=K × M=[0, X _m] × [0, Y _m] × [a _m, a _m] × [b _m, b _m] × [0, π], K represents oval locational space, and M represents oval attribute space, X _mand Y _mrepresent width and the height of image respectively, (a _m, a _m) and (b _m, b _m) representing minimum value and the maximal value of transverse and minor axis respectively, θ ∈ [0, π] represents oval direction, and model of ellipse is as shown in Figure 2.Express picture material with probabilistic model, appearance oval in image is then a stochastic process, describes this stochastic process namely by Gibbs density u (o) represents the potential energy of oval stochastic process, o={o ₁=(x ₁, m ₁) ..., o _n=(x _n, m _n) ∈ χ represent oval stochastic process one configuration, for normaliztion constant; Find the ellipse configuration making Gibbs density maximum, namely separate optimization problem statistics in oval number, be total number of persons's order.

Potential-energy function U (o)=U _p(o)+U _d(o), U _po () represents priori energy, U _do () represents data capacity.

Consider that borehole operation district personnel are blocked sometimes mutually, then show as ellipse in the picture and have overlap, U _po () punishes according to each ellipse and contiguous oval Maximum overlap, definition a (o _i, o _j) ∈ [0,1] represents overlap coefficient, c is normal number, and μ () is ellipse area, μ (o _i∩ o _j) represent oval o _iwith o _joverlapping area, γ _prepresent the weight to overlapping relation punishment, its size changes along with the size of overlapping area between ellipse, and overlapping area is large, then give larger weight, corresponding priori energy is also larger.

Data item U _do () to have expressed in image the degree of confidence that candidate's elliptic region is personnel, safety helmet is yellow oval in the picture, and background is grey, characterizes by both color characteristic difference sizes the degree of confidence that candidate's elliptic region is personnel, namely

o _irepresent candidate's elliptic region, F (o _i) represent the neighborhood of candidate elliptic region, as shown in Figure 3.D _b(o _i, F (o _i)) represent Pasteur's distance of candidate's elliptic region color distribution and its neighborhood color distribution:

$d_B(o_i,F\left(o_i\right))=-\ln \left(\underset{j=1}{\overset{L}{\Σ}}\sqrt{p_j\left(o_i\right)\·q_j\left(o_i\right)}\right)$

Wherein, p _j() and q _j() represents the probability distribution that candidate's elliptic region is corresponding with each bin of its neighborhood color histogram respectively, and L represents the number of color histogram bin, is normalized to [-1,1] scope,

$o_d\left(d_B\right)=\left\{\begin{array}{cc}\left(1-\frac{d_B}{d_0}\right),& d_B<d_0\\ \exp \left(-\frac{d_B-d_0}{D}\right)-1,& d_B>d_0\end{array}\right.$

Wherein, d ₀represent the average of Pasteur's distance of oval personnel targets region and its neighborhood color distribution, D is a scale parameter, o _d(d _b) ∈ [-1,1] have evaluated the similarity of candidate's elliptic region and its neighborhood color distribution, for less Pasteur's distance, o _d(d _b) be on the occasion of, represent that candidate elliptic region is that the degree of confidence of personnel is large, negative value represents that candidate's elliptic region is that the degree of confidence of personnel is low.

The simulated annealing (SA) based on birth and death process is utilized to solve optimization problem, algorithm comprises the following steps:

E1. initialization temperature parameter time discretization step-length δ=δ ₀;

E2. raw step: for each pixel s ∈ I, if do not have target, sentencing probability δ B (s) at s increases a target, wherein, $\&ForAll;s\&Element;I,B\left(s\right)=\frac{zb\left(s\right)}{{\&Sigma;}_{t\&Element;I}b\left(t\right)},b\left(s\right)=1+9\frac{{\max }_{t\&Element;I}{U}_d\left(t\right)-U_d\left(s\right)}{{\max }_{t\&Element;I}{U}_d\left(t\right)-{\min }_{t\&Element;I}{U}_d\left(t\right)},$ Z is for giving some parameter;

E3. ordered steps: calculate configuration target o _idata item u _d(o _i), the order sequence of successively decreasing according to data capacity;

The step of E4. going out: to each target o according to this order _icalculate mortality ratio wherein a _β(o _i)=exp (-β U (o _i)), then target is with probability d (o _i) be destroyed;

E5. convergence test: if all targets are destroyed in step e 4 just added by step e 2, then realize convergence, algorithm terminates, otherwise, reduce temperature T (n+1)=kT (n) and discretize step-length δ (n+1)=δ (n)-Δ δ, then get back to step e 2, wherein n represents cooling number of times, k < 1 is constant, and Δ δ is constant.

Claims (1)

1. a coal mine underground operators method of counting for view-based access control model, is characterized in that: one piece of elliptic region in the personnel targets image in image characterizes, elliptic space χ=[0, the X in definition image _m] × [0, Y _m] × [a _m, a _m] × [b _m, b _m] × [0, π], wherein, X _mand Y _mrepresent width and the height of image respectively, (a _m, a _m) and (b _m, b _m) representing minimum value and the maximal value of transverse and minor axis respectively, θ ∈ [0, π] represents oval direction; Express picture material with probabilistic model, appearance oval in image is then a stochastic process, describes described stochastic process namely by Gibbs density wherein, U (o) represents the potential energy of oval stochastic process, o={o ₁=(x ₁, m ₁) ..., o _n=(x _n, m _n) ∈ χ represent oval stochastic process one configuration, for normaliztion constant; Find the ellipse configuration making Gibbs density maximum, namely separate optimization problem statistics in oval number, be total number of persons's order.