CN106846368A - A kind of oil field real-time video Intelligent Measurement and tracking and device - Google Patents

Abstract

The present invention relates to a kind of oil field real-time video Intelligent Measurement and tracking and device.The method includes：Obtain video flowing；Target detection is carried out to video flowing；When there is target in object detection results, real-time tracking is carried out to target.Be combined for target detection, real-time tracking, oil field business by the present invention, moving target is extracted quickly, accurately, while algorithm is simple, it is easy to implement, the trend of target can be understood with real-time capture, the movement position of positioning target, ensure In Oil Field Exploration And Development production safety.

Description

Oil field real-time video intelligent detection and tracking method and device

Technical Field

The invention relates to a video intelligent detection and tracking method, in particular to an oil field real-time video intelligent detection and tracking method and device.

Background

Facilities such as well sites and station depots in oil fields are sometimes threatened by some unstable factors, such as the intrusion of pedestrian or animal targets. In order to ensure the safety of exploration, development and production of oil fields, intelligent analysis on returned real-time videos is necessary. And if the intrusion of the unknown target is found, detecting in time so as to facilitate subsequent processing.

The existing video detection methods can provide a real-time, visual and real picture for reflecting the monitored object to various places required by people, and the video monitoring systems integrate multiple functions of prevention, monitoring, control evidence obtaining, management and the like and can be used for immediate processing or after-the-fact analysis.

However, the existing video detection and tracking method is not directed to oilfield services. Detection and tracking methods for vehicles, for example: and detecting the vehicles in the video, numbering the marked vehicles, recording vehicle information, and switching to a tracking mode if a driver requests to track the vehicles. But this method is mainly directed to the detection and tracking of vehicles. For another example, for a detection and tracking method for water conservancy control, such as a "water conservancy flood prevention monitoring and early warning method and system based on video monitoring" disclosed in CN103325216A, the method connects an intelligent high-speed dome camera to a video behavior analysis server, a user logs in to monitoring and early warning platform software, a storage unit of the monitoring and early warning platform is provided with a storage path and a storage mode for early warning images and information, and the video behavior analysis server successively calls associated intelligent high-speed dome cameras, which relates to the combination of video monitoring and river flood prevention pollution discharge detection. However, the method is only used for water conservancy prevention and control, and no attempt and exploration are made in the field of oil fields at present.

For an oil field, real-time monitoring is mainly performed on oil wells, oil pipelines and station depots distributed in the field, and any moving body can damage the facilities, so that the detection of a moving target and the subsequent movement of the moving target are more concerned in the emergency monitoring of the oil field.

Disclosure of Invention

Technical problem to be solved

In order to solve the defects of the prior art, the invention provides the method and the device for the real-time video intelligent detection and tracking of the oil field, which combine target detection, real-time tracking and oil field services, extract the moving target quickly and accurately, have simple algorithm and convenient implementation, can capture and position the moving position of the target in real time, know the moving direction of the target and ensure the safety of the exploration, development and production of the oil field.

(II) technical scheme

In order to achieve the purpose, the invention adopts the main technical scheme that:

an oil field real-time video intelligent detection and tracking method comprises the following steps:

101, acquiring a video stream;

102, carrying out target detection on the video stream;

and 103, when the target exists in the target detection result, tracking the target in real time.

Optionally, step 102 specifically includes:

102-1, for any frame in the video stream, acquiring a previous frame image of the any frame and a next frame image of the any frame;

102-2, calculating the difference D (n, n-1) between the any frame and the previous frame and the difference D (n +1, n) between the next frame and the any frame;

102-3, extracting a binary image D (n) of the moving object according to the relation among the D (n, n-1), the D (n +1, n) and a preset threshold A.

Optionally, step 102-2 specifically includes:

D(n,n-1)＝|I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of an arbitrary frame image.

Optionally, step 102-3 specifically includes:

optionally, step 103 specifically includes:

103-1, establishing a template according to the size of the arbitrary frame image S;

103-2, traversing each pixel point (i, j) from the upper left corner of the S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by the (i, j).

Optionally, step 103-1 specifically includes: the size of the template is 0.39 · S.

