CN115482217A - Electric shock prevention video detection method for transformer substation based on Gaussian mixture model separation algorithm - Google Patents
Electric shock prevention video detection method for transformer substation based on Gaussian mixture model separation algorithm Download PDFInfo
- Publication number
- CN115482217A CN115482217A CN202211146602.9A CN202211146602A CN115482217A CN 115482217 A CN115482217 A CN 115482217A CN 202211146602 A CN202211146602 A CN 202211146602A CN 115482217 A CN115482217 A CN 115482217A
- Authority
- CN
- China
- Prior art keywords
- personnel
- contour
- mixture model
- gaussian mixture
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000000203 mixture Substances 0.000 title claims abstract description 55
- 238000001514 detection method Methods 0.000 title claims abstract description 50
- 238000000926 separation method Methods 0.000 title claims abstract description 36
- 230000002265 prevention Effects 0.000 title claims abstract description 22
- 238000012545 processing Methods 0.000 claims abstract description 12
- 238000001914 filtration Methods 0.000 claims abstract description 5
- 238000009826 distribution Methods 0.000 claims description 110
- 238000010586 diagram Methods 0.000 claims description 43
- 230000008859 change Effects 0.000 claims description 19
- 230000002776 aggregation Effects 0.000 claims description 14
- 238000004220 aggregation Methods 0.000 claims description 14
- 230000003044 adaptive effect Effects 0.000 claims description 12
- 230000003068 static effect Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 206010014357 Electric shock Diseases 0.000 description 18
- 230000009286 beneficial effect Effects 0.000 description 12
- 238000000034 method Methods 0.000 description 9
- 238000010276 construction Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000011897 real-time detection Methods 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 241000764238 Isis Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a transformer substation electric shock prevention video detection method based on a Gaussian mixture model separation algorithm, which comprises the following steps of: collecting a video to be detected, and processing the video to be detected based on a Gaussian mixture model separation algorithm to obtain a target image; carrying out image binarization processing on the target image to obtain a binary image; respectively adopting a kernel of 10x10 to perform open operation on the binary image to remove noise points and perform closed operation to connect adjacent regions to obtain a result image; determining the minimum external positive rectangle of the result image by adopting an object contour detection method; filtering the minimum circumscribed positive rectangle according to a preset critical judgment condition to obtain a detection image; and judging whether the outline of the detected image is intersected with the preset warning area or not to generate a detection result.
Description
Technical Field
The invention relates to the technical field of substation electric shock prevention video detection, in particular to a substation electric shock prevention video detection method based on a Gaussian mixture model separation algorithm.
Background
Electric power is a high-risk field of safety production, and in recent years, each unit of an electric power system invests a large amount of resources to strengthen field safety control. The prior supervision has the problems of time and labor waste, untimely risk early warning, difficult evidence obtaining after the accident, low data utilization and the like. With the continuous development of the internet of things, edge computing, cloud computing and artificial intelligence technologies, the edge computing-based video real-time analysis technology is applied.
In the process of transformer substation interval construction, workers may transport higher and longer objects (such as ladders) in violation of construction, and once the objects are too close to or touch the high-voltage wires, high-voltage air breakdown is easily caused to cause electric shock accidents. According to the design specification of 35KV-110KV power transformation, the safe clear distance between the electrified parts of 35KV different phases is 400mm; the safe clear distance between the charged parts of the 110KV out-of-phase is 1000mm.
The application provides a transformer substation electric shock prevention video detection method based on a Gaussian mixture model separation algorithm, wherein a foreground and a background are quickly separated by the Gaussian mixture model separation algorithm, real-time detection is carried out on a transformer substation interval construction process, the accident occurrence probability is reduced, and electric shock risks are prevented.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a transformer substation electric shock prevention video detection method based on a Gaussian mixture model separation algorithm.
A transformer substation electric shock prevention video detection method based on a Gaussian mixture model separation algorithm comprises the following steps:
collecting a video to be detected, and processing the video to be detected based on a Gaussian mixture model separation algorithm to obtain a target image; carrying out image binarization processing on the target image to obtain a binary image; respectively adopting a kernel of 10x10 to perform open operation on the binary image to remove noise points and perform closed operation to connect adjacent regions to obtain a result image; determining the minimum external right rectangle of the result image by adopting an object contour detection method; filtering the minimum circumscribed positive rectangle according to a preset critical judgment condition to obtain a detection image; and judging whether the outline of the detected image is intersected with the preset warning area or not to generate a detection result.
As an embodiment of the invention, the adaptive Gaussian mixture model is estimated according to an online expectation-maximization algorithm to obtain a Gaussian mixture model separation algorithm.
As an embodiment of the present invention, establishing an adaptive gaussian mixture model includes: respectively acquiring the value of any pixel value in the training image at the time N as X N Probability P (X) N ) And the first B distributions are used as a model of the scene background; according to the probability P (X) N ) Determining an update equation of a Gaussian component by using the first B distributions as a model of a scene background; and updating the Gaussian components of the matching test values participating in the establishment of the self-adaptive Gaussian mixture model according to the updating equation until a preset condition is met to obtain the self-adaptive Gaussian mixture model.
