CN108960131A - The anti-people of mechanical garage is strayed into detection method - Google Patents
The anti-people of mechanical garage is strayed into detection method Download PDFInfo
- Publication number
- CN108960131A CN108960131A CN201810706894.4A CN201810706894A CN108960131A CN 108960131 A CN108960131 A CN 108960131A CN 201810706894 A CN201810706894 A CN 201810706894A CN 108960131 A CN108960131 A CN 108960131A
- Authority
- CN
- China
- Prior art keywords
- gradient
- strayed
- garage
- pixel
- controller
- 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
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
Be strayed into detection method the invention discloses a kind of anti-people of mechanical garage, 1), detection system is established: including controller, camera；2), camera in real time shoots mechanical garage vehicles while passing mouth, and by the delivery of video of shooting to controller, the characteristic extracting module of controller extracts the HOG feature of current video frame；3) the HOG feature of current video frame, is collected, and is incorporated into final feature vector and is used for classification；4), HOG feature vector is input in trained support vector machines human body target classifier by the identification module of controller, discriminate whether that someone is strayed into garage, if being judged as, someone is strayed into garage, the emergency stop switch movement of controller control at this time, garage stops movement, while sound-light alarm is alarmed, long-range GSM message alarming for help, detection precisely, can reduce personal injury to the maximum extent.
Description
Technical field
The invention belongs to stereo garage technical fields, and in particular to a kind of anti-people of mechanical garage is strayed into detection method.
Background technique
With the continuous improvement of domestic social and economic level, family's vehicle, public vehicles owning amount be skyrocketed through, by
In the end of the year 2016, national car ownership is up to 1.94 hundred million, wherein the car ownership in 49 cities is more than million, vehicle multidigit
Less, be the problem of parking difficulty it is more and more prominent, bring huge business opportunity and wide market to Mechanical parking system industry.I
State's mechanical three-dimensional parking device industry also enters Rapid development stage, the country and mechanical stereo garage phase from the starting stage
The device manufacturer of pass has developed to several hundred.From the point of view of the market scale by mechanical stereo garage industry in 2017, entirely
Kuomintang-Communist installation mechanical parking space 617386, increase by 5.2% on a year-on-year basis, domestic new projects 2079 (wherein contain automobile
Special elevator), gross sales amount is 1109223.36 ten thousand yuan.
As what mechanical garage used popularizes, associated accident is also increasing, and accident occurs in order to prevent, setting
Some safeguard procedures.Although someone's vehicle is strayed into the regulation of detection device in mechanical garage standard, substantially light beam is hidden
There is blind area or out-of-service time section in gear type, including infrared or laser scan type.As shown in Figure 1, in laser or infrared beam
Scan less than there are check frequencies in region, even if there is personnel to be strayed into the region, detection device can not also detect that personnel are strayed into
Behavior, and increasing scanning light beam quantity can be such that detection device cost obviously increases, and cannot be fully solved light beam scan blind spot
Problem；Second is that laser or infrared beam are blocked by automobile at this time if personnel are to be strayed into garage at vehicles while passing workspace, inspection
Failure of apparatus is surveyed, can not also detect that the personnel in automobile disengaging this period of garage are strayed into behavior, there are out-of-service time sections
Problem, this be also in December, 2,017 one woman of Jiangsu Prov. People's Hospital stereo garage bow to see the mobile phone and be strayed into the accident of being wounded
One of the main reasons, the victim are strayed into garage when vehicle is outputed, at this time the laser at garage rolling screen door or infrared detection system
System is not acted upon by occlusion, is failed the woman of detection identification in time and is strayed into garage behavior and stop garage, picks up the car with normal
Under the lifting platform of state arrive negative one layer, after by the equipment " wounding " of the vehicle of automatic running and carrying vehicle.
