CN101521805A - Helmet-site detecting system on construction field and detecting method thereof - Google Patents

Abstract

The invention discloses a helmet-site detecting system on a construction field and a detecting method thereof. The system comprises a camera and a computer, wherein an image outputted by the camera is an RGB format, the frame per second is larger than 15 fps, and the resolution is 640*480; and the camera is connected with the computer through a LAN/Ethernet port. The detecting method comprises the following steps: (1) processing and segmenting the image with the RGB format to form a human body image; (2) defining a G component of the RGB image of each pixel as G(xi, yi), and summating an X component of the component G(xi, yi) to obtain a one-dimensional component g (yi); (3) calculating the proportional relation of a human body head in an imaging system according to the angle of an imaging camera lens and a human body and determining that the starting point coordinates of the human body head in the Y direction of the one-dimensional component g (yi) is (yS, yE) ; (4) summating g (yi) from yS to yE to obtain g sum; (5) comparing g sum with the G component of the black RGB image; (6) judging that a helmet is not worn if the error is smaller than a set value Epsilon; and (7) judging that the helmet is worn if the error is larger than the set value Epsilon.

Description

On-the-spot detection system of construction site safety cap and method

Technical field

The present invention relates to a kind of safety cap detection system and method, relate in particular to a kind of intelligent video safety cap detection system and method, belong to technical field of video monitoring.

Background technology

At place workmen's safe wearing caps such as industrial and mining enterprises, construction sites is the necessary security safeguard measure, but the situation that breach of security regulation does not wear a safety helmet happens occasionally, and brings potential safety hazard.Therefore in the construction site situation of wearing of safety cap is monitored very necessaryly, adopting video camera that the situation of wearing of construction site safety cap is monitored is method preferably.If manually video image is monitored the very big defective of existence: consume great amount of manpower, but be subject to the various defectives of manpower, take time and effort and efficient is low but rely on.There are some researches show, when a people observes two monitors, can miss 45% useful information in 10 minutes; Can miss 95% useful information after 22 minutes; Attentiveness is disperseed more when the people observes more a plurality of display simultaneously.A large amount of camera pictures does not have enough manpower monitoring, has influence on the validity of monitoring.Therefore we press for a kind of height intellectuality, can automatically select and extract the feature that the meaning represented is arranged, can guarantee higher video analytic system accuracy, with practical value and software, apply it to the construction site safety cap and wear in the monitoring of situation, to guarantee construction safety.

Summary of the invention

The object of the present invention is to provide a kind of height intellectuality, can automatically select and extract the feature that the meaning represented is arranged, can guarantee higher intelligent video safety cap detection system accuracy, with practical value and method, be applied to the monitoring that construction site safety cap such as construction site is worn situation, to guarantee construction safety.

Purpose of the present invention is achieved by the following technical programs:

The on-the-spot detection system of a kind of construction site safety cap, comprise video camera and computer, the capture direction of described video camera be monitored human body and become angle, video camera output image form is a rgb format, the frame per second of image is greater than 15fps, image resolution ratio is 640 * 480, and video camera is connected by the LAN/Ethernet port with computer.

A kind of construction site safety cap in-situ check and test method, this method utilization intelligent video colorful safety cap detection method is carried out analysis monitoring to video image, comprises the following step:

(1) the rgb format image is carried out motion detection, moving Object Segmentation, single matching treatment, human body image is split, the coordinate that the human body image square is four jiaos is (X ₁, Y ₁), (X ₂, Y ₂), (X ₃, Y ₃), (X ₄, Y ₄);

(2) the G component that defines the RGB image of each pixel is G (x _i, y _i), with two dimensional component G (x _i, y _i) the X component press following formula summation:

$g\left(y_i\right)=\underset{i=i_0}{\overset{i=i_m}{\mathrm{\Σ}}}G(x_i,y_i)/(y_{\mathrm{im}}-y_{i0})$ , obtain one dimension component g (y _i);

(3) calculate the proportionate relationship of human body head according to the goniometer of imaging lens and human body, determine that human body head is at one dimension component g (y in imaging system _i) in the starting point coordinate of Y direction be (y _S, y _E);

(4) with g (y _i) from y _STo y _ESummation obtains $g_{\mathrm{sum}}=\underset{y_S}{\overset{y_E}{\mathrm{\Σ}}}y\left(y_i\right)/(y_E-y_S);$

(5) with g _SumWith the G component of the RGB image of black relatively;

(6) if error less than set point ε, judges that then this human body is not for wearing a safety helmet;

(7) if error greater than set point ε, judges that then this human body wears a safety helmet.