As an embodiment of the present invention, the probability P (X) N ) The calculation formula (2) includes:
wherein, w k Is k th Weight parameter of Gaussian component, η (x) N ;θ k ) Is k th Normal distribution of gaussian components;
k th normal distribution of gaussian component eta (x) N ;θ k ) Expressed as:
wherein, mu k Is the mean vector of the component distribution parameters in the D dimension, Σ k =σ k 2 I is k th Covariance of Gaussian component, D is the dimension of input vector x, x is the independent variable of component normal distribution, x N Is the probability of x at time N, I is the identity matrix, σ k Is the variance value of unit distribution;
at the same time, K distributions are based on the fitness value w k /σ k And (6) sorting.
As an embodiment of the present invention, the values of B are:
where the threshold T is the minimum part of the model that the distribution uses as the scene background, B ∈ B, w j The time scale remaining for the jth color.
As an embodiment of the present invention, updating the equation includes:
wherein, ω is k Is the k-th th A gaussian component, 1/α defining the time constant that determines the variation;
at the same time, if none of the K distributions matches the pixel value to be measured, the least likely component is replaced with a distribution having the current value as its mean, initial high variance, and low weight parameters.
As an embodiment of the invention, the estimation of the adaptive Gaussian mixture model according to the online expectation-maximization algorithm comprises the following steps:
the adaptive gaussian mixture model is initially estimated according to the expected sufficient statistical update equations and then switched to the L-receiver window version when the first L samples are processed.
As an embodiment of the present invention, an image binarization process for a target image includes:
wherein x and y are horizontal and vertical coordinates of the pixel point, g (x and y) is the value of the point, thresh is a threshold value, and maxval refers to a new pixel value given when the pixel value exceeds the threshold value.
As an embodiment of the present invention, a transformer substation electric shock protection video detection method based on a gaussian mixture model separation algorithm further includes:
acquiring the contour of the detection image, and generating a personnel contour distribution map of each detection image;
acquiring a yellow personnel profile distribution map which is less than a preset safety distance from a preset warning area in the personnel profile distribution maps;
obtaining an orange personnel contour distribution diagram of which the aggregation degree of the personnel contours is greater than a preset aggregation degree threshold value in the yellow personnel contour distribution diagram; the aggregation degree of the personnel profiles is determined according to the number of the personnel profiles in the yellow personnel profile distribution diagram and the distance between every two personnel profiles;
judging whether a continuous preset orange person contour distribution map exists or not, if so, determining the continuous preset orange person contour distribution map as a red person contour distribution map, and sending out a site risk alarm;
and determining a danger value corresponding to the red personnel contour distribution diagram according to the variation amplitude of the personnel contour in the continuous preset orange personnel contour distribution diagram, and sending reminding alarm information to the manager of the corresponding level according to the danger value.
As an embodiment of the present invention, the determining the danger value corresponding to the red person profile map according to the variation range of the person profile in the continuous preset orange person profile map comprises:
determining the current state of each person contour according to the variation amplitude of the person contours in the continuous preset orange person contour distribution diagram; wherein the current state comprises a motion state and a static state;
when the change amplitude of the current personnel contour within the preset time is smaller than the preset change amplitude, the current personnel contour is in a static state, otherwise, the current personnel contour is in a motion state;
determining the motion trend of the person contour in the motion state based on the change amplitude;
acquiring the nearest distance between each person contour in the motion state and the nearest other person contour in the motion state;
and calculating the risk-involved rate of the personnel corresponding to the personnel profile closest to the preset warning area in the current red personnel profile distribution map according to the movement trend of all the personnel profiles in the movement state and the closest distance between each personnel profile in the movement state and the closest other personnel profile in the movement state, and taking the risk-involved rate as the risk value of the corresponding red personnel profile distribution map.
The beneficial effects of the invention are as follows:
1. the method has the advantages that a Gaussian mixture model separation algorithm is used, a self-adaptive background modeling method is provided, and illumination change can be well processed and shadows can be well recognized;
2. the detection efficiency is high, and real-time video detection can be realized on low-power-consumption edge computing equipment (xavier nx).
