CN108319934A - Safety cap wear condition detection method based on video stream data - Google Patents
Safety cap wear condition detection method based on video stream data Download PDFInfo
- Publication number
- CN108319934A CN108319934A CN201810227004.1A CN201810227004A CN108319934A CN 108319934 A CN108319934 A CN 108319934A CN 201810227004 A CN201810227004 A CN 201810227004A CN 108319934 A CN108319934 A CN 108319934A
- Authority
- CN
- China
- Prior art keywords
- safety cap
- cap
- picture
- video stream
- human body
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/752—Contour matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the safety cap wear condition detection methods based on video stream data, include the following steps：Video stream data obtains；Moving target is extracted；Human body target matches；Head positioning；Safety cap color-match；Safety cap outline；Safety cap wears characteristic matching.The present invention uses the safety cap wear condition detection method based on video stream data, breach single detection means, combine moving target detection, human body target matches, head positioning, safety cap color and outline and safety cap wear characteristic matching technology, substantially increase the discrimination that safety cap and safety cap are correctly worn, utilize method of the present invention, the automatic identification that may be implemented to wear operating personnel's safety cap of construction site situation detects, prevent non-safe wearing cap or hand from the operating personnel of safety cap being taken to enter construction area, and security control units in construction area at different levels is assisted to carry out construction area intelligence supervision, improve the construction area security control level of IT application.
Description
Technical field
The present invention relates to digital video Intellectual Analysis Technology field more particularly to a kind of safety caps based on video stream data
Wear condition detection method is guarantee and detection, the identification of a kind of safety cap safeguard procedures for all kinds of working site personnel
The technology of early warning.
Background technology
Current main detection target means, there is following a few classes：
First, Knowledge based engineering safety cap detection method, as template matches, face characteristic, shape and edge, texture features,
Color characteristic.Application based on HOG (histogram of gradients) feature is exactly a kind of pattern of classics.
Second is that the safety cap detection method based on statistics, such as principal component analysis and safety cap feature, neural network method, branch
Hold vector machine, hidden Markov model, Adaboost algorithm etc..Rudimentary algorithm of the class HAAR graders as machine vision,
It is widely used by industrial quarters.So-called grader refers to just the algorithm classified to safety cap and non-security cap herein,
Machine learning field, many algorithms are all the processes classified to things, clustered.ML machine learning modules in OpenCV carry
Supplied many classification, cluster algorithm, largely used in Practical Project.
Third, the method based on sensors such as RF tags, identifies whether to have worn safety cap.《A kind of safety cap, safety
Cap wearing state monitoring method and system》, it is for the safety guarantee of the people itself most paid close attention to and inadequate, and safety in use
Cap because lose, damage, replace etc. factors, cause the cost of this set system of operation and maintenance higher, lack market application at
This advantage.
Existing machine learning method is directed to the recognition methods of safety cap, is confined to show still under fc-specific test FC collection, but
For under practical widely used scene, it may appear that it is inadaptable, it shows as not detecting, or similar color background is considered as safety
The error detection situation of cap.
Using static classification detector approach, with《A kind of personnel safety cap wear condition based on video analysis is examined in real time
Survey method》, the viola-Jones detectors that it is used, at the scene in use due to misidentify people, can cause to sentence based on this object
Disconnected result is judged by accident.
Movable people's target is detected using dynamic video stream, then carries out safety cap and knows method for distinguishing, such as《Pacify construction site
Full cap wear condition monitoring method》, comparative static classification and Detection device method increases moving target, reduces static similar mesh
Target is interfered.The advantageous accuracy rate for promoting detection target person and judging, but for the adaptation under the several scenes of safety cap, also not
It is enough.And in true deployment, RGB color indicates, for the variation extreme sensitivity of illumination brightness, can not pass through phase
Reflect the color of safety cap to stable threshold value.Since its value threshold variation range is uncertain, cause in practical applications, it is different
Scene will be debugged individually, and use cost and maintenance cost is caused to increase.It is difficult to carry out batch application in the market.
Existing safety cap recognition methods is analyzed, it is substantially special from state sensor, RF tag, static images safety cap
Sign identification, the identification of dynamic video safety cap are classified.Main technical problem is that the universal of actual scene in use is applicable in
Property it is insufficient, cause use cost, implementation and maintenance cost high, it is difficult to be converted into product and be supplied to user.
Invention content
In order to solve the problems mentioned above in the background art, the purpose of the present invention is to propose to based on video stream data
The automatic identification inspection that situation is worn to operating personnel's safety cap of construction site may be implemented in safety cap wear condition detection method
It surveys, prevents non-safe wearing cap or hand from the operating personnel of safety cap being taken to enter construction area.
