CN110163143A - Unlawful practice recognition methods, device and terminal device - Google Patents
Unlawful practice recognition methods, device and terminal device Download PDFInfo
- Publication number
- CN110163143A CN110163143A CN201910414712.0A CN201910414712A CN110163143A CN 110163143 A CN110163143 A CN 110163143A CN 201910414712 A CN201910414712 A CN 201910414712A CN 110163143 A CN110163143 A CN 110163143A
- Authority
- CN
- China
- Prior art keywords
- unlawful practice
- image
- recognized
- images
- recognition methods
- 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
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention is suitable for technical field of image processing, provides a kind of unlawful practice recognition methods, device and terminal device.The unlawful practice recognition methods includes: acquisition images to be recognized；The images to be recognized is pre-processed using thresholding method；Convolutional neural networks model is trained using unlawful practice sample set；The convolutional neural networks model completed based on training identifies pretreated images to be recognized, judges whether there is unlawful practice.Substation field is detected using the unlawful practice recognition methods, the unlawful practice of substation field staff can be found in time, take timely measure, security risk is excluded, improves detection efficiency and accuracy.
Description
Technical field
The invention belongs to technical field of image processing more particularly to a kind of unlawful practice recognition methods, device and terminal to set
It is standby.
Background technique
With the development of national grid, the work such as reconstruction and extension project, daily electric power maintenance and maintenance are more in substation, by
Weakness is realized in some operating personnel's inherently safes, live unlawful practice is caused to take place frequently, for example is crossed over fence, forgotten band safety cap
Equal behaviors.There are huge security risks for similar violation operation, may cause the generation of personal safety accident when serious.Cause
This, finds the unlawful practice of on-site personnel in time and corrects the particularly important of change.
Currently, to there are mainly two types of the detections of operating personnel's unlawful practice in substation: Spot-supervision and monitor video people
Work point analysis and identification.But since both the above mode is carried out judgement identification by people, low efficiency, accuracy is low, real-time is poor,
And a large amount of manpower of waste.
Summary of the invention
In view of this, the embodiment of the invention provides a kind of unlawful practice recognition methods, device and terminal device, to solve
The problem of artificial judgement identification operating personnel's unlawful practice low efficiency, poor accuracy and real-time difference etc. in the prior art.
The first aspect of the embodiment of the present invention provides a kind of unlawful practice recognition methods, comprising:
Obtain images to be recognized；
The images to be recognized is pre-processed using thresholding method；
Convolutional neural networks model is trained using unlawful practice sample set；
The convolutional neural networks model completed based on training identifies pretreated images to be recognized, judges whether
There are unlawful practices.
The second aspect of the embodiment of the present invention provides a kind of unlawful practice identification device, comprising:
Module is obtained, for obtaining images to be recognized；
Preprocessing module, for being pre-processed to the images to be recognized；
Training module, for being trained to convolutional neural networks model；
It identifies judgment module, is identified for pretreated image, judge to whether there is in the images to be recognized
Unlawful practice.
The third aspect of the embodiment of the present invention provides a kind of terminal device, including memory, processor and is stored in
In the memory and the computer program that can run on the processor, when the processor executes the computer program
Realize the step of unlawful practice recognition methods that first aspect of the embodiment of the present invention provides.
The fourth aspect of the embodiment of the present invention provides a kind of computer readable storage medium, the computer-readable storage
Media storage has computer program, and the first aspect such as the embodiment of the present invention is realized when the computer program is executed by processor
The step of unlawful practice recognition methods of offer.
The embodiment of the present invention locates the images to be recognized using thresholding method after obtaining images to be recognized in advance
Reason, is then trained convolutional neural networks model using unlawful practice sample set, the convolution mind finally completed using training
Pretreated images to be recognized is identified through network model, judges whether there is unlawful practice.Utilize above-mentioned violation row
Substation field is detected for recognition methods, can find the unlawful practice of substation field staff in time and prevent to disobey
Rule behavior excludes security risk, improves detection efficiency and accuracy.
Detailed description of the invention
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description is only of the invention some
Embodiment for those of ordinary skill in the art without any creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is the implementation process schematic diagram of unlawful practice recognition methods provided in an embodiment of the present invention；
Fig. 2 is the implementation process schematic diagram of step S102 in Fig. 1；
Fig. 3 is the implementation process schematic diagram of step S1022 in Fig. 2；
Fig. 4 is the implementation process schematic diagram of step S103 in Fig. 1；
Fig. 5 is the interface schematic diagram of display equipment provided in an embodiment of the present invention；
Fig. 6 is the schematic diagram of unlawful practice identification device provided in an embodiment of the present invention；
Fig. 7 is the schematic diagram of terminal device provided in an embodiment of the present invention.