Optionally, step 103-2 specifically includes:

wherein T (i, j) is the brightness value of the template at (i, j), and S (x + i, y + j) is the brightness value of S at (x + i, y + j).

Optionally, after step 103 is executed, the method further includes:

and if no moving target is found after the target is tracked in real time, re-executing the step 101 and the subsequent steps.

In addition, the invention adopts the main technical scheme that:

an oil field real-time video intelligent detection and tracking device, the device includes:

the acquisition module is used for acquiring a video stream;

the detection module is used for carrying out target detection on the video stream acquired by the acquisition module;

the tracking module is used for tracking the target in real time when the target detection result of the detection module has the target;

the detection module is used for acquiring a previous frame image of any frame and a next frame image of any frame for any frame in the video stream; calculating a difference D (n, n-1) between the any one frame and the previous frame and a difference D (n +1, n) between the next frame and the any one frame; extracting a binary image D (n) of the moving target according to the relation among the D (n, n-1), the D (n +1, n) and a preset threshold A;

the detection module is used for calculating D (n, n-1) and D (n +1, n) according to the following formula,

D(n,n-1)＝|I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of any one frame image;

the detection module is used for calculating D (n) according to the following formula,

the tracking module is used for establishing a template according to the size of the arbitrary frame image S; traversing each pixel point (i, j) from the upper left corner of the S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by the (i, j); the size of the template is 0.39 · S;

the tracking module is used for calculating NC according to the following formula,

wherein T (i, j) is the brightness value of the template at (i, j), and S (x + i, y + j) is the brightness value of S at (x + i, y + j).

Optionally, the apparatus further comprises:

and the control module is used for controlling the acquisition module, the detection module and the tracking module to execute again when no moving target is found after the target is tracked in real time.

(III) advantageous effects

The invention has the beneficial effects that: the method comprises the steps of obtaining a video stream, carrying out target detection on the video stream, tracking a target in real time when the target exists in a target detection result, combining target detection, real-time tracking and oilfield services, extracting a moving target quickly and accurately, simultaneously having simple algorithm and convenient implementation, capturing and positioning the moving position of the target in real time, knowing the movement of the target and guaranteeing the safety of exploration, development and production of the oilfield.

Drawings

FIG. 1 is a flow chart of an oilfield real-time video intelligent detection and tracking method according to an embodiment of the present invention;

FIG. 2 is a flow chart of another method for intelligent detection and tracking of real-time video in an oil field according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a normalized correlation-based region matching tracking algorithm according to an embodiment of the present invention;

FIG. 4 is a flow chart of another method for intelligent detection and tracking of real-time video in an oil field according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of an oilfield real-time video intelligent detection and tracking device according to an embodiment of the present invention.

Detailed Description

For the purpose of better explaining the present invention and to facilitate understanding, the present invention will be described in detail by way of specific embodiments with reference to the accompanying drawings.

The existing video detection and tracking method is not specific to oilfield services. In order to solve the problem, according to the characteristics of oil field exploration, development and production, the invention provides an oil field real-time video intelligent detection and tracking method and device, which can carry out target detection on a video stream by acquiring the video stream, track a target in real time when the target detection result has the target, combine target detection, real-time tracking and oil field services, extract a moving target quickly and accurately, have simple algorithm and convenient implementation, can capture and position the moving position of the target in real time, know the moving direction of the target and ensure the safety of oil field exploration, development and production.

Referring to fig. 1, the method for intelligently detecting and tracking an oilfield real-time video provided by the embodiment includes:

101, a video stream is acquired.

And 102, carrying out target detection on the video stream.

Optionally, step 102 specifically includes:

102-1, for any frame in the video stream, acquiring a previous frame image of the any frame and a next frame image of the any frame.

102-2, calculating the difference D (n, n-1) between the any one frame and the previous frame and the difference D (n +1, n) between the next frame and the any one frame.

102-3, extracting the binary image D (n) of the moving object according to the relation between D (n, n-1), D (n +1, n) and the preset threshold A.

Optionally, step 102-2 specifically includes:

D(n,n-1)＝|I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of an arbitrary frame image.