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
fig. 1 is a flowchart of a method for detecting an electric shock prevention video of a transformer substation based on a gaussian mixture model separation algorithm in an embodiment of the present invention;
fig. 2 is a specific flowchart related to an alarm function in a transformer substation electric shock prevention video detection method based on a gaussian mixture model separation algorithm according to an embodiment of the present invention;
fig. 3 is a flow chart for judging whether potential safety hazards exist due to the fact that workers gather in a critical area in the transformer substation anti-electric shock video detection method based on the gaussian mixture model separation algorithm in the embodiment of the invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
Referring to fig. 1, an embodiment of the present invention provides a transformer substation electric shock protection video detection method based on a gaussian mixture model separation algorithm, including:
s101, collecting a video to be detected;
s102, processing a video to be detected based on a Gaussian mixture model separation algorithm to obtain a target image;
s103, carrying out image binarization processing on the target image to obtain a binary image;
s104, respectively adopting a 10x10 kernel to perform open operation on the binary image to remove noise points and perform closed operation to connect adjacent regions to obtain a result image;
s105, determining the minimum circumscribed regular rectangle of the result image by adopting an object contour detection method;
s106, filtering the minimum circumscribed regular rectangle according to a preset critical judgment condition to obtain a detection image;
s107, judging whether the outline of the detected image is intersected with the preset warning area or not, and generating a detection result;
the working principle of the technical scheme is as follows: acquiring a to-be-detected video of a transformer substation, processing the to-be-detected video based on a Gaussian mixture model separation algorithm to obtain a target image, and performing image binarization processing on the target image to obtain a binary image; then, performing open operation on the binary image by using a 10x10 kernel to remove noise generated by video jitter and compression, performing close operation on the binary image by using the 10x10 kernel to connect similar regions to obtain a result image, performing object contour detection on the result image to obtain a minimum external right rectangle of a contour, filtering out a rectangle with a rectangular area smaller than an area threshold value by using the rectangular area, wherein the threshold value is generally set to be within a range of 100-300, and the threshold value is 150; finally setting the outlines of all the filtered objects as recats, the outline of a certain object as rect and the warning region as alarmRect, judging whether the outlines of the detected images are intersected with the preset warning region or not, and generating a detection result; as can be seen from the detection results shown in FIG. 3, the method can well separate the background and the foreground, reduce the noise problems caused by video jitter, illumination change and the like, and can quickly detect the changed part of the video, thereby detecting whether the object intrudes into the alert area;
the beneficial effects of the above technical scheme are: the foreground and the background are quickly separated by using a Gaussian mixture model separation algorithm, real-time detection is carried out on the transformer substation interval construction process, the accident occurrence probability is reduced, and the electric shock risk is prevented.
Referring to fig. 2, in an embodiment, the transformer substation electric shock prevention video detection method based on the gaussian mixture model separation algorithm further includes determining whether to generate alarm information according to a detection result, and continuing to perform subsequent detection on the acquired video after the current detection flow is finished;
the working principle and the beneficial effects of the technical scheme are as follows: on the basis of fig. 1, if there is an intersection between the contour rect of an object and the alert area alarmRect, namely rect ≧ alarmRect >0, an alarm is generated, and the pseudo-code is as follows:
bool alarm=false;
for(varrect inrects)
if(rect∩alarmRect>0)
alarm=true;
the alarm can warn the related personnel in time, and is beneficial to reducing the accident probability.
In one embodiment, the self-adaptive Gaussian mixture model is estimated according to an online expectation-maximization algorithm to obtain a Gaussian mixture model separation algorithm;
the beneficial effects of the above technical scheme are: and the method is beneficial to quickly separating the foreground from the background through a Gaussian mixture model separation algorithm.
In one embodiment, establishing the adaptive gaussian mixture model comprises: respectively acquiring the value of any pixel value in the training image at the time N as X N Probability P (X) N ) And the first B distributions are used as a model of the scene background; according to the probability P (X) N ) Determining an update equation of a Gaussian component by using the first B distributions as a model of a scene background; updating each Gaussian component participating in the establishment of the matching test value of the adaptive Gaussian mixture model according to an updating equation until a preset condition is met to obtain the adaptive Gaussian mixture model;
probability P (X) N ) The calculation formula (2) includes:
wherein, w k Is k th Weight parameter of Gaussian component, η (x) N ;θ k ) Is k th Normal distribution of gaussian components;
k th normal distribution of gaussian component eta (x) N ;θ k ) Expressed as:
wherein, mu k Is the mean vector, Σ, of the component distribution parameters in the D dimension k =σ k 2 I is k th Covariance of Gaussian component, D is the dimension of input vector x, x is the independent variable of component normal distribution, x N Is the probability of x at time N, I is the identity matrix, σ k Variance values distributed as units; further, x is the argument range, x N The representation refers to the probability of a certain value in the range;
k distributions are based on the fitness value w k /σ k Sorting;
the update equation includes:
wherein, ω is k Is the k-th th A gaussian component, 1/α defining the time constant that determines the variation;
meanwhile, if none of the K distributions is matched with the pixel value to be measured, replacing the least probable component with the distribution taking the current value as the mean value, the initial high variance and the low weight parameter of the distribution;
the working principle and the beneficial effects of the technical scheme are as follows: each pixel in the image isIs formed by K Gaussian distribution mixed modeling; a pixel has a value x at time N N The probability of (c) can be written as:wherein w k Is k th A weight parameter of the Gaussian component; eta (x) N ;θ j ) Is k th Normal distribution of components, expressed as Wherein mu k Is the average value of the values of the average, is k th The covariance of the components; k distributions are based on the fitness value w k /σ k Ranking, the first B distributions are used as models of the scene background, where B is estimated as
The threshold T is the minimum part of the background model, B ∈ B, w j The time scale retained for the jth color, w represents the time scale retained for this color, and b refers to the distribution of the first b smaller colors; in other words, it is the minimum prior probability of the background in the scene; background subtraction is performed by labeling a foreground pixel, any pixel having a standard deviation from any B distribution that exceeds 2.5 standard deviations; the Gaussian component of the first matching test value will be updated by the following update equation
Wherein omega k Is the k-th th A Gaussian component; 1/α defines the time constant that determines the change; if none of the K distributions match the pixel value, the least likely component is replaced with a distribution with the current value as its mean, the initial high variance, and the low weight parameter.