Construction site of the present invention safety cap in-situ check and test method, can also realize that this method comprises the following step by another technical scheme:

(1) the rgb format image is carried out motion detection, moving Object Segmentation, single matching treatment, human body image is split, the coordinate that the human body image square is four jiaos is (X ₁, Y ₁), (X ₂, Y ₂), (X ₃, Y ₃), (X ₄, Y ₄);

(2), obtain the picture size of safety cap in this position according to the position coordinates of human body in entire image;

(3) the head image square with human body splits, and the coordinate that the head image square is four jiaos is (X ₁₀, Y ₁₀), (X ₂₀, Y ₂₀), (X ₃, Y ₃), (X ₄, Y ₄);

(4) define the edge line of human body head;

(5) the pixel pixel value within the edge line of human body head is set at 255;

(6) the human body head figure that step (5) is obtained and the figure of safety cap carry out relatedness computation according to following formula:

$\mathrm{cp}(i_0,j_0)=\underset{i}{\mathrm{\Σ}}\underset{j}{\mathrm{\Σ}}p(i+i_0,j+j_0)\·o(i,j)$

x ₁₀≤i ₀≤x ₂₀

y ₁₀≤j ₀≤y ₃

(7) if the degree of correlation greater than the value of defining q, judges that then this human body wears a safety helmet;

(8) if the degree of correlation less than the value of defining q, judges that then this human body does not wear a safety helmet.Requiring on-the-spot employed safety cap is (as colors such as red, green, blues) of non-black,

Compared with prior art, beneficial effect of the present invention is: highly intelligent, and can automatically select and extract and represent the feature of meaning to analyze, be applied to the construction site safety cap and wear in the monitoring of situation, can find the personnel that do not wear a safety helmet real-time and accurately, to guarantee construction safety.

Description of drawings

Fig. 1 is the flow chart of construction site safety cap in-situ check and test method.

Fig. 2 is the flow chart of second kind of construction site safety cap in-situ check and test method.

Fig. 3 is the human figure's image pattern that is partitioned into.

Fig. 4 is human body head image segmentation figure.

Fig. 5 is a human body head image border line chart.

Fig. 6 is that pixel value is set at 255 head figure in the edge line.

Fig. 7 is the figure of safety cap.

Embodiment

The invention will be further described below in conjunction with drawings and Examples.

Construction site of the present invention safety cap detection system, comprise video camera and computer, the capture direction of described video camera be monitored human body certain angle arranged, video camera output image form is a rgb format, the frame per second of image is greater than 15fps, image resolution ratio is 640 * 480, and video camera is connected by the LAN/Ethernet port with computer.

It is (as colors such as red, green, blues) of non-black that the present invention requires on-the-spot employed safety cap, and utilization intelligent video colorful safety cap detection method is carried out analysis monitoring to video image,

Embodiment one:

First kind of construction site safety cap detection method of the present invention as shown in Figure 1, this method comprises the following step:

(1) the rgb format image is carried out motion detection, moving Object Segmentation, single matching treatment, human body image is split, the coordinate that the human body image square is four jiaos is (X ₁, Y ₁), (X ₂, Y ₂), (X ₃, Y ₃), (X ₄, Y ₄), as shown in Figure 3;

(2) the G component that defines the RGB image of each pixel is G (x _i, y _i), with two dimensional component G (x _i, y _i) the X component press following formula summation:

$g\left(y_i\right)=\underset{i=i_0}{\overset{i=i_m}{\mathrm{\Σ}}}G(x_i,y_i)/(y_{\mathrm{im}}-y_{i0})$ , obtain one dimension component g (y _i);

(3) calculate the proportionate relationship of human body head according to the goniometer of imaging lens and human body, determine that human body head is at one dimension component g (y in imaging system _i) in the starting point coordinate of Y direction be (y _S, y _E);

(4) with g (y _i) from y _STo y _ESummation obtains $g_{\mathrm{sum}}=\underset{y_S}{\overset{y_E}{\mathrm{\Σ}}}y\left(y_i\right)/(y_E-y_S);$

(5) with g _SumWith the G component of the RGB image of black relatively;

(6) if error less than set point ε, judges that then this human body is not for wearing a safety helmet;

(7) if error greater than set point ε, judges that then this human body wears a safety helmet.