To achieve the above object, the technical solution that the present invention takes is：
Safety cap wear condition detection method based on video stream data, includes the following steps：
Step 1: video stream data obtains
Camera is arranged in the zone of action of middle operating personnel at the construction field (site), obtains camera video stream, and to video flowing
Data are decoded frame by frame, are converted into Lab space expression, it is made to be converted to corresponding coloured image；
Step 2: moving target is extracted
By carrying out background modeling to the n frame pictures obtained in video, then the moving target in n+1 frame pictures is carried out
Frame is poor, and n+1 frame pixel value I (x, y) are subtracted to the average value u (x, y) of same position pixel in background model, obtain difference d
Difference d (x, y) is then compared by (x, y) with threshold value TH, when difference d (x, y) is more than threshold value TH, then before being labeled as
Sight spot；Otherwise, it is labeled as background dot；Wherein, TH values are determined using adaptive algorithm, that is, between the 3 frame images for calculating each pixel
The average value of frame-to-frame differences and standard deviation and, as the standard TH compared；
Judge whether the moving target continuously moves by the continuous frame in foreground point, if it is continuous to occur, if being not achieved
Continuous N frames occur, then filter；Conversely, the continuous N frames of the moving target occur, and the target centroid coordinate of moving target in N frames
Position pixel is more than i pixel, then is judged as persistent movement, obtains the foreground picture of moving target；Wherein, N=[1,
200], the size of N values is reflected as the time span of object observing, this value is smaller, then the reaction time for providing judgement is faster, more
It is sensitive；I is expressed as the position amount of pixels on both horizontally and vertically, and the value range [1,20] of i, i is smaller, and detection is sensitiveer；
Step 3: human body target matches
The foreground picture of the moving target obtained in interception step 2 is matched with characteristics of human body's model, if more than similar
M is spent, then judges there is human body target in foreground picture, and is entered in next step；Conversely, then judging there is no human body in object to be measured image
Target, and return to step two continues the extraction operation of moving target；Wherein M=[0,1] M values are bigger, indicate that target is behaved
Possibility it is higher；
Step 4: head positioning
There is the foreground picture of human body target to continue to extract to what is obtained in step 3, and intercepts the foreground picture
30% rectangular area of middle Y coordinate maximum value is head and shoulder region, safety cap Matching band of the head and shoulder region as next step；
Step 5: safety cap color-match
Color reduction is carried out to the head and shoulder region picture being truncated in step 4 by Lab colors algorithm, and by area
Similar similar color point is merged connection by the monitoring of block；If do not monitor red, yellow, blue, white blocks or merge after it is red,
Yellow, blue, white blocks region is less than S pixels, then directly ignores, and return to step two；If monitor red, yellow, blue, white blocks and/
Or red, yellow, blue, the white blocks region after merging are more than S pixels, then judge head and shoulder region picture and safety cap color-match, into
Enter in next step；Wherein, S pixels are safety cap minimum pixel required value under different resolution, under 1080 × 720 resolution ratio, S
Pixel adjusting range is 100~1600 pixels, corresponds to the rectangle of 10 × 10~40 × 40 pixels；
Step 6: safety cap outline
The safety cap got in step 5 is provided to the head and shoulder region picture and safety cap contour feature mould of color-match
Type carries out safety cap outline, is judged as that wearing meets defined safety cap type if similarity is more than L, otherwise judges
For the safety cap type for wearing against regulation；The value range of threshold value L is [0,1], and required precision is higher, then closer to 1；
Step 7: safety cap wears characteristic matching
Safety cap is carried out to the head and shoulder region picture of safety cap style complies with set in step 6 and wears characteristic model matching,
If matching is more than similarity P, it is judged as correct safe wearing cap, on the contrary then return to step two；Wherein P=[0,1], precision is wanted
Ask higher, then closer to 1.
In above-mentioned technical proposal, characteristics of human body's model in step 3 is to train grader by neural network model
What training and identification obtained, specific method is：
When training, a large amount of human body pictures are inputted as positive sample, inputs largely without human body picture as negative sample, passes through god
It is trained through network model training grader and learns and obtain characteristics of human body's model；
When identification, the foreground picture of input motion target is instructed by the foreground picture and neural network model of moving target
Matching is identified in the characteristics of human body's model practiced in grader, if more than similarity R, then judges there is human body mesh in foreground picture
Mark；Wherein R=[0,1], R is bigger, then matching degree is higher.
In above-mentioned technical proposal, the model of safety cap contour feature described in step 6 is by the training of SVM training aids and to know
It does not obtain, specific method is：
When training, it is positive sample to input a large amount of correct safety cap pictures, and inputting a large amount of knitting wool caps, peaked cap, baseball cap is
Negative sample is trained by SVM training aids and obtains safety cap contour feature model；
When identification, the head and shoulder region picture of input and safety cap color-match passes through the head and shoulder with safety cap color-match
Region picture is identified with the safety cap contour feature model in SVM training aids and matches, if similarity is more than T, is judged as
Safety cap style complies with set；T=[0,1], T is bigger, then matching degree is higher.
It is to train grader by neural network model that safety cap, which wears characteristic model, in above-mentioned technical proposal, in step 7
What training and identification obtained, specific method is：
When training, a large amount of safe wearing cap personnel pictures are inputted as positive sample, input not safe wearing cap personnel picture
As negative sample, it is trained by training aids and learns and obtain safety cap wearing characteristic model；
When identification, the head and shoulder region picture of safety cap style complies with set is inputted, head zone picture and nerve net are passed through
Safety cap in network model training grader wears characteristic model and matching is identified, if matching is more than similarity U, is judged as
Correct safe wearing cap；Wherein U=[0,1], U is bigger, then matching degree is higher.
In above-mentioned technical proposal, in step 5, under 1080 × 720 resolution ratio, S pixel adjusting ranges are 400 pictures
Element corresponds to the rectangle of 20 × 20 pixels.
In above-mentioned technical proposal, in step 6, the empirical value of the threshold value L takes 0.4.
In above-mentioned technical proposal, in step 7, the empirical value of P takes 0.4.
Compared with prior art, the beneficial effects of the present invention are：
The present invention is matched by human body target and head positioning, avoids the flase drop of the appearance of recognition of face in the prior art；
By safety cap color and outline, the appearance such as other caps such as peaked cap, knitting wool cap may be worn by avoiding operating personnel
Flase drop；By head positioning and safety cap color and outline, realizes the detection of correct safe wearing cap, avoid operation
Personnel's hand takes the flase drop of situations such as safety cap, and further wears aspect ratio pair by safety cap, substantially increases safety cap
The discrimination correctly worn.