Specific embodiment
In being described below, for illustration and not for limitation, the tool of such as particular system structure, technology etc is proposed
Body details, to understand thoroughly the embodiment of the present invention.However, it will be clear to one skilled in the art that there is no these specific
The present invention also may be implemented in the other embodiments of details.In other situations, it omits to well-known system, device, electricity
The detailed description of road and method, in case unnecessary details interferes description of the invention.
In order to illustrate technical solutions according to the invention, the following is a description of specific embodiments.
Referring to Fig. 1, one embodiment of the invention discloses a kind of unlawful practice recognition methods, comprising:
S101 obtains images to be recognized.
The images to be recognized is the picture in field personnel's course of work.
In some embodiments, the video image of front end photographic device is acquired first, and subchannel passes the video image
Transport to backstage；Then preliminary treatment is carried out to the video image received, image is converted to by different video file formats；Most
Described image is screened afterwards to obtain the images to be recognized.
Substation field is provided with more high-definition cameras for acquiring the video image in substation, and is provided with holder
Lens controller is used to control houselights for controlling the rotation of video camera and the zoom of camera lens and focusing, lamp dimmer,
To guarantee in insufficient light or night can also obtain clearly image.Communication channel presses the video image of acquisition
Contracting, and compressed signal is transmitted to backstage by subchannel, and then the signal received is reintegrated and decompressed.Monitoring service
On the one hand device plays out the video image transmitted, observe substation's operation field working condition at any time for inspection personnel；
The video image transmitted is forwarded in system server using TCP/IP technology by another aspect monitoring server, so as to its into
Row processing analysis.
Since the target of image procossing is image, it is therefore desirable to which the video data that will acquire is converted to the text of picture format
Part, so as to subsequent processing.
Finally the image of video data conversion is screened, removes some fuzzy, duplicate images, improves the effect of identification
Rate and accuracy.
S102 pre-processes the images to be recognized using thresholding method.
Since substation equipment is various, and personnel are in dynamic in substation's operation, it is therefore desirable to image into
Row dividing processing extracts important feature in figure, i.e. personage's behavior and dressing.Image segmentation be exactly in image have special culvert
Justice different zones distinguish, these regions be it is mutually disjoint, each region meets the consistent of specific region.Using
The advantages of thresholding method is split the images to be recognized, thresholding method is to calculate easy, consideration image grayscale
Without considering that other information operation efficiency is also high.The images to be recognized is split using thresholding method, can be emphasized out
The feature for needing to extract accelerates the training speed of convolutional neural networks model and improves accuracy rate.
In some embodiments, with reference to Fig. 2, step S102 includes:
S1021 carries out gray proces to the images to be recognized, obtains gray level image.
The images to be recognized is generally the digital picture based on RGB model, carries out gray proces to it first, will be each
Three color datas of pixel, which calculate, is converted to gray value, then in calculated gray value replacement original image
Each pixel, to obtain gray level image.
In some embodiments, gray level image can also be enhanced after obtaining gray level image.
The images to be recognized can generate noise pollution in transmission and conversion process, and picture quality inevitably drops
It is low.Therefore, picture quality can be improved before analyzing image.Being increased using image improves images to be recognized
Quality, feature interested in image only selectively protrudes, pressed down by the reason of image enhancement does not consider image quality decrease
System does not need the feature extracted, and improving image quality, the intelligibility for improving image, reinforces original image at abundant information amount
Interpretation and recognition effect.
In some embodiments, the modes such as piecewise linear transform or logarithmic function greyscale transformation can be selected to carry out image increasing
By force.
S1022 calculates optimal segmenting threshold based on the gray level image.
The adaptive threshold fuzziness of image is realized it is necessary to find the optimal segmenting threshold of image, its segmentation of different images
Threshold value is different, how suitable segmentation threshold is targetedly found according to different images, is the pass for realizing adaptivenon-uniform sampling
Key.
In some embodiments, iterative method can be selected and calculate optimal segmenting threshold, with reference to Fig. 3, step S1022 may include:
S401 obtains the maximum gradation value Z of the gray level imagemaxWith minimum gradation value Zmin, according to the maximum gray scale
Value ZmaxWith the minimum gradation value ZminCalculate first threshold T0, wherein
S402, according to first threshold T0Gray level image is divided into foreground and background, calculates separately being averaged for the prospect
Gray value ZOWith the average gray value Z of the backgroundB。
S403, according to the average gray value Z of the prospectOWith the average gray value Z of the backgroundBCalculate present threshold value T1,
Wherein
S404, if present threshold value T1With first threshold T0Difference be less than preset error value, then first threshold T0For best threshold
Value；Otherwise, by present threshold value T1Value assign first threshold T0, and step S402 is repeated to step S404.
For example, program implementation process can be with are as follows:
1, the maximum gradation value Z of the gray level image is obtainedmaxWith minimum gradation value Zmin, another i=0, initial threshold
2, according to threshold value T0Gray level image is divided into foreground and background, calculates separately the average gray value Z of the prospectO
With the average gray value Z of the backgroundB。
3, according to the average gray value Z of the prospectOWith the average gray value Z of the backgroundBCalculate threshold value Ti+1,
If 4, Ti+1With TiDifference be less than preset error value, then TiFor optimal threshold；Otherwise, i++ repeats step 2 to step
Rapid 4.
In some embodiments, the average gray value Z for calculating separately the prospectOWith the average gray value of the background
ZB, comprising:
S4021 calculates the grey level histogram of the prospect or the grey level histogram of the background.
Grey level histogram is normalized in S4022, obtains the probability of grey level histogram each single item.
The product of S4023, abscissa and corresponding probability to grey level histogram sum to obtain the average gray of the prospect
Value ZOOr the average gray value Z of the backgroundB。
S1023 carries out adaptivenon-uniform sampling to the gray level image according to the optimal segmenting threshold.
S103 is trained neural network model using unlawful practice sample set.
Neural network (Neural Networks, NN) is wide by a large amount of, simple processing unit (referred to as neuron)
The complex networks system for interconnecting and being formed generally, it reflects many essential characteristics of human brain function, is that a height is multiple
Miscellaneous non-linear dynamic learning system.
Deep learning passes through the network of one or more multilayer, and the information in image is extracted and stored, and with
The increase of the network number of plies, image information is constantly abstracted, the final Structural Characteristics for obtaining image.And convolutional Neural
Network is a kind of comprising convolutional calculation and with the feedforward neural network of depth structure, is the representative algorithm of deep learning.
Convolutional neural networks are to inspire the deep learning framework designed based on biological natural vision Cognition Mechanism, mainly
Including convolutional layer, pond layer, full articulamentum.Wherein, convolutional layer refers to the layer for completing image convolution operation, and convolution operation refers to use
One convolution kernel and image corresponding region carry out convolution and obtain a value, then constantly mobile convolution kernel with seek convolution, with complete
The convolution of pairs of whole image.In neural network, the calculating of convolutional layer is usually directed to except convolution operation concept, further includes depth
With step-length concept, wherein depth determines that the neuron number of the same area, i.e., several convolution kernels carry out convolution behaviour to the same area
Make, step-length refers to the number of the mobile pixel of convolution kernel, as shown in figure 5, depth is to be 7*7 with the height and width of input layer
3, the height and width of two filter Fs ilter, each Filter are 3*3 respectively, and depth is 3, and convolution kernel size is that 3*3 is
Example, illustrates the convolution operation schematic diagram of convolutional layer, leftmost input layer (Input Volume) and first filter
(Filter W0) is calculated, and the first layer of input layer and the first layer of Filter W0 carry out operation, the second layer of input layer
With the second layer progress operation of Filter W0, the third layer of input layer and the third layer progress operation of Filter W0, last three
Layer result adds up, and obtains first matrix of consequence of output layer (Output Volume)；And so on, it is leftmost
Input Volume and second filter (Filter W1) are calculated, and second result of Output Volume is obtained
Matrix.Pond layer is between convolutional layer, for the default block area compresses of upper one layer of input data to be worth at one, thus by
Step compression reduces the quantity of data and deconvolution parameter, reduces over-fitting.Full articulamentum is mainly used for learning, the instruction that will be acquired
Practice integrated distribution formula character representation and be mapped to sample labeling space, to obtain the weight of neural network model.
Wherein, convolutional neural networks model mainly has following characteristics: it is locally connected between network, and not full connection, it is this
Connection feature reduces connection quantity not only to save calculation amount, while also highlighting the local characteristics of feature；The weight of network is total
It enjoys, it is the important feature of convolutional neural networks that weight is shared, it is shared by the weight of local receptor field, a small amount of weight can be used
The feature of entire picture described, reduction training complexity largely；Feature extraction and classification at the same carry out, convolution
Network structure of neural network itself has ability in feature extraction, and the feature of image is stored in each neuron, in forward direction and
During back-propagating, by achieving the purpose that learning characteristic to continuing to optimize for weight；The detector proposed can be weak
Flag data is trained.
Neural network model can be the neural network model obtained based on known image data set pre-training.
It is described that convolutional neural networks model is trained using unlawful practice sample set with reference to Fig. 4 in some embodiments
May include:
S1031 initializes convolutional neural networks model.
Initialization convolutional neural networks model refers to the parameter of initialization convolutional neural networks model, to build initial volume
Product neural network model.The parameter for initializing convolutional neural networks model mainly includes initialization convolutional neural networks model middle layer
Connecting quantity between layer, i.e., the weight on side in convolutional neural networks.The parameter of initialization convolutional neural networks model may be used also
To include the number of iterations, batch processing size, learning rate, the neural network number of plies etc. initialized in convolutional neural networks model.
First unlawful practice sample set is inputted the convolutional neural networks model as training sample and carries out the by S1032
One training.
First unlawful practice sample set is inputted into initial convolutional neural networks model and is iterated training, by preceding to biography
It leads, calculate cost using markup information and cost function, by the ginseng in each layer of backpropagation cost function gradient updating
Number, to adjust the weight of initial convolutional neural networks model, until the loss function of the convolutional neural networks model meets
The condition of convergence obtains the convolutional neural networks model by the first training.In some embodiments, the first unlawful practice sample
The image of concentration can be the image for the typical unlawful practice posed for photograph under simple background, or choose in practical unlawful practice image
The better simply unlawful practice image of the background of choosing.
Second unlawful practice sample set is inputted the convolutional Neural net after the first training as test sample by S1033
Network model is tested, and the convolutional neural networks model of training completion is obtained.
Convolutional neural networks model of the second unlawful practice sample set input after the first training is iterated training,
Convolutional neural networks model after the first training is finely adjusted, the convolutional neural networks model of training completion is obtained, makes
It is detected more suitable for the unlawful practice under actual scene.Figure in some embodiments, in the second unlawful practice sample set
As being the unlawful practice image under substation's real scene.
S104, the convolutional neural networks model completed based on training identifies pretreated images to be recognized, sentences
It is disconnected to whether there is unlawful practice.
Pretreated images to be recognized is inputted the convolutional neural networks model that training is completed to identify, is judged wait know
It whether there is unlawful practice in other image.
In some embodiments, the unlawful practice recognition methods can also include recognition result is transmitted to display equipment into
Row display；When the recognition result is shown there are issuing alarm signal when unlawful practice, while being shown in the display equipment
The corresponding violation image of the unlawful practice.In some embodiments, as described in Figure 5, the information shown in the display equipment can
With include: substation's number, date, with the presence or absence of unlawful practice, unlawful practice image and safety code clause etc..
When recognition result is shown there are alarm signal is issued immediately when unlawful practice, for example, sound-light alarm, short massage notice
Deng, while the images to be recognized that will test out unlawful practice is intuitively shown on the display device, so as to staff
It takes timely measure and unlawful practice is handled.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
Corresponding to unlawful practice recognition methods described in foregoing embodiments, Fig. 6 shows provided in an embodiment of the present invention disobey
Advise the schematic diagram of Activity recognition device 10.For ease of description, only the parts related to this embodiment are shown.
With reference to Fig. 6, which includes: to obtain module 11, for obtaining images to be recognized；Pretreatment
Module 12, for being pre-processed to the images to be recognized；Training module 13, for being instructed to convolutional neural networks model
Practice；It identifies judgment module 14, is identified for pretreated image, judged in the images to be recognized with the presence or absence of in violation of rules and regulations
Behavior.
In some embodiments, the acquisition module 11 may include that video data acquiring unit, converting unit and screening are single
Member.Video data acquiring unit is used to obtain the video data of leading portion photographic device acquisition；Converting unit is mainly used for video
Data are converted to image, facilitate subsequent processing；Screening unit is removed some unclear for screening to the image being converted to
Chu, duplicate image.
In some embodiments, the preprocessing module 12 may include gray scale processing unit, optimal threshold solve unit and
Image segmentation unit.Gray scale processing unit is used to carry out gray proces to images to be recognized, and images to be recognized is converted to gray scale
Image；Optimal threshold solves the optimal threshold that unit is used to calculate image, such as can be by iterative method or according to probability density
Function, criterion function combination threshold value equation acquire the optimal threshold of image.Image segmentation unit is used for according to the best threshold acquired
Value is split image.In some embodiments, preprocessing module can also include enhancement unit, in gray scale processing unit
Gray level image is enhanced after images to be recognized is converted to gray level image.
In some embodiments, the training module 13 may include the first model training unit and the second model training list
Member.First model training unit be used for using the first unlawful practice sample set as training sample input convolutional neural networks model into
Row iteration training obtains the convolution by the first training until the loss function of convolutional neural networks model meets the condition of convergence
Neural network model；Second model training unit is used to pass through first for the second unlawful practice sample set as training sample input
Trained convolutional neural networks model is iterated training, the convolutional neural networks model that the training being adjusted is completed.
In some embodiments, the unlawful practice identification device can also include: display module 15, for showing identification knot
Fruit, and the display violation image corresponding with the unlawful practice when there is unlawful practice；Alarm module 16 has in violation of rules and regulations for working as
It alarms when behavior.In some embodiments, alarm module 16 may include acousto-optic warning unit, short message transmission unit acousto-optic report
Alert unit is for issuing sound-light alarm；Short message transmission unit is used to for the information of unlawful practice being sent to the mobile phone of related personnel
In.In some embodiments, alarm module 16 can also include reporting unit；Reporting unit is used for the information of unlawful practice and disobeys
The video of certain period of time is uploaded to higher level command centre where rule behavior.
With reference to Fig. 7, one embodiment of the invention discloses a kind of terminal device, and in the present embodiment, terminal device 7 includes:
Processor 70, memory 71 and it is stored in the computer program that can be run in the memory 71 and on the processor 70
72.The processor 70 realizes each step in above-mentioned unlawful practice recognition methods embodiment when executing the computer program 72
Suddenly, for example, step S101 to S104 as shown in Figure 1.Alternatively, realization when the processor 70 executes the computer program 72
The function of each module/unit in above-mentioned unlawful practice identification device embodiment, such as module 11 shown in Fig. 6 is to the function of module 16
Energy.
Illustratively, the computer program 72 can be divided into one or more program module/units, one
Or multiple program module/units are stored in the memory 71, and are executed by the processor 70, to complete this Shen
Please.One or more of program module/units can be the series of computation machine program instruction section that can complete specific function,
The instruction segment is for describing implementation procedure of the computer program 72 in the terminal device 7.For example, the computer journey
Sequence 72, which can be divided into, obtains module 11, for obtaining images to be recognized；Preprocessing module 12, for the figure to be identified
As being pre-processed；Training module 13, for being trained to convolutional neural networks model；Judgment module 14 is identified, for pre-
Treated, and image is identified, is judged in the images to be recognized with the presence or absence of unlawful practice.
The terminal device can be mobile phone, tablet computer etc. and calculate equipment.The terminal device may include, but not only limit
In processor 70, memory 71.It will be understood by those skilled in the art that Fig. 7 is only the example of terminal device 7, do not constitute
Restriction to terminal device 7 may include perhaps combining certain components or different than illustrating more or fewer components
Component, such as the terminal device 7 can also include input-output equipment, network access equipment, bus etc..
Alleged processor 70 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
The memory 71 can be the internal storage unit of the terminal device 7, such as the hard disk or interior of terminal device 7
It deposits.The memory 71 is also possible to the External memory equipment of the terminal device 7, such as be equipped on the terminal device 7
Plug-in type hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card dodge
Deposit card (Flash Card) etc..Further, the memory 71 can also both include the storage inside list of the terminal device 7
Member also includes External memory equipment.The memory 71 is for storing 7 institute of the computer program 72 and the terminal device
Other programs and data needed.The memory 71 can be also used for temporarily storing the number that has exported or will export
According to.
The embodiment of the invention also provides a kind of computer readable storage medium, computer-readable recording medium storage has meter
Calculation machine program realizes the step in each embodiment described in recognition methods, such as Fig. 1 institute when computer program is executed by processor
The step S101 to step S104 shown.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, is not described in detail or remembers in some embodiment
The part of load may refer to the associated description of other embodiments.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, the specific application and design constraint depending on technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
Scope of the present application.
In embodiment provided herein, it should be understood that disclosed elastic distribution network restoration power method for improving,
Device and terminal device, may be implemented in other ways.For example, lifting device embodiment described above is only to show
Meaning property, for example, in addition the division of the module or unit, only a kind of logical function partition can have in actual implementation
Division mode, such as multiple units or components can be combined or can be integrated into another system or some features can be with
Ignore, or does not execute.Another point, shown or discussed mutual coupling or direct-coupling or communication connection can be logical
Some interfaces are crossed, the INDIRECT COUPLING or communication connection of device or unit can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated module/unit be realized in the form of SFU software functional unit and as independent product sale or
In use, can store in a computer readable storage medium.Based on this understanding, the application realizes above-mentioned implementation
All or part of the process in example method, can also instruct relevant hardware to complete, the meter by computer program
Calculation machine program can be stored in a computer readable storage medium, the computer program when being executed by processor, it can be achieved that on
The step of stating each embodiment of the method.Wherein, the computer program includes computer program code, the computer program generation
Code can be source code form, object identification code form, executable file or certain intermediate forms etc..The computer-readable medium
It may include: any entity or device, recording medium, USB flash disk, mobile hard disk, magnetic that can carry the computer program code
Dish, CD, computer storage, read-only memory (ROM, Read-Only Memory), random access memory (RAM,
Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium etc..It should be noted that described
The content that computer-readable medium includes can carry out increasing appropriate according to the requirement made laws in jurisdiction with patent practice
Subtract, such as in certain jurisdictions, according to legislation and patent practice, computer-readable medium do not include be electric carrier signal and
Telecommunication signal.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (11)
1. a kind of unlawful practice recognition methods characterized by comprising
Obtain images to be recognized；
The images to be recognized is pre-processed using thresholding method；
Convolutional neural networks model is trained using unlawful practice sample set；
The convolutional neural networks model completed based on training identifies pretreated images to be recognized, judges whether there is
Unlawful practice.
2. unlawful practice recognition methods as described in claim 1, which is characterized in that described to utilize unlawful practice sample set to volume
Product neural network model is trained, comprising:
Initialize convolutional neural networks model；
First unlawful practice sample set is inputted into the convolutional neural networks model as training sample and carries out the first training；
The convolutional neural networks model that second unlawful practice sample set inputs after the first training as test sample is carried out
Test obtains the convolutional neural networks model of training completion.
3. unlawful practice recognition methods as described in claim 1, which is characterized in that it is described using thresholding method to it is described to
Identification image is pre-processed, comprising:
Gray proces are carried out to the images to be recognized, obtain gray level image；
Optimal segmenting threshold is calculated based on the gray level image；
Adaptivenon-uniform sampling is carried out to the gray level image according to the optimal segmenting threshold.
4. unlawful practice recognition methods as claimed in claim 3, which is characterized in that described to calculate best point based on gray level image
Cut threshold value, comprising:
S401 obtains the maximum gradation value Z of the gray level imagemaxWith minimum gradation value Zmin, according to the maximum gradation value Zmax
With the minimum gradation value ZminCalculate first threshold T0, wherein
S402, according to first threshold T0Gray level image is divided into foreground and background, calculates separately the average gray value of the prospect
ZOWith the average gray value Z of the backgroundB；
S403, according to the average gray value Z of the prospectOWith the average gray value Z of the backgroundBCalculate present threshold value T1, wherein
S404, if present threshold value T1With first threshold T0Difference be less than preset error value, then first threshold T0For optimal threshold,
Export optimal threshold；Otherwise, by present threshold value T1Value assign first threshold T0, and step S402 is repeated to step S404.
5. unlawful practice recognition methods as claimed in claim 4, which is characterized in that the average gray for calculating separately prospect
Value ZOWith the average gray value Z of the backgroundBInclude:
Calculate the grey level histogram of the prospect or the grey level histogram of the background；
Grey level histogram is normalized, the probability of grey level histogram each single item is obtained；
The product of abscissa and corresponding probability to grey level histogram sums to obtain the average gray value Z of the prospectOOr the back
The average gray value Z of scapeB。
6. unlawful practice recognition methods as described in claim 1, which is characterized in that the unlawful practice recognition methods is also wrapped
It includes:
Recognition result is transmitted to display equipment to show；
When the recognition result is shown there are issuing alarm signal when unlawful practice, while in the display equipment described in display
The corresponding violation image of unlawful practice.
7. such as unlawful practice recognition methods as claimed in any one of claims 1 to 6, which is characterized in that described to obtain figure to be identified
As including:
Subchannel obtains the video data of front end photographic device acquisition；
The video data that will acquire is converted to image；
Described image is screened to obtain the images to be recognized.
8. a kind of unlawful practice identification device characterized by comprising
Module is obtained, for obtaining images to be recognized；
Preprocessing module, for being pre-processed to the images to be recognized；
Training module, for being trained to convolutional neural networks model；
It identifies judgment module, is identified for pretreated image, judged in the images to be recognized with the presence or absence of in violation of rules and regulations
Behavior.
9. unlawful practice identification device as claimed in claim 8, which is characterized in that further include:
Display module, for showing recognition result, and the display violation corresponding with the unlawful practice when there is unlawful practice
Image；
Alarm module, for alarming when there is unlawful practice.
10. a kind of terminal device, which is characterized in that in the memory and can be including memory, processor and storage
The computer program run on the processor, which is characterized in that the processor is realized such as when executing the computer program
The step of any one of claim 1 to 7 unlawful practice recognition methods.
11. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey
Sequence realizes the step of the unlawful practice recognition methods as described in 1 to 7 any one of right when the computer program is executed by processor
Suddenly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910414712.0A CN110163143A (en) | 2019-05-17 | 2019-05-17 | Unlawful practice recognition methods, device and terminal device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910414712.0A CN110163143A (en) | 2019-05-17 | 2019-05-17 | Unlawful practice recognition methods, device and terminal device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110163143A true CN110163143A (en) | 2019-08-23 |
Family
ID=67631220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910414712.0A Pending CN110163143A (en) | 2019-05-17 | 2019-05-17 | Unlawful practice recognition methods, device and terminal device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110163143A (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633872A (en) * | 2019-09-26 | 2019-12-31 | 山东鲁能软件技术有限公司 | Violation behavior identification method and system based on big data analysis |
| CN111144232A (en) * | 2019-12-09 | 2020-05-12 | 国网智能科技股份有限公司 | Transformer substation electronic fence monitoring method based on intelligent video monitoring, storage medium and equipment |
| CN111310612A (en) * | 2020-01-22 | 2020-06-19 | 中国建设银行股份有限公司 | Behavior supervision method and device |
| CN111339933A (en) * | 2020-02-25 | 2020-06-26 | 北京国网富达科技发展有限责任公司 | Transformer substation safety monitoring method and device based on deep learning |
| CN111461655A (en) * | 2020-03-31 | 2020-07-28 | 国网河北省电力有限公司沧州供电分公司 | Personnel management system |
| CN111507320A (en) * | 2020-07-01 | 2020-08-07 | 平安国际智慧城市科技股份有限公司 | Detection method, device, equipment and storage medium for kitchen violation behaviors |
| CN111553305A (en) * | 2020-05-09 | 2020-08-18 | 中国石油天然气集团有限公司 | Violation video identification system and method |
| CN111738290A (en) * | 2020-05-14 | 2020-10-02 | 北京沃东天骏信息技术有限公司 | Image detection method, model construction and training method, device, equipment and medium |
| CN111783744A (en) * | 2020-07-31 | 2020-10-16 | 上海仁童电子科技有限公司 | Operation site safety protection detection method and device |
| CN111950647A (en) * | 2020-08-20 | 2020-11-17 | 连尚（新昌）网络科技有限公司 | Classification model training method and device |
| 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 |
| CN112203053A (en) * | 2020-09-29 | 2021-01-08 | 北京市政建设集团有限责任公司 | Intelligent supervision method and system for subway constructor behaviors |
| CN112530144A (en) * | 2020-11-06 | 2021-03-19 | 华能国际电力股份有限公司上海石洞口第一电厂 | Method and system for warning violation behaviors of thermal power plant based on neural network |
| WO2022037279A1 (en) * | 2020-08-19 | 2022-02-24 | 广西电网有限责任公司贺州供电局 | Operation violation detection method for power transformation site |
| CN114819004A (en) * | 2022-07-04 | 2022-07-29 | 广东电网有限责任公司佛山供电局 | Violation identification method and system based on multi-source data fusion |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567743A (en) * | 2011-12-20 | 2012-07-11 | 东南大学 | Automatic identification method of driver gestures based on video images |
| CN103020596A (en) * | 2012-12-05 | 2013-04-03 | 华北电力大学 | Method for identifying abnormal human behaviors in power production based on block model |
| CN106709447A (en) * | 2016-12-21 | 2017-05-24 | 华南理工大学 | Abnormal behavior detection method in video based on target positioning and characteristic fusion |
| CN108174165A (en) * | 2018-01-17 | 2018-06-15 | 重庆览辉信息技术有限公司 | Electric power safety operation and O＆M intelligent monitoring system and method |
| CN108932479A (en) * | 2018-06-06 | 2018-12-04 | 上海理工大学 | A kind of human body anomaly detection method |
-
2019
- 2019-05-17 CN CN201910414712.0A patent/CN110163143A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567743A (en) * | 2011-12-20 | 2012-07-11 | 东南大学 | Automatic identification method of driver gestures based on video images |
| CN103020596A (en) * | 2012-12-05 | 2013-04-03 | 华北电力大学 | Method for identifying abnormal human behaviors in power production based on block model |
| CN106709447A (en) * | 2016-12-21 | 2017-05-24 | 华南理工大学 | Abnormal behavior detection method in video based on target positioning and characteristic fusion |
| CN108174165A (en) * | 2018-01-17 | 2018-06-15 | 重庆览辉信息技术有限公司 | Electric power safety operation and O＆M intelligent monitoring system and method |
| CN108932479A (en) * | 2018-06-06 | 2018-12-04 | 上海理工大学 | A kind of human body anomaly detection method |
Non-Patent Citations (3)
| Title |
|---|
| 杨帆 等: "《精通图像处理经典算法（MATLAB版）》", 30 April 2014, 北京航空航天大学出版社 * |
| 码迷: "灰度图像--图像分割 阈值处理之平均阈值", 《HTTP://WWW.MAMICODE.COM/INFO-DETAIL-501991.HTML》 * |
| 谭升: "【数字图像处理】7.1：灰度图像-图像分割 阈值处理之平均阈值", 《HTTPS://FACE2AI.COM/DIP-7-1-%E7%81%B0%E5%BA%A6%E5%9B%BE%E5%83%8F-%E5%9B%BE%E5%83%8F%E5%88%86%E5%89%B2-%E9%98%88%E5%80%BC%E5%A4%84%E7%90%86%E4%B9%8B%E5%B9%B3%E5%9D%87%E9%98%88%E5%80%BC/》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110633872A (en) * | 2019-09-26 | 2019-12-31 | 山东鲁能软件技术有限公司 | Violation behavior identification method and system based on big data analysis |