Optionally, step 102-3 specifically includes:

and 103, when the target exists in the target detection result, tracking the target in real time.

Optionally, step 103 specifically includes:

103-1, establishing a template according to the size of any frame of image S.

103-2, traversing each pixel point (i, j) from the upper left corner of S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by (i, j).

Optionally, step 103-1 specifically includes: the size of the template was 0.39 · S.

Optionally, step 103-2 specifically includes:

where T (i, j) is the luminance value of the template at (i, j) and S (x + i, y + j) is the luminance value of S at (x + i, y + j).

Optionally, after step 103 is executed, the method further includes:

and if no moving target is found after the target is tracked in real time, re-executing the step 101 and the subsequent steps.

The beneficial effect of this embodiment is: the video stream is acquired, the video stream is subjected to target detection, when a target exists in a target detection result, the target is tracked in real time, the target detection, the real-time tracking and the oilfield service are combined, the moving target is extracted quickly and accurately, meanwhile, the algorithm is simple, the implementation is convenient, the moving position of the target can be captured and positioned in real time, the movement direction of the target is known, and the safety of oilfield exploration, development and production is guaranteed.

The following describes the oilfield real-time video intelligent detection and tracking method provided by the present invention again with reference to the flow shown in fig. 2.

201, a video stream is acquired.

Specifically, the video stream is read and then stored in the memory.

202, target detection is performed on the video stream.

Specifically, a three-frame difference algorithm is adopted for target detection.

The basic idea of the frame difference algorithm is to perform difference processing on two or more adjacent frames of images in a video sequence, and then compare the difference with a preset threshold value A to obtain a pixel point of a moving part.

In practice, this can be achieved by the following steps.

202-1, for any frame in the video stream, acquiring a previous frame image of any frame and a next frame image of any frame.

202-2, calculating the difference D (n, n-1) between any frame and the previous frame and the difference D (n +1, n) between the next frame and any frame.

Wherein D (n, n-1) ═ I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

Wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of an arbitrary frame image.

202-3, extracting the binary image D (n) of the moving object according to the relation between D (n, n-1), D (n +1, n) and the preset threshold A.

Wherein,

the three-frame difference algorithm is suitable for the situation that the background does not change in a short time, and when the object motion is found, the object motion is detected.

And 203, when the target exists in the target detection result, tracking the target in real time.

Specifically, a region matching tracking algorithm based on normalized correlation is used for real-time tracking.

The normalized correlation-based region matching tracking algorithm performs matching tracking by calculating normalized correlation coefficients of the template and the region covered by the template (see fig. 3).

In particular, the method can be realized by the following steps.

203-1, establishing a template according to the size of any frame of image S.

The created template is generally a rectangle with a small aspect ratio S.

Alternatively, the size of the template is 0.39 · S.

203-2, traversing each pixel point (i, j) from the upper left corner of S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by (i, j).

Wherein,

t (i, j) is the luminance value of the template at (i, j), and S (x + i, y + j) is the luminance value of S at (x + i, y + j).

And when the template is matched and calculated with the pixels in the S area one by one, obtaining a matrix related to NC. The value of the NC matrix element is between 0 and 1, the larger the value is, the better the matching effect is, the 0 indicates that the template and the covered image area have no correlation, namely the matching effect is the worst, and the 1 indicates that the template and the covered image area have the highest correlation, namely the matching effect is the best. Meanwhile, the influence caused by the sudden change of illumination can be reduced through the normalization correlation processing.

204, if no moving target is found after the target is tracked in real time, re-executing step 201 and the subsequent steps.

If no moving object is found, go to step 201 and continue the real-time video stream reading.

The method described in the steps 201 to 204 performs real-time video intelligent monitoring of the oil field by adopting a mode of combining a three-frame difference algorithm and a normalized correlation-based area matching tracking algorithm, and for the oil field, the method mainly performs real-time monitoring on oil wells, oil pipelines and station libraries distributed in the field, and any moving body can damage the facilities, so that the emergency monitoring of the oil field can pay more attention to the detection of a moving target and the subsequent movement direction thereof, various functional modules are integrated together, the three-frame difference algorithm is combined with the oil field service, the moving target is extracted quickly and accurately, and meanwhile, the algorithm is simple and is convenient to implement. The moving position of the target can be captured and positioned in real time through the normalized related region matching tracking, the moving direction of the target is known, and the safety of the exploration, development and production of the oil field is guaranteed.