The present invention uses the safety cap wear condition detection method based on video stream data, breaches single detection hand
Section, combines moving target detection, human body target matching, head positioning, safety cap color and outline and safety cap is worn
Characteristic matching technology is worn, the discrimination that safety cap and safety cap are correctly worn is substantially increased, utilizes side of the present invention
Method may be implemented the automatic identification detection for wearing situation to operating personnel's safety cap of construction site, prevent non-safe wearing cap
Or hand takes the operating personnel of safety cap to enter construction area, substitutes the traditional mode of manual patrol inspection, saves manpower, and auxiliary
It helps security control units in construction area at different levels to carry out construction area intelligence supervision, improves the construction area security control level of IT application.
Description of the drawings
Fig. 1 is the flow chart of the safety cap wear condition detection method provided by the invention based on video stream data；
Fig. 2 is the stream that characteristics of human body's model of the present invention trains classifier training and identification by neural network model
Cheng Tu；
Fig. 3 is flow chart of the safety cap contour feature model of the present invention by SVM training aids training and identification；
Fig. 4 is that safety cap of the present invention wearing characteristic model passes through neural network model training classifier training and knowledge
Other flow chart.
Specific implementation mode
Below in conjunction with the drawings and specific embodiments, the present invention is described in further detail：
As shown in Figures 1 to 4, the invention discloses a kind of safety cap wear condition detection side based on video stream data
Method includes the following steps：
Step 1: video stream data obtains
Camera is arranged in the zone of action of middle operating personnel at the construction field (site), obtains camera video stream, and to video flowing
Data are decoded frame by frame, are converted into Lab space expression, it is made to be converted to corresponding coloured image；
Step 2: moving target is extracted
By carrying out background modeling to the n frame pictures obtained in video, then the moving target in n+1 frame pictures is carried out
Frame is poor, and n+1 frame pixel value I (x, y) are subtracted to the average value u (x, y) of same position pixel in background model, obtain difference d
Difference d (x, y) is then compared by (x, y) with threshold value TH, when difference d (x, y) is more than threshold value TH, then before being labeled as
Sight spot；Otherwise, it is labeled as background dot；Wherein, TH values are determined using adaptive algorithm, that is, between the 3 frame images for calculating each pixel
The average value of frame-to-frame differences and standard deviation and, as the standard TH compared；
Judge whether the moving target continuously moves by the continuous frame in foreground point, if it is continuous to occur, if being not achieved
(N=[1,200], the size of N values are reflected as the time span of object observing, this value is smaller, then provides judgement for continuous N frames appearance
Reaction time it is faster, it is sensitiveer), then filter；Conversely, the continuous N frames of the moving target occur, and in N frames moving target mesh
The position pixel of mark center-of-mass coordinate is more than i pixel, and (i is expressed as the position amount of pixels on both horizontally and vertically, the value of i
Range [1,20], i is smaller, and detection is sensitiveer), then it is judged as persistent movement, obtains the foreground picture of moving target；
Step 3: human body target matches
The foreground picture of the moving target obtained in interception step 2 is matched with characteristics of human body's model, if more than similar
M is spent, then judges there is human body target in foreground picture, and is entered in next step；Conversely, then judging there is no human body in object to be measured image
Target, and return to step two continues the extraction operation of moving target；Wherein M=[0,1] M values are bigger, indicate that target is behaved
Possibility it is higher；
Step 4: head positioning
There is the foreground picture of human body target to continue to extract to what is obtained in step 3, and intercepts the foreground picture
30% rectangular area of middle Y coordinate maximum value is head and shoulder region, safety cap Matching band of the head and shoulder region as next step；
Step 5: safety cap color-match
Color reduction is carried out to the head and shoulder region picture being truncated in step 4 by Lab colors algorithm, and by area
Similar similar color point is merged connection by the monitoring of block；If do not monitor red, yellow, blue, white blocks or merge after it is red,
Yellow, blue, white blocks region is less than S pixels, then directly ignores, and return to step two；If monitor red, yellow, blue, white blocks and/
Or red, yellow, blue, the white blocks region after merging are more than S pixels, then judge head and shoulder region picture and safety cap color-match, into
Enter in next step；Wherein, S pixels are safety cap minimum pixel required value under different resolution, under 1080 × 720 resolution ratio, S
Pixel adjusting range is 100~1600 pixels (rectangle for corresponding to 10 × 10~40 × 40 pixels)；Preferably, S pixels tune
Whole ranging from 400 pixels correspond to the rectangle of 20 × 20 pixels；
Step 6: safety cap outline
The safety cap got in step 5 is provided to the head and shoulder region picture and safety cap contour feature mould of color-match
Type carries out safety cap outline, is judged as that wearing meets defined safety cap type if similarity is more than L, otherwise judges
For the safety cap type for wearing against regulation；The value range of threshold value L is [0,1], and required precision is higher, then closer to 1.It is excellent
Choosing, the empirical value of the threshold value L takes 0.4；
Step 7: safety cap wears characteristic matching
Safety cap is carried out to the head and shoulder region picture of safety cap style complies with set in step 6 and wears characteristic model matching,
If matching is more than similarity P, it is judged as correct safe wearing cap, on the contrary then return to step two；Wherein P=[0,1], precision is wanted
Ask higher, then closer to 1.Preferably, the empirical value of P takes 0.4.