Embodiment two:

Second kind of construction site safety cap detection method of the present invention: as shown in Figure 2, this method comprises the following step:

(1) the rgb format image is carried out motion detection, moving Object Segmentation, single matching treatment, human body image is split, the coordinate that the human body image square is four jiaos is (X ₁, Y ₁), (X ₂, Y ₂), (X ₃, Y ₃), (X ₄, Y ₄), as shown in Figure 3;

(2), obtain the picture size of safety cap, as shown in Figure 7 in this position according to the position coordinates of human body in entire image;

(3) the head image square with human body splits, and the coordinate that the head image square is four jiaos is (X ₁₀, Y ₁₀), (X ₂₀, Y ₂₀), (X ₃, Y ₃), (X ₄, Y ₄), as shown in Figure 4;

(4) take out the head image square, define the edge line of human body head, as shown in Figure 5;

(5) the pixel pixel value within the edge line of human body head is set at 255, image becomes black, as shown in Figure 6;

(6) the human body head figure that step (5) is obtained and the figure of safety cap carry out relatedness computation according to following formula:

$\mathrm{cp}(i_0,j_0)=\underset{i}{\mathrm{\Σ}}\underset{j}{\mathrm{\Σ}}p(i+i_0,j+j_0)\·o(i,j)$

x ₁₀≤i ₀≤x ₂₀

y ₁₀≤j ₀≤y ₃

(7) if the degree of correlation greater than the value of defining q, judges that then this human body wears a safety helmet;

(8) if the degree of correlation less than the value of defining q, judges that then this human body does not wear a safety helmet.

In addition to the implementation, the present invention can also have other execution modes, and all employings are equal to the technical scheme of replacement or equivalent transformation formation, all drop in the protection range of requirement of the present invention.

Claims (3)

1. on-the-spot detection system of construction site safety cap, comprise video camera and computer, it is characterized in that, the capture direction of described video camera be monitored human body and become angle, video camera output image form is a rgb format, the frame per second of image is greater than 15fps, and image resolution ratio is 640 * 480, and video camera is connected by the LAN/Ethernet port with computer.

2. construction site safety cap in-situ check and test method, it is non-black that this method requires safety cap, this method comprises the following step:

(1) the rgb format image is carried out motion detection, moving Object Segmentation, single matching treatment, human body image is split, the coordinate that the human body image square is four jiaos is (X ₁, Y ₁), (X ₂, Y ₂), (X ₃, Y ₃), (X ₄, Y ₄);

(2) the G component that defines the RGB image of each pixel is G (x _i, y _i), with two dimensional component G (x _i, y _i) the X component press following formula summation:

$g\left(y_i\right)=\underset{i=i_0}{\overset{i=i_m}{\mathrm{\Σ}}}G(x_i,y_i)/(y_{\mathrm{im}}-y_{i0}),$ Obtain one dimension component g (y _i);

(3) calculate the proportionate relationship of human body head according to the goniometer of imaging lens and human body, determine that human body head is at one dimension component g (y in imaging system _i) in the starting point coordinate of Y direction be (y _S, y _E);

(4) with g (y _i) from y _STo y _ESummation obtains $g_{\mathrm{sum}}=\underset{y_S}{\overset{y_E}{\mathrm{\Σ}}}y\left(y_i\right)/(y_E-y_S);$

(5) with g _SumWith the G component of the RGB image of black relatively;

(6) if error less than set point ε, judges that then this human body is not for wearing a safety helmet;

(7) if error greater than set point ε, judges that then this human body wears a safety helmet.

3, a kind of construction site safety cap in-situ check and test method, it is non-black that this method requires safety cap, this method comprises the following step:

(1) the rgb format image is carried out motion detection, moving Object Segmentation, single matching treatment, human body image is split, the coordinate that the human body image square is four jiaos is (X ₁, Y ₁), (X ₂, Y ₂), (X ₃, Y ₃), (X ₄, Y ₄);

(2), obtain the picture size of safety cap in this position according to the position coordinates of human body in entire image;

(3) the head image square with human body splits, and the coordinate that the head image square is four jiaos is (X ₁₀, Y ₁₀), (X ₂₀, Y ₂₀), (X ₃, Y ₃), (X ₄, Y ₄);

(4) define the edge line of human body head;

(5) the pixel pixel value within the edge line of human body head is set at 255;

(6) the human body head figure that step (5) is obtained and the figure of safety cap carry out relatedness computation according to following formula:

$\mathrm{cp}(i_0,j_0)=\underset{i}{\mathrm{\Σ}}\underset{j}{\mathrm{\Σ}}p(i+i_0,j+j_0)\·o(i,j)$

x ₁₀≤i ₀≤x ₂₀

y ₁₀≤j ₀≤y ₃

(7) if the degree of correlation greater than the value of defining q, judges that then this human body wears a safety helmet;

(8) if the degree of correlation less than the value of defining q, judges that then this human body does not wear a safety helmet.

Images