Fig. 4 shows a practical application flow of the oilfield real-time video intelligent detection and tracking method provided by the embodiment. After the real-time video stream is accessed, the video stream is read through the video image reading module, the read video stream is subjected to target detection through the moving target detection module, if a moving target is found, the video stream is tracked through the real-time tracking module, and if the moving target is not found, the target detection is carried out through the moving target detection module again until the monitoring is finished.

The beneficial effect of this embodiment is: the video stream is acquired, the video stream is subjected to target detection, when a target exists in a target detection result, the target is tracked in real time, the target detection, the real-time tracking and the oilfield service are combined, the moving target is extracted quickly and accurately, meanwhile, the algorithm is simple, the implementation is convenient, the moving position of the target can be captured and positioned in real time, and the moving direction of the target is known, so that the number of times of field manual inspection of important facilities in the oilfield can be reduced, the continuous operation can be realized, the illegal intrusion target can be found and recorded in real time, compared with the manual inspection, the human resource investment cost of safety guarantee work can be reduced, and the safety guarantee coefficient of the important facilities in the oilfield can be improved.

Based on the same conception, the invention also provides an oil field real-time video intelligent detection and tracking device, and the problem solving principle of the device is similar to that of the oil field real-time video intelligent detection and tracking method, so that the implementation of the device can refer to the implementation of the oil field real-time video intelligent detection and tracking method, and repeated parts are not repeated.

Referring to fig. 5, the real-time video intelligent detection and tracking device for oil field includes:

an obtaining module 501, configured to obtain a video stream;

a detection module 502, configured to perform target detection on the video stream acquired by the acquisition module 501;

a tracking module 503, configured to track the target in real time when the target detection result of the detection module 502 includes the target;

a detection module 502, configured to obtain, for any frame in a video stream, a previous frame image of the any frame and a next frame image of the any frame; calculating the difference D (n, n-1) between the any one frame and the previous frame and the difference D (n +1, n) between the next frame and the any one frame; extracting a binary image D (n) of the moving target according to the relation between D (n, n-1), D (n +1, n) and a preset threshold A;

a detection module 502 for calculating D (n, n-1) and D (n +1, n) according to the following formula,

D(n,n-1)＝|I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of any one frame image;

a detection module 502 for calculating D (n) according to the formula,

a tracking module 503, configured to establish a template according to the size of any frame of image S; traversing each pixel point (i, j) from the upper left corner of the S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by the (i, j); the size of the template is 0.39 · S;

a tracking module 503 for calculating NC according to the following formula,

where T (i, j) is the luminance value of the template at (i, j) and S (x + i, y + j) is the luminance value of S at (x + i, y + j).

Optionally, the apparatus further comprises:

and the control module is used for controlling the acquisition module, the detection module and the tracking module to execute again when no moving target is found after the target is tracked in real time.

The beneficial effect of this embodiment is: the video stream is acquired, the video stream is subjected to target detection, when a target exists in a target detection result, the target is tracked in real time, the target detection, the real-time tracking and the oilfield service are combined, the moving target is extracted quickly and accurately, meanwhile, the algorithm is simple, the implementation is convenient, the moving position of the target can be captured and positioned in real time, and the moving direction of the target is known, so that the number of times of field manual inspection of important facilities in the oilfield can be reduced, the continuous operation can be realized, the illegal intrusion target can be found and recorded in real time, compared with the manual inspection, the human resource investment cost of safety guarantee work can be reduced, and the safety guarantee coefficient of the important facilities in the oilfield can be improved.

Claims (10)

1. An oil field real-time video intelligent detection and tracking method is characterized by comprising the following steps:

101, acquiring a video stream;

102, carrying out target detection on the video stream;

and 103, when the target exists in the target detection result, tracking the target in real time.

2. The method according to claim 1, wherein step 102 specifically comprises:

102-1, for any frame in the video stream, acquiring a previous frame image of the any frame and a next frame image of the any frame;

102-2, calculating the difference D (n, n-1) between the any frame and the previous frame and the difference D (n +1, n) between the next frame and the any frame;

102-3, extracting a binary image D (n) of the moving object according to the relation among the D (n, n-1), the D (n +1, n) and a preset threshold A.

3. The method according to claim 2, wherein step 102-2 specifically comprises:

D(n,n-1)＝|I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of an arbitrary frame image.

4. The method according to claim 2, wherein step 102-3 specifically comprises:

$D\left(n\right)=\left\{\begin{array}{cc}1& D(n,n-1)>A\∩D(n+1,n)>A\\ 0& D(n,n-1)\≤A\∩D(n+1,n)\≤A\end{array}\right..$

5. the method according to claim 1, wherein step 103 specifically comprises:

103-1, establishing a template according to the size of the arbitrary frame image S;

103-2, traversing each pixel point (i, j) from the upper left corner of the S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by the (i, j).

6. The method according to claim 5, wherein step 103-1 specifically comprises: the size of the template is 0.39 · S.

7. The method according to claim 5, wherein step 103-2 specifically comprises:

$NC=\frac{\underset{i,j}{\Σ}\left(T\right(i,j)\·S(x+i,y+j\left)\right)}{\sqrt{\underset{i,j}{\Σ}T{(i,j)}^2\·\underset{i,j}{\Σ}S{(x+i,y+j)}^2}};$

wherein T (i, j) is the brightness value of the template at (i, j), and S (x + i, y + j) is the brightness value of S at (x + i, y + j).

8. The method of claim 1, wherein after step 103 is performed, further comprising:

and if no moving target is found after the target is tracked in real time, re-executing the step 101 and the subsequent steps.

9. The utility model provides an oil field real-time video intellectual detection system and tracking means which characterized in that, the device includes:

the acquisition module is used for acquiring a video stream;

the detection module is used for carrying out target detection on the video stream acquired by the acquisition module;

the tracking module is used for tracking the target in real time when the target detection result of the detection module has the target;

the detection module is used for acquiring a previous frame image of any frame and a next frame image of any frame for any frame in the video stream; calculating a difference D (n, n-1) between the any one frame and the previous frame and a difference D (n +1, n) between the next frame and the any one frame; extracting a binary image D (n) of the moving target according to the relation among the D (n, n-1), the D (n +1, n) and a preset threshold A;

the detection module is used for calculating D (n, n-1) and D (n +1, n) according to the following formula,

D(n,n-1)＝|I_n(x,y)-I_n-1(x,y)|；

D(n+1,n)＝|I_n+1(x,y)-I_n(x,y)|；

wherein, (x, y) is the coordinate of the pixel point, I_n(x, y) is a pixel value of any one frame image;

the detection module is used for calculating D (n) according to the following formula,

$D\left(n\right)=\left\{\begin{array}{cc}1& D(n,n-1)>A\∩D(n+1,n)>A\\ 0& D(n,n-1)\≤A\∩D(n+1,n)\≤A\end{array}\right.;$

the tracking module is used for establishing a template according to the size of the arbitrary frame image S; traversing each pixel point (i, j) from the upper left corner of the S, and calculating the normalized correlation NC of the template in the image area S (x, y) covered by the (i, j); the size of the template is 0.39 · S;

the tracking module is used for calculating NC according to the following formula,

$NC=\frac{\underset{i,j}{\Σ}\left(T\right(i,j)\·S(x+i,y+j\left)\right)}{\sqrt{\underset{i,j}{\Σ}T{(i,j)}^2\·\underset{i,j}{\Σ}S{(x+i,y+j)}^2}};$

wherein T (i, j) is the brightness value of the template at (i, j), and S (x + i, y + j) is the brightness value of S at (x + i, y + j).

10. The apparatus of claim 9, further comprising:

and the control module is used for controlling the acquisition module, the detection module and the tracking module to execute again when no moving target is found after the target is tracked in real time.