As shown in Fig. 2, characteristics of human body's model in step 3 is to train classifier training by neural network model
It is obtained with identification, specific method is：
When training, a large amount of human body pictures are inputted as positive sample, inputs largely without human body picture as negative sample, passes through god
It is trained through network model training grader and learns and obtain characteristics of human body's model；
When identification, the foreground picture of input motion target is instructed by the foreground picture and neural network model of moving target
Matching is identified in the characteristics of human body's model practiced in grader, if more than similarity R, then judges there is human body mesh in foreground picture
Mark；Wherein R=[0,1], R is bigger, then matching degree is higher.
As shown in figure 3, the model of safety cap contour feature described in step 6 is obtained by the training of SVM training aids and identification
, specific method is：
When training, it is positive sample to input a large amount of correct safety cap pictures, and inputting a large amount of knitting wool caps, peaked cap, baseball cap is
Negative sample is trained by SVM training aids and obtains safety cap contour feature model；
When identification, the head and shoulder region picture of input and safety cap color-match passes through the head and shoulder with safety cap color-match
Region picture is identified with the safety cap contour feature model in SVM training aids and matches, if similarity is more than T, is judged as
Safety cap style complies with set；T=[0,1], T is bigger, then matching degree is higher.
As shown in figure 4, it is to train classifier training by neural network model that safety cap, which wears characteristic model, in step 7
It is obtained with identification, specific method is：
When training, a large amount of safe wearing cap personnel pictures are inputted as positive sample, input not safe wearing cap personnel picture
As negative sample, it is trained by training aids and learns and obtain safety cap wearing characteristic model；
When identification, the head and shoulder region picture of safety cap style complies with set is inputted, head zone picture and nerve net are passed through
Safety cap in network model training grader wears characteristic model and matching is identified, if matching is more than similarity U, is judged as
Correct safe wearing cap；Wherein U=[0,1], U is bigger, then matching degree is higher.
Finally illustrate, the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although with reference to compared with
Good embodiment describes the invention in detail, it will be understood by those of ordinary skill in the art that, it can be to the skill of the present invention
Art scheme is modified or replaced equivalently, and without departing from the objective and range of technical solution of the present invention, should all be covered at this
In the right of invention.
Claims (7)
1. the safety cap wear condition detection method based on video stream data, which is characterized in that include the following steps：
Step 1: video stream data obtains
Camera is arranged in the zone of action of middle operating personnel at the construction field (site), obtains camera video stream, and to video stream data
It is decoded frame by frame, is converted into Lab space expression, it is made to be converted to corresponding coloured image；
Step 2: moving target is extracted
By carrying out background modeling to the n frame pictures obtained in video, frame then is carried out to the moving target in n+1 frame pictures
N+1 frame pixel value I (x, y) are subtracted the average value u (x, y) of same position pixel in background model by difference, obtain difference d (x,
Y), then difference d (x, y) is compared with threshold value TH, when difference d (x, y) is more than threshold value TH, is then labeled as foreground
Point；Otherwise, it is labeled as background dot；Wherein, TH values are determined using adaptive algorithm, that is, calculate frame between 3 frame images of each pixel
Between poor average value and standard deviation and, as the standard TH compared；
Judge whether the moving target continuously moves by the continuous frame in foreground point, if it is continuous to occur, if continuous N is not achieved
Frame occurs, then filters；Conversely, the continuous N frames of the moving target occur, and in N frames the target centroid coordinate of moving target position
Pixel is more than i pixel, then is judged as persistent movement, obtains the foreground picture of moving target；Wherein, [1,200] N=, N values
Size, be reflected as the time span of object observing, this value is smaller, then the reaction time for providing judgement is faster, sensitiveer；I tables
The position amount of pixels being shown as on both horizontally and vertically, the value range [1,20] of i, i is smaller, and detection is sensitiveer；
Step 3: human body target matches
The foreground picture of moving target obtained in interception step 2 is matched with characteristics of human body's model, if more than similarity M,
Then judge there is human body target in foreground picture, and enters in next step；Conversely, then judging there is no human body mesh in object to be measured image
Mark, and return to step two continues the extraction operation of moving target；Wherein M=[0,1] M values are bigger, indicate that target is people's
Possibility is higher；
Step 4: head positioning
There is the foreground picture of human body target to continue to extract to what is obtained in step 3, and intercepts Y in the foreground picture
30% rectangular area of coordinate maximum value is head and shoulder region, safety cap Matching band of the head and shoulder region as next step；
Step 5: safety cap color-match
Color reduction is carried out to the head and shoulder region picture being truncated in step 4 by Lab colors algorithm, and by block
Similar similar color point is merged connection by monitoring；If do not monitor red, yellow, blue, white blocks or merge after it is red, yellow,
Blue, white blocks region is less than S pixels, then directly ignores, and return to step two；If monitoring red, yellow, blue, white blocks and/or conjunction
Red, yellow, blue, white blocks region after and are more than S pixels, then head and shoulder region picture and safety cap color-match are judged, under
One step；Wherein, S pixels are safety cap minimum pixel required value under different resolution, under 1080 × 720 resolution ratio, S pixels
Adjusting range is 100~1600 pixels, corresponds to the rectangle of 10 × 10~40 × 40 pixels；
Step 6: safety cap outline
By the safety cap got in step 5 provide the head and shoulder region picture of color-match and safety cap contour feature model into
Row safety cap outline is judged as that wearing meets defined safety cap type if similarity is more than L, otherwise is judged as wearing
Wear safety cap type against regulation；The value range of threshold value L is [0,1], and required precision is higher, then closer to 1；
Step 7: safety cap wears characteristic matching
Safety cap is carried out to the head and shoulder region picture of safety cap style complies with set in step 6 and wears characteristic model matching, if
With more than similarity P, being then judged as correct safe wearing cap, on the contrary then return to step two；Wherein P=[0,1], required precision is got over
Height, then closer to 1.