| CN111144232A (en) * | 2019-12-09 | 2020-05-12 | 国网智能科技股份有限公司 | Transformer substation electronic fence monitoring method based on intelligent video monitoring, storage medium and equipment |
| CN111310612A (en) * | 2020-01-22 | 2020-06-19 | 中国建设银行股份有限公司 | Behavior supervision method and device |
| CN111339933A (en) * | 2020-02-25 | 2020-06-26 | 北京国网富达科技发展有限责任公司 | Transformer substation safety monitoring method and device based on deep learning |
| CN111461655A (en) * | 2020-03-31 | 2020-07-28 | 国网河北省电力有限公司沧州供电分公司 | Personnel management system |
| CN111553305B (en) * | 2020-05-09 | 2023-05-05 | 中国石油天然气集团有限公司 | System and method for identifying illegal videos |
| CN111553305A (en) * | 2020-05-09 | 2020-08-18 | 中国石油天然气集团有限公司 | Violation video identification system and method |
| CN111738290A (en) * | 2020-05-14 | 2020-10-02 | 北京沃东天骏信息技术有限公司 | Image detection method, model construction and training method, device, equipment and medium |
| CN111738290B (en) * | 2020-05-14 | 2024-04-09 | 北京沃东天骏信息技术有限公司 | Image detection method, model construction and training method, device, equipment and medium |
| CN111507320A (en) * | 2020-07-01 | 2020-08-07 | 平安国际智慧城市科技股份有限公司 | Detection method, device, equipment and storage medium for kitchen violation behaviors |
| CN111783744A (en) * | 2020-07-31 | 2020-10-16 | 上海仁童电子科技有限公司 | Operation site safety protection detection method and device |
| WO2022037279A1 (en) * | 2020-08-19 | 2022-02-24 | 广西电网有限责任公司贺州供电局 | Operation violation detection method for power transformation site |
| CN111950647A (en) * | 2020-08-20 | 2020-11-17 | 连尚（新昌）网络科技有限公司 | Classification model training method and device |
| 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 |
| CN112203053A (en) * | 2020-09-29 | 2021-01-08 | 北京市政建设集团有限责任公司 | Intelligent supervision method and system for subway constructor behaviors |
| CN112530144A (en) * | 2020-11-06 | 2021-03-19 | 华能国际电力股份有限公司上海石洞口第一电厂 | Method and system for warning violation behaviors of thermal power plant based on neural network |
| CN112530144B (en) * | 2020-11-06 | 2022-06-28 | 华能国际电力股份有限公司上海石洞口第一电厂 | Method and system for warning violation behaviors of thermal power plant based on neural network |
| CN114819004A (en) * | 2022-07-04 | 2022-07-29 | 广东电网有限责任公司佛山供电局 | Violation identification method and system based on multi-source data fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163143A (en) | Unlawful practice recognition methods, device and terminal device | |
| CN112651978B (en) | Sublingual microcirculation image segmentation method and device, electronic equipment and storage medium | |
| CN104063686B (en) | Crop leaf diseases image interactive diagnostic system and method | |
| CN111653023A (en) | Intelligent factory supervision method | |
| CN106650804B (en) | A kind of face sample cleaning method and system based on deep learning feature | |
| CN110363204A (en) | A kind of object expression method based on multitask feature learning | |
| CN111242865A (en) | Fundus image enhancement method based on generation type countermeasure network | |
| CN106778910A (en) | Deep learning system and method based on local training | |
| CN116168348B (en) | Security monitoring method, system and storage medium based on image processing | |
| CN110363114A (en) | A kind of person works' condition detection method, device and terminal device | |
| CN108764287B (en) | Target detection method and system based on deep learning and packet convolution | |
| CN114140844A (en) | Face silence living body detection method and device, electronic equipment and storage medium | |
| CN111047598B (en) | Deep learning-based ultraviolet discharge light spot segmentation method and device for power transmission and transformation equipment | |
| CN116433518A (en) | Fire image smoke removing method based on improved Cycle-Dehaze neural network | |
| CN109829441B (en) | Facial expression recognition method and device based on course learning | |
| Guo et al. | A deep learning method for the recognition of solar radio burst spectrum | |
| CN113139431A (en) | Image saliency target detection method based on deep supervised learning | |
| CN117253192A (en) | Intelligent system and method for silkworm breeding | |
| CN112488213A (en) | Fire picture classification method based on multi-scale feature learning network | |
| Zhang et al. | A novel DenseNet Generative Adversarial network for Heterogenous low-Light image enhancement | |
| CN116740808A (en) | Animal behavior recognition method based on deep learning target detection and image classification | |
| CN115620170A (en) | Multi-source optical remote sensing image water body extraction method based on SegFormer | |
| CN115731620A (en) | Method for detecting counter attack and method for training counter attack detection model | |
| CN111160213B (en) | Illegal boarding and disembarking detection method, system and storage medium based on deep learning | |
| CN113516003A (en) | Identification model-based identification method and device applied to intelligent security |
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: 20190823 |