In one embodiment, the adaptive Gaussian mixture model is estimated according to an online expectation-maximization algorithm, comprising:
starting to estimate the self-adaptive Gaussian mixture model according to an expected sufficient statistical update equation, and then switching to an L-receiver window version when the previous L samples are processed;
the working principle and the beneficial effects of the technical scheme are as follows: the Gaussian mixture model is estimated by an expected sufficient statistical update equation, and then the model is switched to an L-receiver window version when the previous L samples are processed; the expected fully statistically updated equation provides a good estimate at the beginning before all L samples can be collected; this initial estimation improves the accuracy of the estimation, also improves the performance of the tracker, allowing fast convergence on a stable background model; the L-receiver window update equation takes precedence over the most recent data, so the tracker can adapt to the change of the environment;
the left column shows the online EM algorithm for which sufficient statistics are expected, while the right column shows the L-receiver window version:
the tracker can adapt to the change of the environment, and the performance and the estimation accuracy of the tracker are improved.
The online EM algorithm for which sufficient statistics are expected is shown in the left column, and the L-receiver window version is shown in the right column.
In one embodiment, the image binarization processing is performed on the target image, and comprises the following steps:
wherein x and y are horizontal and vertical coordinates of the pixel point, g (x and y) is the value of the point, thresh is a threshold value, and maxval is a new pixel value given when the pixel value exceeds the threshold value;
in the present embodiment, maxval =255, thresh =128;
the beneficial effects of the above technical scheme are: it is beneficial to highlight the outline of the target image.
Referring to fig. 3, in an embodiment, a transformer substation electric shock prevention video detection method based on a gaussian mixture model separation algorithm further includes:
s201, acquiring the contour of the detected image, and generating a personnel contour distribution map of each detected image;
s202, obtaining a yellow personnel contour distribution map which is less than a preset safety distance from a preset warning area in the personnel contour distribution maps;
s203, obtaining an orange personnel contour distribution map of which the aggregation degree of the personnel contours in the yellow personnel contour distribution map is greater than a preset aggregation degree threshold; the aggregation degree of the personnel outlines is determined according to the number of the personnel outlines in the yellow personnel outline distribution diagram and the distance between every two personnel outlines;
s204, judging whether a continuous preset orange personnel contour distribution diagram exists, if so, determining the continuous preset orange personnel contour distribution diagram as a red personnel contour distribution diagram, and sending out a site risk alarm;
s205, determining a danger value corresponding to the red personnel contour distribution diagram according to the variation amplitude of the personnel contour in the continuous preset orange personnel contour distribution diagram, and sending reminding alarm information to the manager of the corresponding level according to the danger value;
the working principle of the technical scheme is as follows: acquiring the contour of the detection image, and generating a personnel contour distribution map of each detection image; then, a yellow personnel contour distribution map with a preset safety distance less than a preset safety distance from a preset warning area in the personnel contour distribution maps is obtained; wherein, the yellow person contour distribution map is a partial or complete person contour distribution map; obtaining an orange personnel contour distribution diagram of which the aggregation degree of the personnel contours is greater than a preset aggregation degree threshold value in the yellow personnel contour distribution diagram; the aggregation degree of the personnel profiles is determined according to the number of the personnel profiles in the yellow personnel profile distribution diagram and the distance between every two personnel profiles; the more the number of the person outlines in the yellow person outline distribution map is, the higher the aggregation degree is; the closer the distance between every two personnel contours in the yellow personnel contour distribution diagram is, the higher the aggregation degree is; continuously acquiring the outline of the detected image, and reserving a preset personnel outline image each time; for example, 4 person contour diagrams need to be reserved, and 4 person contour diagrams are reserved currently, after the next person contour diagram is completely distinguished, the reserved sequence is added, and the person contour diagram with the earliest time in the original reserved sequence is deleted; judging whether a continuous preset orange personnel contour distribution diagram exists, if so, determining the continuous preset orange personnel contour distribution diagram as a red personnel contour distribution diagram, and sending out a site risk alarm; for example, 4 person contour diagrams need to be reserved, the preset number is 3, and 3 continuous orange person contour distribution diagrams exist in the currently reserved 4 person contour diagrams, and then the continuous 3 person contour diagrams are determined as red person contour distribution diagrams; finally, determining a danger value corresponding to the red personnel contour distribution diagram according to the variation amplitude of the personnel contour in the continuous preset orange personnel contour distribution diagram, and sending reminding alarm information to the managers of the corresponding level according to the danger value;
the beneficial effects of the above technical scheme are: at the transformer substation during operation, the staff is because of the work needs, can have the possibility of gathering official working, when personnel gather, then the condition such as collision fall down appears likely to appear, if gathering personnel is in warning regional edge this moment, it is then possible to make the personnel of falling down fall down in warning region this time to fall down, thereby cause the potential safety hazard, through above-mentioned technical scheme, additionally set up safe fall down the distance (predetermine safe distance promptly), and carry out the analysis to personnel's gathering degree, thereby further reduce accident occurrence probability, because constructor's personnel quality differs, constructor arouses when the alarm that the system sent exists unable to attach attention, consequently, set up the danger value, send the managers that corresponds danger alarm to corresponding grade according to the danger value, the warning effect of protection against electric shock has further been improved.
