CN106355723A - Method based on images for counting operation personnel in mine - Google Patents

Abstract

The invention discloses a method based on images for counting operation personnel in a mine. The method comprises the steps: observing an operating zone with one or a plurality of cameras, dispersing the operating zone into small regions, constructing a camera observed primitive dictionary D, and solving x to obtain the number of personnel by using the relation among a characterization camera observed value y, a small region duty status value x and the dictionary D (as shown in the specification). The method can automatically count the number of the operation personnel in a coal mine without initiative cooperation of the personnel, thereby realizing non-interference real-time monitoring.

Description

Mine operation personnel method of counting based on image

Technical field

The present invention relates to coal mine underground operators method of counting, more particularly, to a kind of mine operation personnel based on image Method of counting.

Background technology

In order to prevent colliery spy, major accident occurs, and colliery forbids overdetermination person to produce, monitor in real time coal mine operation people Number is to prevent one of effective measures of overdetermination person's production.At present, coal mine operation number is to carry mark by personnel in the pit Card counting, the many people of identification card system None- identified one card phenomenon, and often skip occurs, thus be unable to accurate counting colliery Underground operators.

Need a kind of coal mine operation people solving or at least improving intrinsic one or more problems in prior art Member's method of counting.

Content of the invention

It is an object of the invention to provide a kind of mine operation personnel method of counting based on image, the method can be to coal Under mine, operation area number counts automatically, cooperates with one's own initiative without personnel, realizes noiseless monitoring.

According to a kind of embodiment form, provide a kind of mine operation personnel method of counting based on image, the inventive method Technical scheme comprise the steps:

A. in operation area periphery, one or more camera is installed, each viewing field of camera overwrite job area；

B. operation area is separated into n not overlapping subregion, with one-dimensional vector x ∈ { 0,1 }ⁿExpression n sub-regions duty State, 1 expression this subregion someone, 0 represents that this subregion is unmanned；

C. use ore deposit staff's model of scattering of some typical bodies, build camera looks into fee primitive dictionary d ∈ { 0,1 }^m×n, row in dictionary d_jConstruction method is: operation area has and only one of which model personnel are located at j-th subregion, will observe all cameras of operation area Observe vector cascade, that is, assume that the observation vector of p-th camera isThen Wherein m_pFor the image pixel number of p-th camera,P is the camera number of configuration；

When d. counting, operation area configuration camera obtains observing vector, solve In x, in x the number of " 1 " be operation area number, wherein z ∈ { 0,1 }^mFor noise vector, computing on the right of equation is boolean Algebraic operation.

Brief description

By following explanation, accompanying drawing embodiment becomes aobvious and has seen, it is at least one preferred only be described in conjunction with the accompanying But the way of example of non-limiting example is given.

Fig. 1 is the principle schematic of the inventive method.

Specific embodiment

Fig. 1 is the principle schematic of the inventive method.In operation area periphery, one or more camera is installed, each camera Visual field all overwrite job areas, by operation area be divided into the sizes such as n as 70cm × 70cm subregion it is assumed that every sub-regions A people how can only be accommodated, the duty state of n sub-regions is with vector x ∈ { 0,1 }ⁿRepresent, 1 expression this subregion someone, 0 table Show that this subregion is unmanned.

Build camera looks into fee primitive dictionary d ∈ { 0,1 }^m×n, column vector d in dictionary d_jBuild with the following method: it is assumed that joining Putting camera has p, with height in 170cm about body weight be about the underground operators of average weight as model, in operation When area has and only model personnel are located at j-th subregion, each camera records a width operation area scene image, by each phase The pixel belonging to personnel in the scene image of machine record is labeled as 1, and other is context marker is 0, and the image after labelling is used one Dimensional vector represents referred to as observes vector it is assumed that the observation vector of p-th camera isSight by all cameras Examine vector cascade, that is,Whereinm_pImage pixel for p-th camera Number, t represents transposition.

During counting, configure cameras record operation area scene image, obtain observing vector WhereinFor the observation vector of p-th camera, solveIn x, in x, 1 number is operation The number in area, wherein z ∈ { 0,1 }^mFor noise vector, computing on the right of equation is boolean calculation.

To solveProblem state as:

$\begin{array}{ccc}\stackrel{\&overbar;}{x}=\arg {\min }_{x\&element;{\{0,1\}}^n}\left|setc\left(x\right)\right|& s.t.& \left|setc(y\&circleplus;d\·x)\right|\≤e\end{array}$

Wherein, setc (x) :={ j:x_j=1 } in expression x, element value is 1 index set

| | represent the number seeking contained element in set

E is maximum noise digit, i.e. | setc (z) |≤e

CauseIn imply x in duty state vector x_j=1 element and dictionary row d_jCorresponding relation, therefore The problems referred to above can use following Algorithm for Solvings:

|input paramete: y, d, e, β

1. initialization setγ is empty set,

2.for j=1,2 ..., n

if|setc(d_j) setc (y) |≤e, then

Wherein,.. represent the difference asking two set

3. ask

4. update

$\begin{array}{c}setc\left(\stackrel{\&overbar;}{y}\right)\&doubleleftarrow;setc\left(\stackrel{\&overbar;}{y}\right)\∪setc\left(d_j\right)\\ setc\left(r\right)\&doubleleftarrow;setc\left(r\right)\backslash setc\left(d_j\right)\end{array}$

5.If e is ＞ e, repeat step 3～5

6. exportIt is personnel's number.

Claims (1)

1. a kind of mine operation personnel method of counting based on image is it is characterised in that comprise the following steps:

A. in operation area periphery, one or more camera is installed, each viewing field of camera overwrite job area；

B. operation area is separated into n not overlapping subregion；

C. the miner using scattering of some typical bodies, as personnel's model, builds camera looks into fee primitive dictionary d ∈ { 0,1 }^m×n, row in dictionary d_jConstruction method is: operation area has and only one of which model personnel are located at j-th subregion, and operation area configures camera looks into fee Operation area, cascades the observation vector of each camera, that is, assumes that the observation vector of p-th camera isThenWherein m_pFor the image pixel number of p-th camera,P is configuration Camera number；

When d. counting, operation area configuration camera looks into fee operation area obtains observing vectorSolveIn x, x ∈ { 0,1 }ⁿRepresent n sub-regions duty state, 1 expression this subregion someone, 0 this sub-district of expression Domain is unmanned, and in x, the number of " 1 " is the number of operation area, z ∈ { 0,1 }^mFor noise vector,The computing of middle the right For boolean calculation.