Summary of the invention
The purpose of the present invention is to provide a kind of anti-people of mechanical garage to be strayed into detection method, solves mechanical in the prior art
It is substantially that light beam blocks type, including infrared or laser scan type that the anti-people in formula garage, which is strayed into detection device, and there are blind area or failures
The technical issues of period.
In order to solve the above-mentioned technical problem the present invention, adopts the following technical scheme that
The anti-people of mechanical garage is strayed into detection method, includes the following steps:
1) it, establishes detection system: including controller, emergency stop switch, and being mounted on mechanical garage vehicles while passing mouth just
The camera of top is connected between emergency stop switch, camera and controller by conducting wire, includes characteristic extracting module in controller
And identification module；
2), camera in real time shoots mechanical garage vehicles while passing mouth, and by the delivery of video of shooting to control
Device, the characteristic extracting module of controller extract the HOG feature of current video frame；
3) the HOG feature of current video frame, is collected, and is incorporated into final feature vector and is used for classification；
4), using the INRIA data set comprising different types of human body target picture as support vector machines (SVM) training
The database of study, the HOG feature and corresponding label, label for extracting the positive negative sample of database are+1 or -1, are input to support
It is trained study in vector machine, obtains the classifier based on human body target detection identification.
5) Classification and Identification, is carried out using feature vector of the classifier to current video frame, judges whether that someone is strayed into vehicle
Library；
If 6) be judged as that someone is strayed into garage, the emergency stop switch movement of controller control at this time, garage stops movement, simultaneously
Carry out sound-light alarm, voice is pacified, long-range GSM message alarming for help.
The present invention is strayed into the personal injury and equipment damage problem that may cause for mechanical garage personnel, has developed one
Set mechanical garage is anti-to be strayed into early warning system, and farthest to reduce equipment cost, system is existing using current existing garage
Monitor video image develops a set of personnel and is strayed into Activity recognition software, and with garage system emergency stop switch, live acousto-optic report
Alert system connection issues emergent stop signal stop garage when the personnel of identifying are strayed into behavior, at the same carry out live sound-light alarm,
Field speech, which is pacified, rescues the personnel for being strayed into garage with long-range GSM message alarming for help, calling rescue personnel, and detection is accurate,
Personal injury can be reduced to the maximum extent.
Compared with other character description methods, HOG has many good qualities.Firstly, since HOG is the local grid in image
It is operated on unit, so it can keep good invariance to image geometry and optical deformation, both deformation only can
It appears on bigger space field.Secondly, returning in thick airspace sampling, fine direction sampling and stronger indicative of local optical
Under the conditions of one change etc., as long as pedestrian is generally able to maintain erect posture, pedestrian can be allowed there are some subtle limbs dynamic
Make, these subtle movements can be ignored without influencing detection effect.Therefore, HOG feature is particularly suitable for doing in image
Human body target detection, can effectively solve the problem that on-site supervision image irradiation condition in garage is poor, and contrast is low, human body target dress ornament becomes
The difficulties such as change, attitudes vibration, the randomness of human motion and randomness.
It is further improved, the HOG feature that current video frame is extracted in the step 2) includes the following steps:
2.1) detection window, is set, greyscale transform process is carried out to the video frame of acquisition；
2.2) normalization of color space, is carried out to input picture using Gramma correction method；
2.3) gradient of each pixel of image, is calculated；
2.4), creating unit lattice construct gradient orientation histogram for each cell；
2.5), that the combination of multiple cells is blocking, the gradient orientation histogram of block is normalized；
2.6) it, collects all pieces in detection space of HOG feature and forms the HOG feature vector for indicating the video frame.
It is further improved, further includes that voice pacifies module, when someone is strayed into garage, controller control voice pacifies module
Work, pacifies trapped person.It is the separately installed module of system that field speech, which pacifies module, passes through serial ports with controller
Communication interface connection, when the personnel of detecting are strayed into behavior, send instructions to voice and pacifies submodule, be allowed to provide voice automatically
Prompt is pacified, the mood of personnel's impatience is strayed into mitigate, is allowed to bear with receiving rescue, avoids being strayed into personnel and hastily voluntarily takes off
Secondary injury caused by tired.
It is further improved, calculates the gradient of image abscissa and ordinate direction, and calculate each location of pixels accordingly
Gradient direction value；Derivation operations can not only capture profile, the shadow and some texture informations, moreover it is possible to the shadow that further weakened light shines
It rings.The gradient calculation method of pixel is as follows at some described pixel (x, y):
3.1) convolution algorithm, is done to image with [- 1,0,1] gradient operator, obtains the direction x (horizontal direction, to be positive to the right
Direction) gradient component Gx(x,y)；
3.2), then [1,0, -1] is usedTGradient operator does convolution algorithm to image, obtain the direction y (vertical direction, with to
Upper is positive direction) gradient component Gy(x,y)；
3.3) gradient magnitude and the direction of the pixel, are calculated according to the following formula:
Gx(x, y)=H (x+1, y)-H (x-1, y)；
Gy(x, y)=H (x, y+1)-H (x, y-1)；
Wherein, Gx(x,y)、Gy(x, y) and H (x, y) respectively indicate the level side in image at some pixel (x, y)
To gradient, vertical gradient and pixel value；
Then the gradient magnitude G (x, y) at the pixel (x, y) and gradient direction α (x, y) are respectively as follows:
α (x, y)=tan-1[Gy(x,y)/Gx(x,y)]。
It is further improved, the creating unit lattice, the method for constructing each cell building histogram of gradients is as follows:
4.1) several cells, are divided the image into, each cell includes multiple pixels, the picture element matrix of n*n is constituted,
n≥3；
4.2) gradient information of this n*n pixel, m >=5, are counted using the histogram of m bin；
4.2.1), 0~180 ° of the gradient direction range of each cell is divided into m equal portions direction block, each direction for 180 °
Block corresponds to an angular interval；
4.2.2), according to the respective gradient direction α (x, y) of n*n pixel in cell, judge that each pixel falls into correspondence
Direction block in, calculate all directions block in pixel quantity, to pixel gradient direction each in cell in histogram
It is weighted projection (being mapped to fixed angular range), weight is calculated according to the gradient magnitude G (x, y) of the pixel, is obtained
The gradient orientation histogram of the cell, i.e. the m dimensional feature vector of the cell.
It is further improved, the variation of the variation and foreground-background contrast shone due to local light, so that gradient intensity
Variation range is very big.This just needs to normalize gradient intensity.Normalization can be further to illumination, shade and edge
It is compressed.The combination of all cells is blocking, and the method being normalized on block is as follows: each unit lattice are combined into
Big, coconnected piece of space, then in a block feature vector of all cells be together in series just obtain the block HOG it is special
Sign；Multiple pieces are that mutual is overlapped, then the feature of a cell can repeatedly be appeared in different results last feature to
In amount；Block descriptor (vector) after normalization is known as HOG descriptor.
Compared with prior art, the beneficial effects of the present invention are:
The present invention is strayed into the personal injury and equipment damage problem that may cause for mechanical garage personnel, has developed one
Set mechanical garage is anti-to be strayed into early warning system, and farthest to reduce equipment cost, system is existing using current existing garage
Monitor video image develops a set of personnel and is strayed into Activity recognition software, and with garage system emergency stop switch, live acousto-optic report
Alert system connection issues emergent stop signal stop garage when the personnel of identifying are strayed into behavior, at the same carry out live sound-light alarm,
Field speech, which is pacified, rescues the personnel for being strayed into garage with long-range GSM message alarming for help, calling rescue personnel.
By extracting the HOG feature of video frame, HOG feature vector is input to trained support vector machines by identification module
In human body target classifier, discriminate whether that someone is strayed into garage, this method efficiently solves garage on-site supervision image light
The human bodies such as, contrast low, human body target dress ornament variation, attitudes vibration, the randomness of human motion and randomness poor according to condition
Difficulties in target detection.