In one embodiment, determining the danger value corresponding to the red person profile map according to the variation amplitude of the person profile in the continuous orange person profile map comprises:
determining the current state of each person contour according to the variation amplitude of the person contours in the continuous preset orange person contour distribution diagram; wherein the current state comprises a motion state and a static state;
when the change amplitude of the current personnel contour within the preset time is smaller than the preset change amplitude, the current personnel contour is in a static state, otherwise, the current personnel contour is in a motion state;
determining the motion trend of the person contour in the motion state based on the change amplitude;
acquiring the nearest distance between each person contour in the motion state and the nearest other person contour in the motion state;
calculating the risk-involved rate of the personnel corresponding to the personnel profile closest to the preset warning area in the current red personnel profile distribution map according to the movement tendency of all personnel profiles in the movement state and the closest distance between each personnel profile in the movement state and the nearest other personnel profile in the movement state, and taking the risk-involved rate as the risk value of the corresponding red personnel profile distribution map;
the working principle of the technical scheme is as follows: according to the connectionContinuously presetting the variation amplitude of the personnel outlines in the orange personnel outline distribution diagram to determine the current state of each personnel outline; wherein the current state comprises a motion state and a static state; the variation amplitude refers to that any person contour in the current orange person contour distribution diagram determines the person contour of the same person according to the current position and any person contour in the next orange person contour distribution diagram according to the current position, and the variation amplitude is determined according to the action amplitudes of the two person contours; when the change amplitude of the current personnel contour within the preset time is smaller than the preset change amplitude, the current personnel contour is in a static state, otherwise, the current personnel contour is in a motion state; the preset time refers to the acquisition time of the continuous preset orange personnel contour distribution diagram, and the variation amplitude of the current personnel contour within the preset time refers to the variation amplitude of the current personnel contour in the continuous preset orange personnel contour distribution diagram; determining the motion trend of the person contour in the motion state based on the change amplitude; the movement trend is a movement direction, including a direction movement facing to the nearest other person contour and a direction movement opposite to the nearest other person contour; acquiring the nearest distance between each person contour in the motion state and the nearest other person contour in the motion state; the minimum distance between two persons in motion states when moving; finally, calculating the risk-involved rate of the personnel corresponding to the personnel profile closest to the preset warning area in the current red personnel profile distribution map according to the movement trend of all personnel profiles in the movement state and the closest distance between each personnel profile in the movement state and the nearest other personnel profile in the movement state, and taking the risk-involved rate as the risk value of the corresponding red personnel profile distribution map; typically starting from the collision of the outermost personnel contour of the gathered personnel towards the inner personnel contour; the inner-bound personnel are the personnel closest to the preset warning area; the calculation formula of the risk involved rate is preferably as follows:wherein dangerous is the risk-involved rate of the person corresponding to the person profile closest to the preset warning region in the current red person profile distribution map, and m isThe number of the person profiles in the current red person profile map, g is the current state of the person profile, i =1 represents that the person profile is the outermost person profile, i = m-1 represents that the person profile is the person profile closest to the innermost person profile, and Q g,i A weight value corresponding to the current state of the current personnel contour, when g is a motion state, Q g,i =1, when g is in a stationary state, Q g,i =0,d is the tendency of the contour of the person to move, H d,i A preset collision rate corresponding to the movement trend of the current person profile, wherein when the movement trend of the ith person profile faces to the (i + 1) th person profile, H d,i = V, when the motion of the ith person profile tends not to face the (i + 1) th person profile, H d,i =v,V>>v, L is the closest distance between the ith and (i + 1) th person's contours, L l,i The risk weight value of the distance between the ith personal outline and the (i + 1) th personal outline is L l,i The larger;
the beneficial effects of the above technical scheme are: the probability of collision of the personnel is determined through the change amplitude of the personnel outline, so that the risk involved rate of the closest working personnel in the preset warning area is determined, the danger value corresponding to the red personnel outline distribution diagram is obtained, and the warning effect of preventing electric shock is facilitated.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A transformer substation electric shock prevention video detection method based on a Gaussian mixture model separation algorithm is characterized by comprising the following steps: collecting a video to be detected, and processing the video to be detected based on a Gaussian mixture model separation algorithm to obtain a target image; carrying out image binarization processing on the target image to obtain a binary image; respectively adopting a 10x10 kernel to perform open operation on the binary image to remove noise points and perform closed operation to connect adjacent regions to obtain a result image; determining the minimum external positive rectangle of the result image by adopting an object contour detection method; filtering the minimum circumscribed positive rectangle according to a preset critical judgment condition to obtain a detection image; and judging whether the outline of the detected image is intersected with the preset warning area or not to generate a detection result.