2. the safety cap wear condition detection method according to claim 1 based on video stream data, it is characterised in that：Step
Characteristics of human body's model in rapid three trains classifier training and identification to obtain by neural network model, specific side
Method is：
When training, a large amount of human body pictures are inputted as positive sample, inputs largely without human body picture as negative sample, passes through nerve net
Network model training grader, which is trained, to be learnt and obtains characteristics of human body's model；
When identification, the foreground picture of input motion target passes through foreground picture and the neural network model training point of moving target
Matching is identified in characteristics of human body's model in class device, if more than similarity R, then judges there is human body target in foreground picture；Its
Middle R=[0,1], R is bigger, then matching degree is higher.
3. the safety cap wear condition detection method according to claim 1 based on video stream data, which is characterized in that step
Safety cap contour feature model described in rapid six is obtained by the training of SVM training aids and identification, and specific method is：
When training, it is positive sample to input a large amount of correct safety cap pictures, and it is negative sample to input a large amount of knitting wool caps, peaked cap, baseball cap
This, is trained by SVM training aids and obtains safety cap contour feature model；
When identification, the head and shoulder region picture of input and safety cap color-match passes through the head and shoulder region with safety cap color-match
Picture is identified with the safety cap contour feature model in SVM training aids and matches, if similarity is more than T, is judged as safety
Cap style complies with set；T=[0,1], T is bigger, then matching degree is higher.
4. the safety cap wear condition detection method according to claim 1 based on video stream data, which is characterized in that step
Safety cap is worn characteristic model and is obtained by neural network model training classifier training and identification in rapid seven, specific side
Method is：
When training, a large amount of safe wearing cap personnel pictures are inputted as positive sample, input not safe wearing cap personnel picture conduct
Negative sample is trained by training aids and learns and obtain safety cap wearing characteristic model；
When identification, the head and shoulder region picture of safety cap style complies with set is inputted, head zone picture and neural network mould are passed through
Safety cap in type training grader wears characteristic model and matching is identified, if matching is more than similarity U, is judged as correct
Safe wearing cap；Wherein U=[0,1], U is bigger, then matching degree is higher.
5. the safety cap recognition methods according to claim 1 based on video stream data, it is characterised in that：In step 5,
Under 1080 × 720 resolution ratio, S pixel adjusting ranges are 400 pixels, correspond to the rectangle of 20 × 20 pixels.
6. the safety cap recognition methods according to claim 1 based on video stream data, it is characterised in that：In step 6,
The empirical value of the threshold value L takes 0.4.
7. the safety cap recognition methods according to claim 1 based on video stream data, it is characterised in that：In step 7, P
Empirical value take 0.4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810227004.1A CN108319934A (en) | 2018-03-20 | 2018-03-20 | Safety cap wear condition detection method based on video stream data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810227004.1A CN108319934A (en) | 2018-03-20 | 2018-03-20 | Safety cap wear condition detection method based on video stream data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108319934A true CN108319934A (en) | 2018-07-24 |
Family
ID=62898933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810227004.1A Pending CN108319934A (en) | 2018-03-20 | 2018-03-20 | Safety cap wear condition detection method based on video stream data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108319934A (en) |
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109117827A (en) * | 2018-09-05 | 2019-01-01 | 武汉市蓝领英才科技有限公司 | Work clothes work hat wearing state automatic identifying method and alarm system based on video |
| CN109255298A (en) * | 2018-08-07 | 2019-01-22 | 南京工业大学 | Safety helmet detection method and system in dynamic background |
| CN109635697A (en) * | 2018-12-04 | 2019-04-16 | 国网浙江省电力有限公司电力科学研究院 | Electric operating personnel safety dressing detection method based on YOLOv3 target detection |
| CN109670441A (en) * | 2018-12-14 | 2019-04-23 | 广东亿迅科技有限公司 | A kind of realization safety cap wearing knows method for distinguishing, system, terminal and computer readable storage medium |
| CN109672863A (en) * | 2018-12-24 | 2019-04-23 | 海安常州大学高新技术研发中心 | A kind of construction personnel's safety equipment intelligent monitoring method based on image recognition |
| CN109697430A (en) * | 2018-12-28 | 2019-04-30 | 成都思晗科技股份有限公司 | The detection method that working region safety cap based on image recognition is worn |
| CN109766869A (en) * | 2019-01-23 | 2019-05-17 | 中国建筑第八工程局有限公司 | A kind of artificial intelligent detecting method of safety cap wearing based on machine vision |
| CN110119686A (en) * | 2019-04-17 | 2019-08-13 | 电子科技大学 | A kind of safety cap real-time detection method based on convolutional neural networks |
| CN110263609A (en) * | 2019-01-27 | 2019-09-20 | 杭州品茗安控信息技术股份有限公司 | A kind of automatic identifying method of safety cap wear condition |
| CN110414339A (en) * | 2019-06-21 | 2019-11-05 | 武汉倍特威视系统有限公司 | Hearing room personnel's close contact recognition methods based on video stream data |
| CN110427812A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Colliery industry driving not pedestrian detection method based on video stream data |
| CN110427808A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Police uniform recognition methods based on video stream data |
| CN110427811A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Skeleton based on video stream data is fought recognition methods |
| CN110619314A (en) * | 2019-09-24 | 2019-12-27 | 杭州宇泛智能科技有限公司 | Safety helmet detection method and device and electronic equipment |
| CN110781833A (en) * | 2019-10-28 | 2020-02-11 | 杭州宇泛智能科技有限公司 | Authentication method and device and electronic equipment |
| CN110796049A (en) * | 2019-10-18 | 2020-02-14 | 国家电网有限公司 | Production worker safety helmet wearing detection method and system based on image processing |
| CN110852170A (en) * | 2019-10-12 | 2020-02-28 | 北京文安智能技术股份有限公司 | Personnel safety helmet detection method, device and system |
| CN110889376A (en) * | 2019-11-28 | 2020-03-17 | 创新奇智(南京)科技有限公司 | Safety helmet wearing detection system and method based on deep learning |
| CN111083441A (en) * | 2019-12-18 | 2020-04-28 | 广州穗能通能源科技有限责任公司 | Construction site monitoring method and device, computer equipment and storage medium |
| CN111079731A (en) * | 2019-12-03 | 2020-04-28 | 中冶赛迪重庆信息技术有限公司 | Configuration system, method, equipment and medium based on safety helmet identification monitoring system |
| CN111191581A (en) * | 2019-12-27 | 2020-05-22 | 深圳供电局有限公司 | Safety helmet detection method and device based on electric power construction and computer equipment |
| CN111401310A (en) * | 2020-04-08 | 2020-07-10 | 天津中科智能识别产业技术研究院有限公司 | Kitchen health safety supervision and management method based on artificial intelligence |
| CN111414873A (en) * | 2020-03-26 | 2020-07-14 | 广州粤建三和软件股份有限公司 | Alarm prompting method, device and alarm system based on wearing state of safety helmet |
| CN111428641A (en) * | 2020-03-24 | 2020-07-17 | 深圳供电局有限公司 | Secure dressing detection method and device, computer equipment and readable storage medium |
| CN111597903A (en) * | 2020-04-17 | 2020-08-28 | 江门明浩电力工程监理有限公司 | Intelligent monitoring box for distribution network and monitoring method thereof |
| CN111623751A (en) * | 2020-05-31 | 2020-09-04 | 张冬梅 | System for identifying the inclination of a visor and corresponding terminal |
| CN111616457A (en) * | 2020-05-06 | 2020-09-04 | 国网浙江省电力有限公司衢州供电公司 | Intelligent safety helmet for construction safety |
| CN112084986A (en) * | 2020-09-16 | 2020-12-15 | 国网福建省电力有限公司营销服务中心 | Real-time safety helmet detection method based on image feature extraction |
| CN112149513A (en) * | 2020-08-28 | 2020-12-29 | 成都飞机工业（集团）有限责任公司 | Industrial manufacturing site safety helmet wearing identification system and method based on deep learning |
| CN112232313A (en) * | 2020-03-06 | 2021-01-15 | 杭州宇泛智能科技有限公司 | Method and device for detecting wearing state of personal safety helmet in video and electronic equipment |
| CN112949354A (en) * | 2019-12-10 | 2021-06-11 | 顺丰科技有限公司 | Method and device for detecting wearing of safety helmet, electronic equipment and computer-readable storage medium |
| CN113158851A (en) * | 2021-04-07 | 2021-07-23 | 浙江大华技术股份有限公司 | Wearing safety helmet detection method and device and computer storage medium |
| US20210326901A1 (en) * | 2020-04-20 | 2021-10-21 | International Business Machines Corporation | Managing apparel to facilitate compliance |
| CN113627406A (en) * | 2021-10-12 | 2021-11-09 | 南方电网数字电网研究院有限公司 | Abnormal behavior detection method and device, computer equipment and storage medium |
| CN118230245A (en) * | 2024-03-29 | 2024-06-21 | 连云港酷思茂建材有限公司 | Intelligent identification system based on visual detection data |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110314595A1 (en) * | 2009-02-18 | 2011-12-29 | Nobumasa Kobayashi | Work-mask wearing device and helmet using the same |
| CN103745226A (en) * | 2013-12-31 | 2014-04-23 | 国家电网公司 | Dressing safety detection method for worker on working site of electric power facility |
| CN104036575A (en) * | 2014-07-01 | 2014-09-10 | 江苏省南京市公路管理处公路科学研究所 | Safety helmet wearing condition monitoring method on construction site |
| CN104077581A (en) * | 2014-07-21 | 2014-10-01 | 国家电网公司 | Smart video identification method for wearing safety helmet rightly |
| CN106156765A (en) * | 2016-08-30 | 2016-11-23 | 南京邮电大学 | safety detection method based on computer vision |
| CN106372662A (en) * | 2016-08-30 | 2017-02-01 | 腾讯科技（深圳）有限公司 | Helmet wearing detection method and device, camera, and server |
| CN106503716A (en) * | 2016-09-13 | 2017-03-15 | 中国电力科学研究院 | A kind of safety cap recognition methods that is extracted based on color and contour feature and system |
| CN107679524A (en) * | 2017-10-31 | 2018-02-09 | 天津天地伟业信息系统集成有限公司 | A kind of detection method of the safety cap wear condition based on video |
-
2018
- 2018-03-20 CN CN201810227004.1A patent/CN108319934A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110314595A1 (en) * | 2009-02-18 | 2011-12-29 | Nobumasa Kobayashi | Work-mask wearing device and helmet using the same |
| CN103745226A (en) * | 2013-12-31 | 2014-04-23 | 国家电网公司 | Dressing safety detection method for worker on working site of electric power facility |
| CN104036575A (en) * | 2014-07-01 | 2014-09-10 | 江苏省南京市公路管理处公路科学研究所 | Safety helmet wearing condition monitoring method on construction site |
| CN104077581A (en) * | 2014-07-21 | 2014-10-01 | 国家电网公司 | Smart video identification method for wearing safety helmet rightly |
| CN106156765A (en) * | 2016-08-30 | 2016-11-23 | 南京邮电大学 | safety detection method based on computer vision |
| CN106372662A (en) * | 2016-08-30 | 2017-02-01 | 腾讯科技（深圳）有限公司 | Helmet wearing detection method and device, camera, and server |
| CN106503716A (en) * | 2016-09-13 | 2017-03-15 | 中国电力科学研究院 | A kind of safety cap recognition methods that is extracted based on color and contour feature and system |
| CN107679524A (en) * | 2017-10-31 | 2018-02-09 | 天津天地伟业信息系统集成有限公司 | A kind of detection method of the safety cap wear condition based on video |
Non-Patent Citations (1)
| Title |
|---|
| 丁红军等: "《视频侦查》", 31 October 2015, 中国人民公安大学出版社 * |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109255298A (en) * | 2018-08-07 | 2019-01-22 | 南京工业大学 | Safety helmet detection method and system in dynamic background |
| CN109117827B (en) * | 2018-09-05 | 2020-11-24 | 武汉市蓝领英才科技有限公司 | Video-based method for automatically identifying wearing state of work clothes and work cap and alarm system |
| CN109117827A (en) * | 2018-09-05 | 2019-01-01 | 武汉市蓝领英才科技有限公司 | Work clothes work hat wearing state automatic identifying method and alarm system based on video |
| CN109635697A (en) * | 2018-12-04 | 2019-04-16 | 国网浙江省电力有限公司电力科学研究院 | Electric operating personnel safety dressing detection method based on YOLOv3 target detection |
| CN109670441B (en) * | 2018-12-14 | 2024-02-06 | 广东亿迅科技有限公司 | Method, system, terminal and computer readable storage medium for realizing wearing recognition of safety helmet |
| CN109670441A (en) * | 2018-12-14 | 2019-04-23 | 广东亿迅科技有限公司 | A kind of realization safety cap wearing knows method for distinguishing, system, terminal and computer readable storage medium |
| CN109672863A (en) * | 2018-12-24 | 2019-04-23 | 海安常州大学高新技术研发中心 | A kind of construction personnel's safety equipment intelligent monitoring method based on image recognition |
| CN109697430A (en) * | 2018-12-28 | 2019-04-30 | 成都思晗科技股份有限公司 | The detection method that working region safety cap based on image recognition is worn |
| CN109766869A (en) * | 2019-01-23 | 2019-05-17 | 中国建筑第八工程局有限公司 | A kind of artificial intelligent detecting method of safety cap wearing based on machine vision |
| CN110263609A (en) * | 2019-01-27 | 2019-09-20 | 杭州品茗安控信息技术股份有限公司 | A kind of automatic identifying method of safety cap wear condition |
| CN110119686A (en) * | 2019-04-17 | 2019-08-13 | 电子科技大学 | A kind of safety cap real-time detection method based on convolutional neural networks |
| CN110119686B (en) * | 2019-04-17 | 2020-09-25 | 电子科技大学 | Safety helmet real-time detection method based on convolutional neural network |
| CN110427812A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Colliery industry driving not pedestrian detection method based on video stream data |
| CN110427811A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Skeleton based on video stream data is fought recognition methods |
| CN110427808A (en) * | 2019-06-21 | 2019-11-08 | 武汉倍特威视系统有限公司 | Police uniform recognition methods based on video stream data |
| CN110414339A (en) * | 2019-06-21 | 2019-11-05 | 武汉倍特威视系统有限公司 | Hearing room personnel's close contact recognition methods based on video stream data |
| CN110619314A (en) * | 2019-09-24 | 2019-12-27 | 杭州宇泛智能科技有限公司 | Safety helmet detection method and device and electronic equipment |
| CN110852170A (en) * | 2019-10-12 | 2020-02-28 | 北京文安智能技术股份有限公司 | Personnel safety helmet detection method, device and system |
| CN110796049A (en) * | 2019-10-18 | 2020-02-14 | 国家电网有限公司 | Production worker safety helmet wearing detection method and system based on image processing |
| CN110781833A (en) * | 2019-10-28 | 2020-02-11 | 杭州宇泛智能科技有限公司 | Authentication method and device and electronic equipment |
| CN110889376A (en) * | 2019-11-28 | 2020-03-17 | 创新奇智(南京)科技有限公司 | Safety helmet wearing detection system and method based on deep learning |
| CN111079731A (en) * | 2019-12-03 | 2020-04-28 | 中冶赛迪重庆信息技术有限公司 | Configuration system, method, equipment and medium based on safety helmet identification monitoring system |
| CN112949354A (en) * | 2019-12-10 | 2021-06-11 | 顺丰科技有限公司 | Method and device for detecting wearing of safety helmet, electronic equipment and computer-readable storage medium |
| CN111083441A (en) * | 2019-12-18 | 2020-04-28 | 广州穗能通能源科技有限责任公司 | Construction site monitoring method and device, computer equipment and storage medium |
| CN111191581B (en) * | 2019-12-27 | 2024-04-12 | 深圳供电局有限公司 | Safety helmet detection method and device based on electric power construction and computer equipment |
| CN111191581A (en) * | 2019-12-27 | 2020-05-22 | 深圳供电局有限公司 | Safety helmet detection method and device based on electric power construction and computer equipment |
| CN112232313A (en) * | 2020-03-06 | 2021-01-15 | 杭州宇泛智能科技有限公司 | Method and device for detecting wearing state of personal safety helmet in video and electronic equipment |
| CN111428641A (en) * | 2020-03-24 | 2020-07-17 | 深圳供电局有限公司 | Secure dressing detection method and device, computer equipment and readable storage medium |
| CN111414873A (en) * | 2020-03-26 | 2020-07-14 | 广州粤建三和软件股份有限公司 | Alarm prompting method, device and alarm system based on wearing state of safety helmet |
| CN111414873B (en) * | 2020-03-26 | 2021-04-30 | 广州粤建三和软件股份有限公司 | Alarm prompting method, device and alarm system based on wearing state of safety helmet |
| CN111401310A (en) * | 2020-04-08 | 2020-07-10 | 天津中科智能识别产业技术研究院有限公司 | Kitchen health safety supervision and management method based on artificial intelligence |
| CN111401310B (en) * | 2020-04-08 | 2023-08-29 | 天津中科智能识别产业技术研究院有限公司 | Kitchen sanitation safety supervision and management method based on artificial intelligence |
| CN111597903A (en) * | 2020-04-17 | 2020-08-28 | 江门明浩电力工程监理有限公司 | Intelligent monitoring box for distribution network and monitoring method thereof |
| US11645600B2 (en) * | 2020-04-20 | 2023-05-09 | International Business Machines Corporation | Managing apparel to facilitate compliance |
| US20210326901A1 (en) * | 2020-04-20 | 2021-10-21 | International Business Machines Corporation | Managing apparel to facilitate compliance |
| CN111616457A (en) * | 2020-05-06 | 2020-09-04 | 国网浙江省电力有限公司衢州供电公司 | Intelligent safety helmet for construction safety |
| CN111623751B (en) * | 2020-05-31 | 2021-11-30 | 上海域唯医疗科技有限公司 | System for identifying the inclination of a visor and corresponding terminal |
| CN111623751A (en) * | 2020-05-31 | 2020-09-04 | 张冬梅 | System for identifying the inclination of a visor and corresponding terminal |
| CN112149513A (en) * | 2020-08-28 | 2020-12-29 | 成都飞机工业（集团）有限责任公司 | Industrial manufacturing site safety helmet wearing identification system and method based on deep learning |
| CN112084986A (en) * | 2020-09-16 | 2020-12-15 | 国网福建省电力有限公司营销服务中心 | Real-time safety helmet detection method based on image feature extraction |
| CN113158851B (en) * | 2021-04-07 | 2022-08-09 | 浙江大华技术股份有限公司 | Wearing safety helmet detection method and device and computer storage medium |
| CN113158851A (en) * | 2021-04-07 | 2021-07-23 | 浙江大华技术股份有限公司 | Wearing safety helmet detection method and device and computer storage medium |
| CN113627406A (en) * | 2021-10-12 | 2021-11-09 | 南方电网数字电网研究院有限公司 | Abnormal behavior detection method and device, computer equipment and storage medium |
| CN118230245A (en) * | 2024-03-29 | 2024-06-21 | 连云港酷思茂建材有限公司 | Intelligent identification system based on visual detection data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108319934A (en) | Safety cap wear condition detection method based on video stream data | |
| CN108460358A (en) | Safety cap recognition methods based on video stream data | |
| CN106295551B (en) | A kind of personnel safety cap wear condition real-time detection method based on video analysis | |
| CN107133564B (en) | Tooling cap detection method | |
| CN110837784B (en) | Examination room peeping and cheating detection system based on human head characteristics | |
| Graf et al. | Multi-modal system for locating heads and faces | |
| CN104063722B (en) | A kind of detection of fusion HOG human body targets and the safety cap recognition methods of SVM classifier | |
| US7916904B2 (en) | Face region detecting device, method, and computer readable recording medium | |
| CN110008831A (en) | A kind of Intellectualized monitoring emerging system based on computer vision analysis | |
| CN106446926A (en) | Transformer station worker helmet wear detection method based on video analysis | |
| CN112396658B (en) | Indoor personnel positioning method and system based on video | |
| US7620216B2 (en) | Method of tracking a human eye in a video image | |
| CN108549884A (en) | A kind of biopsy method and device | |
| CN103279737B (en) | A kind of behavioral value method of fighting based on space-time interest points | |
| CN104951773A (en) | Real-time face recognizing and monitoring system | |
| CN106682578B (en) | Weak light face recognition method based on blink detection | |
| CN103942539B (en) | A kind of oval accurate high efficiency extraction of head part and masking method for detecting human face | |
| CN110263609A (en) | A kind of automatic identifying method of safety cap wear condition | |
| CN106846734A (en) | A kind of fatigue driving detection device and method | |
| CN104616438A (en) | Yawning action detection method for detecting fatigue driving | |
| CN105844245A (en) | Fake face detecting method and system for realizing same | |
| CN105893963B (en) | A kind of method of the best frame easy to identify of single pedestrian target in screening video | |
| CN114612823A (en) | Personnel behavior monitoring method for laboratory safety management | |
| CN109741565A (en) | Coal-mine fire identifying system and method | |
| CN209657454U (en) | Coal-mine fire identifying system |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180724 |
|
| RJ01 | Rejection of invention patent application after publication |