Detailed description of the invention
Fig. 1 is that infrared in the prior art or laser scan type people's vehicle is strayed into structure of the detecting device schematic diagram.
Fig. 2 is present system entirety frame.
Fig. 3 is the flow chart that the HOG feature of current video frame is extracted in the present invention.
Specific embodiment
To keep the purpose of the present invention and technical solution clearer, below in conjunction with the embodiment of the present invention to skill of the invention
Art scheme is clearly and completely described.
The anti-people of mechanical garage is strayed into detection method, includes the following steps:
1) it, establishes detection system: including controller, emergency stop switch, and being mounted on mechanical garage vehicles while passing mouth just
The camera of top is connected between emergency stop switch, camera and controller by conducting wire, includes characteristic extracting module in controller
And identification module；
2), camera in real time shoots mechanical garage vehicles while passing mouth, and by the delivery of video of shooting to control
Device, the characteristic extracting module of controller extract the HOG feature of current video frame；
3) the HOG feature of current video frame, is collected, and is incorporated into final feature vector and is used for classification；
4), using the INRIA data set comprising different types of human body target picture as support vector machines (SVM) training
The database of study extracts the HOG feature and corresponding label (+1 or -1) of the positive negative sample of database, is input to support vector machines
In be trained study, obtain one based on human body target detection identification classifier.
5) Classification and Identification, is carried out using feature vector of the classifier to current video frame, judges whether that someone is strayed into vehicle
Library；
If 6) be judged as that someone is strayed into garage, the emergency stop switch movement of controller control at this time, garage stops movement, simultaneously
Carry out sound-light alarm, voice is pacified, long-range GSM message alarming for help.
The present invention is strayed into the personal injury and equipment damage problem that may cause for mechanical garage personnel, has developed one
Set mechanical garage is anti-to be strayed into early warning system, and farthest to reduce equipment cost, system is existing using current existing garage
Monitor video image develops a set of personnel and is strayed into Activity recognition software, and with garage system emergency stop switch, live acousto-optic report
Alert system connection issues emergent stop signal stop garage when the personnel of identifying are strayed into behavior, at the same carry out live sound-light alarm,
Field speech, which is pacified, rescues the personnel for being strayed into garage with long-range GSM message alarming for help, calling rescue personnel, system entirety frame
Frame is as shown in Figure 2.
Compared with other character description methods, HOG has many good qualities.Firstly, since HOG is the local grid in image
It is operated on unit, so it can keep good invariance to image geometry and optical deformation, both deformation only can
It appears on bigger space field.Secondly, returning in thick airspace sampling, fine direction sampling and stronger indicative of local optical
Under the conditions of one change etc., as long as pedestrian is generally able to maintain erect posture, pedestrian can be allowed there are some subtle limbs dynamic
Make, these subtle movements can be ignored without influencing detection effect.Therefore, HOG feature is particularly suitable for doing in image
Human body target detection.By extracting the HOG feature of video frame, HOG feature vector is input to trained by identification module
In support vector machines human body target classifier, differentiate to whether someone is strayed into garage.Efficiently solve garage scene
Monitoring image illumination condition is poor, contrast is low, the variation of human body target dress ornament, attitudes vibration, human motion randomness and with
Difficulties in the human bodies target detection such as machine.
In the present embodiment, the HOG feature that current video frame is extracted in the step 2) includes the following steps, such as Fig. 3 institute
Show:
2.1) detection window, is set, greyscale transform process is carried out to the video frame of acquisition；
2.2) normalization of color space, is carried out to input picture using Gramma correction method；
2.3) gradient of each pixel of image, is calculated；
2.4), creating unit lattice construct gradient orientation histogram for each cell；
2.5), that the combination of multiple cells is blocking, the gradient orientation histogram of block is normalized；
2.6) it, collects all pieces in detection space of HOG feature and forms the HOG feature vector for indicating the video frame.
It in the present embodiment, further include that voice pacifies module, when someone is strayed into garage, controller control voice pacifies mould
Block work, pacifies trapped person.It is the separately installed module of system that field speech, which pacifies module, passes through string with controller
The connection of port communications interface, when the personnel of detecting are strayed into behavior, send instructions to voice and pacifies submodule, be allowed to provide language automatically
Sound pacifies prompt, to mitigate the mood for being strayed into personnel's impatience, is allowed to bear with receiving rescue, avoids being strayed into personnel hastily voluntarily
Secondary injury caused by getting rid of poverty.
In the present embodiment, the gradient calculation method of pixel is as follows at some described pixel (x, y):
4.1) convolution algorithm, is done to image with [- 1,0,1] gradient operator, obtains the direction x (horizontal direction, to be positive to the right
Direction) gradient component Gx(x,y)；
4.2) convolution algorithm then, is done to image with [1,0, -1] T gradient operator, obtain the direction y (vertical direction, with to
Upper is positive direction) gradient component Gy(x,y)；
4.3) gradient magnitude and the direction of the pixel, are calculated according to the following formula:
Gx(x, y)=H (x+1, y)-H (x-1, y)；
Gy(x, y)=H (x, y+1)-H (x, y-1)；
Wherein, Gx(x,y)、Gy(x, y) and H (x, y) respectively indicate the level side in image at some pixel (x, y)
To gradient, vertical gradient and pixel value；
Then the gradient magnitude G (x, y) at the pixel (x, y) and gradient direction α (x, y) are respectively as follows:
α (x, y)=tan-1[Gy(x,y)/Gx(x,y)]。
In the present embodiment, the creating unit lattice, the method for constructing each cell building histogram of gradients are as follows:
5.1) several cells, are divided the image into, each cell includes multiple pixels, constitutes the picture element matrix of 6*6；
In other embodiments, the quantity of pixel can be 8*8 or 16*16 etc. in each cell.
5.2) gradient information of this 6*6 pixel, is counted using the histogram of 9 bin；
5.2.1), 0~180 ° of the gradient direction range of each cell is divided into 9 equal portions direction blocks, each direction for 180 °
Corresponding one 20 ° of the angular interval of block；
5.2.2), according to the respective gradient direction of 6*6 pixel in cell, judge that each pixel falls into corresponding direction
In block, the quantity of pixel in all directions block is calculated, pixel each in cell is added in histogram with gradient direction
Power projection (being mapped to fixed angular range), weight are calculated according to the gradient magnitude G (x, y) of the pixel, then obtain the list
The gradient orientation histogram of first lattice, i.e. 9 dimensional feature vectors of the cell.
In the present embodiment, described that the combination of all cells is blocking, it include 3*3 cell in each piece, block is enterprising
The method of row normalized is as follows: each unit lattice being combined into big, coconnected piece of space, then all lists in a block
The feature vector of first lattice, which is together in series, just obtains the HOG feature of the block；Multiple pieces are that mutual is overlapped, then the spy of a cell
Sign can repeatedly be appeared in last feature vector with different results；Block descriptor (vector) after normalization is known as HOG
Descriptor.
Block is characterized by three parameters: the number of cell in each piece, the number of pixel in each cell, every
The histogram number of active lanes of a cell.Such as: the optimal parameter of pedestrian detection is arranged so that 3 × 3 cells/block, 6 × 6 pictures
Element/cell, 9 histogram bin.The then intrinsic dimensionality of the block are as follows: 3*3*9.
Do not done in the present invention illustrate be the prior art or can be realized by the prior art, and the present invention
Described in specific implementation case be only preferable case study on implementation of the invention, practical range not for the purpose of limiting the invention.
Equivalent changes and modifications made by i.e. all contents according to scope of the present invention patent all should be used as technology scope of the invention.
Claims (6)
1. the anti-people of mechanical garage is strayed into detection method, which comprises the steps of:
1) it, establishes detection system: including controller, emergency stop switch, and being mounted on right above mechanical garage vehicles while passing mouth
Camera, connected between emergency stop switch, camera and controller by conducting wire, include characteristic extracting module and knowledge in controller
Other module；
2), camera in real time shoots mechanical garage vehicles while passing mouth, and by the delivery of video of shooting to controller, control
The characteristic extracting module of device processed extracts the HOG feature of current video frame；
3) the HOG feature of current video frame, is collected, and is incorporated into final feature vector and is used for classification；
4), the number using the INRIA data set comprising different types of human body target picture as support vector machines training study
According to library, the HOG feature of the extraction positive negative sample of database and corresponding label are input in support vector machines and are trained study,
Obtain the classifier based on human body target detection identification；
5) Classification and Identification, is carried out using feature vector of the classifier to current video frame, judges whether that someone is strayed into garage；
If 6) be judged as that someone is strayed into garage, the emergency stop switch movement of controller control at this time, garage stops movement, carries out simultaneously
Sound-light alarm, voice pacify, long-range GSM message alarming for help.
2. the anti-people of mechanical garage according to claim 1 is strayed into detection method, which is characterized in that mentioned in the step 2)
The HOG feature of current video frame is taken to include the following steps:
2.1) detection window, is set, greyscale transform process is carried out to the video frame of acquisition；
2.2) normalization of color space, is carried out to input picture using Gramma correction method；
2.3) gradient of each pixel of image, is calculated；
2.4), creating unit lattice construct gradient orientation histogram for each cell；
2.5), that the combination of multiple cells is blocking, the gradient orientation histogram of block is normalized；
2.6) it, collects all pieces in detection space of HOG feature and forms the HOG feature vector for indicating the video frame.
3. the anti-people of mechanical garage according to claim 2 is strayed into detection method, which is characterized in that further include that voice is pacified
Module, when someone is strayed into garage, controller control voice pacifies module work, pacifies trapped person.
4. the anti-people of mechanical garage according to claim 2 is strayed into detection method, which is characterized in that some described pixel
The gradient calculation method of pixel is as follows at point (x, y):
4.1) convolution algorithm, is done to image with [- 1,0,1] gradient operator, obtains the direction x (horizontal direction, to the right for pros
To) gradient component Gx(x,y)；
4.2) convolution algorithm then, is done to image with [1,0, -1] T gradient operator, obtains the direction y (vertical direction, to be upwards
Positive direction) gradient component Gy(x,y)；
4.3) gradient magnitude and the direction of the pixel, are calculated according to the following formula:
Gx(x, y)=H (x+1, y)-H (x-1, y)；
Gy(x, y)=H (x, y+1)-H (x, y-1)；
Wherein, Gx(x,y)、Gy(x, y) and H (x, y) respectively indicate the ladder of the horizontal direction in image at some pixel (x, y)
Degree, vertical gradient and pixel value；
Then the gradient magnitude G (x, y) at the pixel (x, y) and gradient direction α (x, y) are respectively as follows:
5. the anti-people of mechanical garage according to claim 2 is strayed into detection method, which is characterized in that the creating unit
Lattice, the method for constructing each cell building histogram of gradients are as follows:
4.1) several cells, are divided the image into, each cell includes multiple pixels, constitute the picture element matrix of n*n, n >=
3；
4.2) gradient information of this n*n pixel, m >=5, are counted using the histogram of m bin；
4.2.1), 0~180 ° of the gradient direction range of each cell is divided into m equal portions direction block, each direction block pair for 180 °
Answer an angular interval；
4.2.2), according to the respective gradient direction α (x, y) of n*n pixel in cell, judge that each pixel falls into corresponding side
Into block, the quantity of pixel in all directions block is calculated, pixel each in cell is carried out in histogram with gradient direction
Weighted projection, weight are calculated according to the gradient magnitude G (x, y) of the pixel, obtain the gradient orientation histogram of the cell, i.e.,
The m dimensional feature vector of the cell.
6. the anti-people of mechanical garage according to claim 2 is strayed into detection method, which is characterized in that described by all units
Lattice combination is blocking, and the method being normalized on block is as follows: each unit lattice are combined into big, space is coconnected
Block, then the feature vector of all cells is together in series and just obtains the HOG feature of the block in a block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810706894.4A CN108960131A (en) | 2018-06-29 | 2018-06-29 | The anti-people of mechanical garage is strayed into detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810706894.4A CN108960131A (en) | 2018-06-29 | 2018-06-29 | The anti-people of mechanical garage is strayed into detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108960131A true CN108960131A (en) | 2018-12-07 |
Family
ID=64484761
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810706894.4A Pending CN108960131A (en) | 2018-06-29 | 2018-06-29 | The anti-people of mechanical garage is strayed into detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108960131A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109727426A (en) * | 2019-01-23 | 2019-05-07 | 南京市特种设备安全监督检验研究院 | A kind of mechanical garage personnel are strayed into monitoring identification early warning system and detection method |
| CN110482390A (en) * | 2019-08-29 | 2019-11-22 | 南京市特种设备安全监督检验研究院 | A kind of escalator/moving sidewalk angle personnel are strayed into monitoring identification early warning system and monitoring recognition methods |
| CN111845736A (en) * | 2020-06-16 | 2020-10-30 | 江苏大学 | Vehicle collision early warning system triggered by distraction monitoring and control method |
| CN114387720A (en) * | 2022-01-12 | 2022-04-22 | 深圳市春盛海科技有限公司 | Method for identifying safety monitoring image of iron rolling door matched with AI (Artificial Intelligence) operation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102955461A (en) * | 2011-08-31 | 2013-03-06 | 鸿富锦精密工业（深圳）有限公司 | System and method for monitoring mechanical-type parking spaces |
| CN205172115U (en) * | 2015-11-18 | 2016-04-20 | 广东明和智能设备有限公司 | Safety protection system for stereo garage |
| CN105667425A (en) * | 2016-03-01 | 2016-06-15 | 北京爱国小男孩科技有限公司 | Automobile antitheft and intra-automobile personnel danger alarm device and method |
| CN205772562U (en) * | 2016-07-07 | 2016-12-07 | 上海三洋电梯有限公司 | There is voice and pacify the elevator station reporting device of function |
| CN107038842A (en) * | 2017-05-23 | 2017-08-11 | 西安科技大学 | In-car is by lock personal security warning system |
| CN107720466A (en) * | 2017-11-20 | 2018-02-23 | 上海贝思特电气有限公司 | Elevator voice pacifies method and elevator voice pacifies circuit |
| CN207216639U (en) * | 2017-09-12 | 2018-04-10 | 江南嘉捷电梯股份有限公司 | A kind of trolley autostairs personnel are strayed into detection means |
-
2018
- 2018-06-29 CN CN201810706894.4A patent/CN108960131A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102955461A (en) * | 2011-08-31 | 2013-03-06 | 鸿富锦精密工业（深圳）有限公司 | System and method for monitoring mechanical-type parking spaces |
| CN205172115U (en) * | 2015-11-18 | 2016-04-20 | 广东明和智能设备有限公司 | Safety protection system for stereo garage |
| CN105667425A (en) * | 2016-03-01 | 2016-06-15 | 北京爱国小男孩科技有限公司 | Automobile antitheft and intra-automobile personnel danger alarm device and method |
| CN205772562U (en) * | 2016-07-07 | 2016-12-07 | 上海三洋电梯有限公司 | There is voice and pacify the elevator station reporting device of function |
| CN107038842A (en) * | 2017-05-23 | 2017-08-11 | 西安科技大学 | In-car is by lock personal security warning system |
| CN207216639U (en) * | 2017-09-12 | 2018-04-10 | 江南嘉捷电梯股份有限公司 | A kind of trolley autostairs personnel are strayed into detection means |
| CN107720466A (en) * | 2017-11-20 | 2018-02-23 | 上海贝思特电气有限公司 | Elevator voice pacifies method and elevator voice pacifies circuit |
Non-Patent Citations (3)
| Title |
|---|
| 徐渊 等: "结合SVM分类器与HOG特征提取的行人检测", 《计算机工程》 * |
| 武光利，王锦沼: "基于HOG特征与SVM分类器的行人检测研究", 《中国有线电视》 * |
| 赵伟: "基于机器视觉的列检库作业人员安全检测系统研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109727426A (en) * | 2019-01-23 | 2019-05-07 | 南京市特种设备安全监督检验研究院 | A kind of mechanical garage personnel are strayed into monitoring identification early warning system and detection method |
| CN110482390A (en) * | 2019-08-29 | 2019-11-22 | 南京市特种设备安全监督检验研究院 | A kind of escalator/moving sidewalk angle personnel are strayed into monitoring identification early warning system and monitoring recognition methods |
| CN111845736A (en) * | 2020-06-16 | 2020-10-30 | 江苏大学 | Vehicle collision early warning system triggered by distraction monitoring and control method |
| CN114387720A (en) * | 2022-01-12 | 2022-04-22 | 深圳市春盛海科技有限公司 | Method for identifying safety monitoring image of iron rolling door matched with AI (Artificial Intelligence) operation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960131A (en) | The anti-people of mechanical garage is strayed into detection method | |
| CN110348312A (en) | A kind of area video human action behavior real-time identification method | |
| CN113516076B (en) | Attention mechanism improvement-based lightweight YOLO v4 safety protection detection method | |
| CN101968848B (en) | Video monitoring method and system and video monitoring alarm system | |
| CN111310556A (en) | Drowning prevention safety supervision system based on primary and middle school student area and monitoring method thereof | |
| CN112183472A (en) | Method for detecting whether test field personnel wear work clothes or not based on improved RetinaNet | |
| CN107103300B (en) | Off-duty detection method and system | |
| CN202771441U (en) | Face recognition monitoring system | |
| CN112634561A (en) | Safety alarm method and system based on image recognition | |
| CN108898098A (en) | Early stage video smoke detection method based on monitor supervision platform | |
| CN111908288A (en) | TensorFlow-based elevator safety system and method | |
| CN115512387A (en) | Construction site safety helmet wearing detection method based on improved YOLOV5 model | |
| CN116189052A (en) | Security method, system, intelligent terminal and storage medium based on video stream analysis | |
| CN110222216B (en) | Multifunctional intelligent defense system | |
| CN111539264A (en) | Ship flame detection positioning system and detection positioning method | |
| KR20230097854A (en) | Method and system for recognizing dangerous behavior of workers in power plant | |
| CN109727426A (en) | A kind of mechanical garage personnel are strayed into monitoring identification early warning system and detection method | |
| CN113052139A (en) | Deep learning double-flow network-based climbing behavior detection method and system | |
| CN111539254A (en) | Target detection method, target detection device, electronic equipment and computer-readable storage medium | |
| CN115171006B (en) | Detection method for automatically identifying person entering electric power dangerous area based on deep learning | |
| CN116052035A (en) | Power plant personnel perimeter intrusion detection method based on convolutional neural network | |
| CN113920535B (en) | Electronic region detection method based on YOLOv5 | |
| CN115830294A (en) | Identification method and device applied to personnel management and control in infrastructure construction operation field | |
| CN108408521A (en) | A kind of big data analysis system and method based on elevator passenger daily behavior | |
| CN112699745A (en) | Method for positioning trapped people on fire scene |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181207 |