2. The transformer substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm as claimed in claim 1, wherein the adaptive Gaussian mixture model is estimated according to an online expectation-maximization algorithm to obtain the Gaussian mixture model separation algorithm.
3. The transformer substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm as claimed in claim 2, wherein the establishing of the adaptive Gaussian mixture model comprises: respectively acquiring the value of any pixel value in the training image at the time N as X N Probability P (X) N ) And the first B distributions are used as a model of the scene background; according to the probability P (X) N ) Determining an update equation of a Gaussian component by using the first B distributions as a model of a scene background; and updating the Gaussian components of the matching test values participating in the establishment of the self-adaptive Gaussian mixture model according to the updating equation until a preset condition is met to obtain the self-adaptive Gaussian mixture model.
4. The substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm according to claim 3, characterized in that the probability P (X) is N ) The calculation formula (2) includes:
wherein w k Is k th Weight parameter of Gaussian component, η (x) N ;θ k ) Is k th Normal distribution of gaussian components;
k th normal distribution of Gaussian components eta (x) N ;θ k ) RepresentComprises the following steps:
wherein, mu k Is the mean vector, Σ, of the component distribution parameters in the D dimension k =σ k 2 I is k th Covariance of Gaussian component, D is the dimension of input vector x, x is the independent variable of component normal distribution, x N Is the probability of x at time N, I is the identity matrix, σ k Is the variance value of unit distribution;
at the same time, K distributions are based on the fitness value w k /σ k And (6) sorting.
5. The substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm according to claim 3, wherein the value of B is as follows:
where the threshold T is the minimum part of the model that the distribution uses as the scene background, B ∈ B, w j The proportion of time remaining for the jth color.
6. The substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm according to claim 3, wherein the updating equation comprises:
wherein, ω is k Is the kth th A gaussian component, 1/α defining the time constant that determines the change;
meanwhile, if none of the K distributions matches the pixel value to be measured, the least likely component is replaced with the distribution having the current value as its mean, initial high variance, and low weight parameters.
7. The substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm as claimed in claim 2, wherein the estimation of the adaptive Gaussian mixture model according to the online expectation-maximization algorithm comprises:
the adaptive gaussian mixture model is initially estimated based on the expected full statistical update equation and then switched to the L-receiver window version when the first L samples are processed.
8. The transformer substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm as claimed in claim 1, wherein image binarization processing is performed on a target image, and comprises the following steps:
wherein x and y are horizontal and vertical coordinates of the pixel point, g (x and y) is the value of the point, thresh is a threshold value, and maxval refers to a new pixel value given when the pixel value exceeds the threshold value.
9. The transformer substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm as claimed in claim 1, further comprising:
acquiring the contour of the detection image, and generating a personnel contour distribution map of each detection image;
acquiring a yellow personnel contour distribution map which is less than a preset safety distance from a preset warning area in the personnel contour distribution map;
obtaining an orange person contour distribution diagram of which the aggregation degree of the person contours in the yellow person contour distribution diagram is greater than a preset aggregation degree threshold; the aggregation degree of the personnel profiles is determined according to the number of the personnel profiles in the yellow personnel profile distribution diagram and the distance between every two personnel profiles;
judging whether a continuous preset orange person contour distribution map exists or not, if so, determining the continuous preset orange person contour distribution map as a red person contour distribution map, and sending out a site risk alarm;
and determining a danger value corresponding to the red personnel contour distribution diagram according to the variation amplitude of the personnel contour in the continuous preset orange personnel contour distribution diagram, and sending reminding alarm information to the managers of the corresponding grade according to the danger value.
10. The substation electric shock prevention video detection method based on the Gaussian mixture model separation algorithm according to claim 9, wherein the step of determining the danger value corresponding to the red personnel contour distribution map according to the variation amplitude of the personnel contour in the continuous preset orange personnel contour distribution map comprises the following steps:
determining the current state of each personnel contour according to the variation amplitude of the personnel contours in the continuous preset orange personnel contour distribution diagram; wherein the current state comprises a motion state and a static state;
when the change amplitude of the current personnel contour within the preset time is smaller than the preset change amplitude, the current personnel contour is in a static state, otherwise, the current personnel contour is in a motion state;
determining the motion trend of the person contour in the motion state based on the change amplitude;
acquiring the nearest distance between each person contour in the motion state and the nearest other person contour in the motion state;
and calculating the risk-involved rate of the personnel corresponding to the personnel profile closest to the preset warning area in the current red personnel profile distribution map according to the movement trend of all the personnel profiles in the movement state and the closest distance between each personnel profile in the movement state and the closest other personnel profile in the movement state, and taking the risk-involved rate as the risk value of the corresponding red personnel profile distribution map.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211146602.9A CN115482217B (en) | 2022-09-21 | 2022-09-21 | Transformer substation electric shock prevention video detection method based on Gaussian mixture model separation algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211146602.9A CN115482217B (en) | 2022-09-21 | 2022-09-21 | Transformer substation electric shock prevention video detection method based on Gaussian mixture model separation algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115482217A true CN115482217A (en) | 2022-12-16 |
CN115482217B CN115482217B (en) | 2024-05-10 |
Family
ID=84423693
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211146602.9A Active CN115482217B (en) | 2022-09-21 | 2022-09-21 | Transformer substation electric shock prevention video detection method based on Gaussian mixture model separation algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115482217B (en) |
Citations (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4916450A (en) * | 1988-05-12 | 1990-04-10 | Radar Control Systems Corporation | Radar system for headway control of a vehicle |
CN102180139A (en) * | 2011-04-22 | 2011-09-14 | 钟延沁 | Proximity trigger safety air bag and safety control method thereof |
CN102826423A (en) * | 2011-06-14 | 2012-12-19 | 株式会社日立制作所 | Elevator safe operation system and safe operation method |
DE102013214305A1 (en) * | 2013-07-22 | 2015-01-22 | Bayerische Motoren Werke Aktiengesellschaft | Method for avoiding a collision of a vehicle with at least one object, driver assistance system and vehicle |
CN105953762A (en) * | 2016-07-07 | 2016-09-21 | 国网山东省电力公司经济技术研究院 | Power transmission line modeling method used for net distance measurement and distance measurement device thereof |
CN108274469A (en) * | 2018-04-23 | 2018-07-13 | 东北大学 | Vacuum mechanical-arm anticollision detecting system and detection method based on multidimensional vision sensor |
WO2018130016A1 (en) * | 2017-01-10 | 2018-07-19 | 哈尔滨工业大学深圳研究生院 | Parking detection method and device based on monitoring video |
JP2018206107A (en) * | 2017-06-06 | 2018-12-27 | 東芝デベロップメントエンジニアリング株式会社 | Approaching object detecting device |
CN111985388A (en) * | 2020-08-18 | 2020-11-24 | 深圳市自行科技有限公司 | Pedestrian attention detection driving assistance system, device and method |
CN113012315A (en) * | 2021-02-23 | 2021-06-22 | 浙江三一装备有限公司 | Safety monitoring system and method for working machine and working machine |
CN113052125A (en) * | 2021-04-09 | 2021-06-29 | 内蒙古科电数据服务有限公司 | Construction site violation image recognition and alarm method |
CN113344874A (en) * | 2021-06-04 | 2021-09-03 | 温州大学 | Pedestrian boundary crossing detection method based on Gaussian mixture modeling |
CN113890193A (en) * | 2021-09-29 | 2022-01-04 | 国网内蒙古东部电力有限公司呼伦贝尔供电公司 | Working state monitoring method and system for electric tracing system of transformer substation circuit breaker |
CN114261389A (en) * | 2021-11-29 | 2022-04-01 | 东风商用车有限公司 | Running track collision risk analysis method of automatic driving commercial vehicle |
CN114708508A (en) * | 2022-04-21 | 2022-07-05 | 广东电网有限责任公司东莞供电局 | Method for identifying moving small animal |
CN114913420A (en) * | 2022-05-17 | 2022-08-16 | 中国矿业大学 | Dynamic obstacle monitoring system and monitoring method for anti-impact drilling robot |
-
2022
- 2022-09-21 CN CN202211146602.9A patent/CN115482217B/en active Active
Patent Citations (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4916450A (en) * | 1988-05-12 | 1990-04-10 | Radar Control Systems Corporation | Radar system for headway control of a vehicle |
CN102180139A (en) * | 2011-04-22 | 2011-09-14 | 钟延沁 | Proximity trigger safety air bag and safety control method thereof |
CN102826423A (en) * | 2011-06-14 | 2012-12-19 | 株式会社日立制作所 | Elevator safe operation system and safe operation method |
DE102013214305A1 (en) * | 2013-07-22 | 2015-01-22 | Bayerische Motoren Werke Aktiengesellschaft | Method for avoiding a collision of a vehicle with at least one object, driver assistance system and vehicle |
CN105953762A (en) * | 2016-07-07 | 2016-09-21 | 国网山东省电力公司经济技术研究院 | Power transmission line modeling method used for net distance measurement and distance measurement device thereof |
WO2018130016A1 (en) * | 2017-01-10 | 2018-07-19 | 哈尔滨工业大学深圳研究生院 | Parking detection method and device based on monitoring video |
JP2018206107A (en) * | 2017-06-06 | 2018-12-27 | 東芝デベロップメントエンジニアリング株式会社 | Approaching object detecting device |
CN108274469A (en) * | 2018-04-23 | 2018-07-13 | 东北大学 | Vacuum mechanical-arm anticollision detecting system and detection method based on multidimensional vision sensor |
CN111985388A (en) * | 2020-08-18 | 2020-11-24 | 深圳市自行科技有限公司 | Pedestrian attention detection driving assistance system, device and method |
CN113012315A (en) * | 2021-02-23 | 2021-06-22 | 浙江三一装备有限公司 | Safety monitoring system and method for working machine and working machine |
CN113052125A (en) * | 2021-04-09 | 2021-06-29 | 内蒙古科电数据服务有限公司 | Construction site violation image recognition and alarm method |
CN113344874A (en) * | 2021-06-04 | 2021-09-03 | 温州大学 | Pedestrian boundary crossing detection method based on Gaussian mixture modeling |
CN113890193A (en) * | 2021-09-29 | 2022-01-04 | 国网内蒙古东部电力有限公司呼伦贝尔供电公司 | Working state monitoring method and system for electric tracing system of transformer substation circuit breaker |
CN114261389A (en) * | 2021-11-29 | 2022-04-01 | 东风商用车有限公司 | Running track collision risk analysis method of automatic driving commercial vehicle |
CN114708508A (en) * | 2022-04-21 | 2022-07-05 | 广东电网有限责任公司东莞供电局 | Method for identifying moving small animal |
CN114913420A (en) * | 2022-05-17 | 2022-08-16 | 中国矿业大学 | Dynamic obstacle monitoring system and monitoring method for anti-impact drilling robot |
Non-Patent Citations (2)
Title |
---|
SIMONE MAXAND: "Identification of independent structural shocks in the presence of multiple Gaussian components", 《ECONOMETRICS AND STATISTICS》, vol. 16, 3 November 2018 (2018-11-03), pages 55 - 68 * |
朱红岷 等: "基于图像处理的变电站视频智能分析研究", 《计算机工程与应用》, vol. 54, no. 07, 1 April 2018 (2018-04-01), pages 264 - 270 * |
Also Published As
Publication number | Publication date |
---|---|
CN115482217B (en) | 2024-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106412501B (en) | A kind of the construction safety behavior intelligent monitor system and its monitoring method of video | |
CN106407928B (en) | Transformer composite insulator casing monitoring method and system based on raindrop identification | |
CN105404847B (en) | A kind of residue real-time detection method | |
CN110414400B (en) | Automatic detection method and system for wearing of safety helmet on construction site | |
CN111428617A (en) | Video image-based distribution network violation maintenance behavior identification method and system | |
CN107247946B (en) | Behavior recognition method and device | |
JP2007128513A (en) | Scene analysis | |
CN107911663A (en) | A kind of elevator passenger hazardous act intelligent recognition early warning system based on Computer Vision Detection | |
CN101727672A (en) | Method for detecting, tracking and identifying object abandoning/stealing event | |
CN104463253B (en) | Passageway for fire apparatus safety detection method based on adaptive background study | |
CN104392461B (en) | A kind of video tracing method based on textural characteristics | |
CN113096075B (en) | Invader detection method and system for power transmission line winding foreign matter cleaner | |
CN103945197B (en) | Electric power facility external force damage prevention early warning scheme based on Video Motion Detection technology | |
CN111738336A (en) | Image detection method based on multi-scale feature fusion | |
CN113469654B (en) | Multi-level safety control system of transformer substation based on intelligent algorithm fuses | |
CN113297926B (en) | Behavior detection and recognition method and system | |
CN114782897A (en) | Dangerous behavior detection method and system based on machine vision and deep learning | |
CN114782988A (en) | Construction environment-oriented multi-stage safety early warning method | |
Rentao et al. | Indoor smoking behavior detection based on yolov3-tiny | |
CN104077591A (en) | Intelligent and automatic computer monitoring system | |
CN115100813A (en) | Intelligent community system based on digital twins | |
CN107704818A (en) | A kind of fire detection system based on video image | |
Wang et al. | Vision-based highway traffic accident detection | |
CN113179389A (en) | System and method for identifying crane jib of power transmission line dangerous vehicle | |
CN115482217B (en) | Transformer substation electric shock prevention video detection method based on Gaussian mixture model separation algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |