CN108345854A - Information processing method, device, system based on image analysis and storage medium - Google Patents
Information processing method, device, system based on image analysis and storage medium Download PDFInfo
- Publication number
- CN108345854A CN108345854A CN201810128166.XA CN201810128166A CN108345854A CN 108345854 A CN108345854 A CN 108345854A CN 201810128166 A CN201810128166 A CN 201810128166A CN 108345854 A CN108345854 A CN 108345854A
- Authority
- CN
- China
- Prior art keywords
- information
- people
- region
- stream
- described image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- 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
- G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
This application provides a kind of information processing methods based on image analysis, are applied to cloud platform, and the cloud platform connects each monitoring node disposed in network, the method includes:The image of the monitoring node corresponding region is received from the monitoring node;Described image is acquired in real time by the image capture device in the monitoring node;Individual identification processing is carried out to described image, the result handled according to the individual identification determines the associated people information in the region;It is generated and the associated management information in the region according to the people information;The associated management information in the region is sent to the corresponding management node in the region.Correspondingly, present invention also provides corresponding system, device and storage mediums.
Description
Technical field
This application involves information technology field more particularly to a kind of information processing method based on image analysis, device, it is
System and storage medium.
Background technology
Currently, with the development of internet, Internet of Things also fast-developing and extensive use, this is computer development in science and technology
Inevitable outcome, Internet of Things refer to by various information sensing devices, any required monitoring of acquisition in real time, connection, interactive object
Or the information of the various needs such as process, the huge network to be formed is combined with internet, the purpose is to realize object and object, object with
People, the connection of all articles and network facilitate identification, management and control.
Invention content
The embodiment of the present application proposes a kind of information processing method based on image analysis, is applied to cloud platform, the cloud
The each monitoring node disposed in platform connection network, the method includes:The monitoring node pair is received from the monitoring node
Answer the image in region;Described image is acquired in real time by the image capture device in the monitoring node;Described image is carried out a
Body identifying processing, the result handled according to the individual identification determine the associated people information in the region;According to the stream of people
Information generates and the associated management information in the region;The associated management information in the region is sent to the corresponding pipe in the region
Manage node.
The embodiment of the present application proposes a kind of information processing unit based on image analysis, is applied to cloud platform, the cloud
The each monitoring node disposed in platform connection network, described device include:Receiving module receives the prison from the monitoring node
Control the image of node corresponding region;Described image is acquired in real time by the image capture device in the monitoring node;Identification module,
Individual identification processing is carried out to described image, the result handled according to the individual identification determines that the associated stream of people in the region believes
Breath;Analysis module generates and the associated management information in the region according to the people information;Sending module sends the area
The associated management information in domain is to the corresponding management node in the region.
The embodiment of the present application proposes a kind of information processing system based on image analysis, including:It is deployed in network
The cloud platform and management node of multiple monitoring nodes, the multiple monitoring node of connection;Wherein, the monitoring node, including figure
As collecting device, the image of the monitoring node corresponding region is acquired in real time by described image collecting device, and by the figure
As being sent to the cloud platform;The cloud platform receives the image of the monitoring node corresponding region from the monitoring node, to institute
It states image and carries out individual identification processing, the result handled according to the individual identification determines the associated people information in the region,
According to the people information generate with the associated management information in the region, send the associated management information in the region to described
The corresponding management node in region
In some instances, the cloud platform includes individual identification server and analysis alarm server;Wherein, described
Body identifies that server receives the image of the monitoring node corresponding region from the monitoring node, and individual identification is carried out to described image
Processing;The result of individual identification processing is sent to the analysis alarm server;Wherein, the analysis alarm server, connects
Receive the result for the individual identification processing that the individual identification server is sent;The result handled according to the individual identification is true
The fixed associated people information in region generates and the associated management information in the region, transmission institute according to the people information
The associated management information in region is stated to the corresponding management node in the region.
In some instances, the individual identification server carries out individual identification processing to described image, including:To described
Image carries out characteristics of human body's identification and/or individual behavior analysis.
In some instances, the management node includes:Management platform and/or manager's client.
The application also proposed a kind of non-volatile computer readable storage medium storing program for executing, be stored with computer-readable instruction, can
So that at least one processor executes the process described above.
As can be seen that it can be with by information processing method above based on image analysis and system by above technical scheme
Find out, method and system provided by the present application can the quick, intelligent stream of people for extracting each region of the image data based on magnanimity
Information, and according to the generation management information of people information intelligence, artificial operation is avoided, it improves for each region (ratio
Such as market, tourist attraction) management operating efficiency.
Description of the drawings
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without having to pay creative labor, may be used also for those of ordinary skill in the art
With obtain other attached drawings according to these attached drawings.
Fig. 1 is illustrated by the applicable network structure of the information processing system based on image analysis of the application one embodiment
Figure;
Fig. 2 a are the functional module structure schematic diagrames of information processing system of the application one embodiment based on image analysis;
Fig. 2 b are the work(of the individual identification module of the information processing system based on image analysis of the application one embodiment
It can modular structure schematic diagram;
Fig. 2 c are the density of stream of people distribution plane figure described in the application one embodiment;
Fig. 2 d are that the terminal device for being equipped with remote monitoring module described in the application one embodiment shows density of stream of people
The example of distribution plane figure;
Fig. 3 is the specific implementation flow chart of the information processing method based on image analysis of the application one embodiment;
Fig. 4 is the specific implementation flow chart of the information processing method based on image analysis of the application one embodiment;
Fig. 5 is the structural schematic diagram of the information processing unit based on image analysis of the application one embodiment;And
Fig. 6 is the hardware architecture diagram of the computing device described in the application one embodiment.
Specific implementation mode
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation describes, it is clear that described embodiment is only a part of the embodiment of the present invention, instead of all the embodiments.Based on this
Embodiment in invention, every other reality obtained by those of ordinary skill in the art without making creative efforts
Example is applied, shall fall within the protection scope of the present invention.
It is succinct and intuitive in order to what is described, hereafter by describing several representative embodiments come to the solution of the present invention
It is illustrated.A large amount of details is only used for helping to understand the solution of the present invention in embodiment.However, it will be apparent that the technology of the present invention
Scheme can be not limited to these details when realizing.In order to avoid unnecessarily having obscured the solution of the present invention, some embodiment party
Formula is not described meticulously, but only gives frame.Hereinafter, " comprising " refers to " including but not limited to ", " root
According to ... " refer to " according at least to ..., but be not limited to according only to ... ".Hereinafter it is not specifically stated the quantity of an ingredient
When, it is meant that the ingredient is either one or more, or can be regarded as at least one.
Currently, user plane simply can not be retrieved and be analyzed using tactics of human sea, therefore to the image data of magnanimity
The monitoring to the stream of people using technology of Internet of things realization is needed, and then facilitates the identification, management and control of administrative staff.The application's
Embodiment proposes a kind of information processing method, device and system based on image analysis.
Fig. 1 shows the network that the information processing system based on image analysis that some embodiments of the application provide is applicable in
Structural schematic diagram.As shown in Figure 1, this network 100 includes at least:Image capture device 11, network 12, computing device 13 and service
Device 14.
In some embodiments of the present application, above-mentioned image capture device 11 can be mounted in camera everywhere, than
Camera such as at market, station public place, the image information for acquiring corresponding region, said one or more
A image capture device 11 and Internet of Things communication module together form a monitoring node of information processing system.
Network 12 may include cable network and wireless network.As shown in Figure 1, netting side, image capture device in access
11 and terminal device 13 can wirelessly or wired mode is linked into network 12;And in core net side, clothes
Business device 14 is connected to network 12 generally by wired mode.Certainly, server 14 can also be wirelessly connected to
Network 12.
Computing device 13 can be the intelligent terminals such as smart mobile phone, PAD and tablet computer, PC, can also
It is the server of another management platform, such as the Police Information management platform of public security bureau.At least one computing device 13 can be with
The management node for constituting information processing system in the application, is equipped with remote monitoring module, for receiving and showing service thereon
The management information in the region that the image and server 14 that device 14 is obtained from image capture device 11 are sent, such as warning information
Deng.
Server 14 is the cloud platform for connecting each monitoring node disposed in network, includes this information processing system thereon
In individual identification module and analysis alarm module, monitoring, security protection service are provided together with computing device 13, for example, calculating set
Standby 13 can show the image information that image capture device 11 acquires, and the analysis alarm module when there is exception on server 14 can be to
Remote monitoring module on computing device 13 sends a warning message, so that it shows to remind user etc..Above-mentioned server 14 can
Can also be the cluster server that multiple servers form to be individual server.
On the basis of above-mentioned network structure shown in FIG. 1, embodiments herein provides a kind of based on image analysis
Information processing system.Fig. 2 a show the group of the information processing system 200 provided by the embodiments of the present application based on image analysis
At structural schematic diagram.As shown in Figure 2 a, which may include:One or more monitoring node 21 includes Internet of Things Network Communication
Module 201 and image capture device 11, cloud platform 22 include individual identification module 202 and analysis alarm module 203 and one
Or multiple management nodes 23 include remote monitoring module 204.
Wherein, above-mentioned monitoring node 21, acquires the image of the monitoring node corresponding region in real time, and described image is sent out
Send the cloud platform 22 for each monitoring node disposed into connection network.Wherein, the monitoring node includes one or more figures
As collecting device and Internet of Things communication module 201, the image capture device in above-mentioned monitoring node is for acquiring the monitoring section
The image information of point corresponding region can be each camera for being located at floor, garage and road etc.;And above-mentioned monitoring section
Internet of Things communication module 201 in point includes the communication modules such as GPRS communication cards, 3G/4G/5G modules, for cloud platform 22 and
Management node 23 communicates, for example the described image information that above-mentioned image capture device acquires is sent to above-mentioned cloud platform 22.
In some instances, the acquisition for the analysis alarm server that monitoring node 21 may also respond to cloud platform 22 refers to
It enables, acquires the image of the corresponding target individual of the face characteristic information to collect evidence.
Cloud platform 22, for receiving the image of the monitoring node corresponding region from the monitoring node, to described image into
The processing of row individual identification, the result handled according to the individual identification determines the associated people information in the region, according to described
People information generates and the associated management information in the region, sends the associated management information in the region to the region and corresponds to
Management node.
Management node 23 receives the above-mentioned management information that cloud platform 22 is sent.
In some instances, cloud platform 22 includes individual identification server 221 and analysis alarm server 222;Wherein, institute
The image that individual identification server 221 receives the monitoring node corresponding region from the monitoring node is stated, described image is carried out
Individual identification processing;The result of individual identification processing is sent to the analysis alarm server 222;Wherein, the analysis is accused
Alert server 222 receives the result for the individual identification processing that the individual identification server is sent;At the individual identification
The result of reason determines the associated people information in the region, is generated according to the people information and is believed with the associated management in the region
Breath, sends the associated management information in the region to the corresponding management node in the region.
Here, individual identification module 202, above-mentioned analysis alarm server are installed on above-mentioned individual identification server 221
Analysis alarm module 203 is installed on 222.It should be noted that above-mentioned individual identification server 221 and above-mentioned analysis alarm clothes
Device 222 of being engaged in is the same server or different servers, that is, individual identification module 202 and analysis alarm module 203 can be with
It on the same server can also be on different server.
Specifically, the individual identification server 221 in cloud platform 22 receives the Internet of Things communication module in monitoring node 21
The image in the corresponding region of monitoring node that 201 image capture devices sent acquire in real time;To above-mentioned image capture device
People and/or object in the image collected carry out individual identification processing, and the result of individual identification processing is sent to cloud platform
22 analysis alarm server 222 is so that it carries out subsequent analysis processing.Wherein, the individual identification, which is handled, includes:To described
Image carries out characteristics of human body's identification and/or individual behavior analysis.Here, the people in above-mentioned image can be single or multiple;On
It includes photo and video etc. to state image;Above-mentioned user is the use using the management node of the information processing system based on image analysis
Family, such as administrator, monitoring personnel etc., wherein above-mentioned management node includes:Management platform and/or manager's client, such as
Above-mentioned management platform can be the information management of the Police Information management platform of public security bureau, the information management platform in market, scenic spot
Platform or other platforms for needing to complete the mechanism for monitoring and managing based on people information, above-mentioned manager's client can be with
It is the management client that the administrative staff of the mechanisms such as market, scenic spot, public security bureau use.
Further, Fig. 2 b show the information processing system 200 provided by the embodiments of the present application based on image analysis
Cloud platform 22 in individual identification module 202 on individual identification server 221 functional module structure schematic diagram.Such as Fig. 2 b institutes
Show, individual identification module 202 may include characteristics of human body's recognition unit 2021 and individual behavior analytic unit 2022.
In some instances, above-mentioned individual identification server 221 carries out first nerves network by multiple sample images
Training, obtains the first nerves network model identified for characteristics of human body;Wherein, individual identification is carried out to described image when described
Processing includes when carrying out characteristics of human body's identification to described image, and above-mentioned individual identification server 221 is further defeated by described image
Enter the first nerves network model;According to the output of the first nerves network model, determine that the human body in described image is special
Reference ceases;Characteristics of human body's information in determining described image is sent to the analysis alarm server 222.Wherein,
Above-mentioned multiple sample images can be preset or being acquired from actual scene, it is also possible to dynamic input.
Specifically, characteristics of human body's recognition unit 2021 in the individual identification module 202 of above-mentioned individual identification server 221
First nerves network is trained by multiple sample images, obtains the first nerves network mould identified for characteristics of human body
Type;Described image is further inputted into the first nerves network model;According to the output of the first nerves network model, really
Determine characteristics of human body's information in described image;And characteristics of human body's information in determining described image is sent to described point
Analyse alarm server 222.
Further, it includes carrying out Head recognition and face to described image to carry out characteristics of human body's identification to described image
Identification, characteristics of human body's recognition unit 2021 further carry out Head recognition and recognition of face to described image, in above-mentioned image
Characteristics of human body's information includes number of people information and face information.
In some instances, the analysis alarm server 222 receives the figure that the individual identification server is sent
Characteristics of human body's information as in;Further according to the number of people information in characteristics of human body's information in described image, the area is determined
Number in domain and/or density of stream of people data;According to the number and/or density of stream of people data in the region, obtain determining with
The associated people information in region, the people information include:In density of stream of people distribution, the distribution of stream of people's movement locus, number
At least one.
Specifically, the analysis alarm module 203 of above-mentioned analysis alarm server 222 can be determined according to above-mentioned image it is above-mentioned
Number of people information in image determines the image section with number of people feature in image, and then is determined according to above-mentioned number of people information
Number in corresponding region simultaneously calculates the density of stream of people data in above-mentioned zone;And it further determines that and is associated with the region
People information, such as the region density of stream of people distribution, stream of people's movement locus and number distribution etc..
For example, the image input that above-mentioned individual identification server 221 acquires the image capture device in a certain market
One neural network model can obtain the number of people information in the above-mentioned image in the market, and the people's header is sent to
State analysis alarm server 222;Above-mentioned analysis alarm server 222 determines the number in the market according to above-mentioned number of people information
And/or density of stream of people data, and then determine the stream of peoples such as the density of stream of people distribution in the market, the distribution of stream of people's movement locus, number
Information.
In some instances, the analysis alarm server 222 receives the institute that the individual identification server 221 is sent
State characteristics of human body's information in image;According to the face characteristic information in characteristics of human body's information in described image, determine described in
The identification information of the identification information of target individual in region, the target individual is used to indicate whether that there are the target individuals;
According to the identification information of target individual in the region, determine and the associated people information in the region, the people information packet
It includes:The position of target individual and/or the movement locus of target individual.
For example, the correspondence area that above-mentioned individual identification server 221 can acquire an image capture device at a certain station
Individual in the image in domain carries out recognition of face and obtains face characteristic information, and above-mentioned face characteristic information is sent to above-mentioned point
Analyse alarm server 222;Above-mentioned analysis alarm server 222 determines target individual according to face characteristic information in above-mentioned zone
People information, here, the people information of above-mentioned target individual can be the position of target individual and the movement of target individual
Track etc., for example, individual identification server 221 is in the image of the corresponding region of the image capture device acquisition at a certain station
Individual carry out recognition of face obtain face characteristic information, by above-mentioned face characteristic information be sent to analysis alarm server 222,
Alarm server 222 is analyzed through overmatching, determines that target individual is the escaped criminal that public security system is ordered to arrest, then analyzes alarm server
222 generate matching warning information, and the management which is sent to the management node and public security system at station is put down
Platform is so that the administrative staff and public security officer at station take corresponding measure, and are adopted to the transmission of the corresponding monitoring node at above-mentioned station
Collection instructs so that it acquires the image of above-mentioned target individual collects evidence.
Further, characteristics of human body's recognition unit 2021 for identification the face characteristic information in described image first
Network model may include depth convolutional neural networks and recurrent neural network.
Wherein, depth convolutional neural networks receive the image of the image capture device acquisition of Internet of Things communication module 201,
Extract one or more convolutional neural networks features of above-mentioned image, and by the said one of extraction or multiple convolutional neural networks
Feature (CNN features:Convolutional Neural Network features) input recurrent neural network.
Here, difference lies in convolutional neural networks contain input layer, volume to convolutional neural networks with general neural network
Lamination, pond layer and full articulamentum.Wherein, above-mentioned input layer is responsible for input picture, such as the image of a 32*32*3;Convolution
Layer is responsible for extracting above-mentioned input picture feature by convolutional calculation;Above-mentioned pond layer compresses the characteristic pattern of input, a side
Face can make characteristic pattern become smaller, and simplify network calculations complexity;On the other hand Feature Compression is carried out, main feature is extracted;It is above-mentioned
Full articulamentum connects all features of above-mentioned convolutional neural networks, and outputs it and give grader and divide features described above
Class.Specifically, can have multiple convolutional layers in convolutional neural networks, therefore can be by the more convolutional Neural net of the convolution number of plies
Network is known as depth convolutional neural networks.In a convolutional layer of convolutional neural networks, several characteristic planes are generally comprised,
In each characteristic plane be made of the neuron of some rectangular arrangeds, each neuron is only connect with part adjacent bed neuron,
And the neuron of same characteristic plane shares weights, shared weights are exactly convolution kernel here.Convolution kernel is generally with random decimal
The form of matrix initializes, the rational weights in the training process of network by learning to obtain.Utilize convolutional layer
The input of convolution kernel and the convolutional layer carries out convolution, can obtain the convolutional neural networks feature (CNN features) of the convolutional layer.On
It is the connection reduced between each layer of network to state the direct benefit that shared weights (convolution kernel) are brought, while reducing over-fitting again
Risk.And CNN is characterized as the corresponding feature of the convolutional layer of the different depth of convolutional neural networks, the CNN feature generations of different depth
The abstract characteristics of table different levels, CNN depths of features is deeper, and CNN features are more abstract, conversely, CNN depth is more shallow, CNN features are got over
Specifically.In this way, the CNN features using different depth can more comprehensively describe face so that the result of recognition of face is more forced
Very, effect is more preferable.
Without loss of generality it is assumed that some region of number is 4 people, can be with if density of stream of people is bigger in the region
Greater value is set.Full articulamentum 1, full articulamentum 2, full articulamentum 3 and full articulamentum 4 one are extracted using 52 layers of convolutional neural networks
Totally four layers of CNN features can then obtain the CNN features of 4 different depths.
Recurrent neural network learns above-mentioned one or more convolutional neural networks features successively, and makes each recurrent neural list
Member all exports the number of people rectangle frame information of each convolutional neural networks feature, wherein above-mentioned number of people rectangle frame information is the region
The number of people and its corresponding location information of the different CNN features of interior correspondence, the number of people of different CNN features correspond to different probability,
Namely different exact level.For example, there is 4 people in a region, then each recurrent neural unit can identify corresponding difference
The number of people rectangle frame information of CNN features and each number of people rectangle frame information include being distributed 4 numbers of people of different location.Here,
Recurrent neural network can be long Memory Neural Networks (LSTM in short-term:Long short-term Memory).
It is much partly overlapping, such as a certain due to the number of people rectangle frame information of each recurrent neural unit output
There are 4 people in region, then has 4 in the number of people rectangle frame information of one CNN feature of correspondence of the same recurrent neural unit output
Number of people rectangle frame, and 4 number of people rectangle frames are probably overlapped, and therefore, recurrent neural network further will be above-mentioned
Number of people rectangle frame information carries out non-maximum suppression processing, and the larger number of people rectangle frame of degree of overlapping is deleted, final face is obtained
Recognition result.Wherein, non-maximum suppression processing inhibits the principle of non-maximum element, will overlap using search local maximum
It spends larger number of people rectangle frame to delete, obtains final face recognition result namely face characteristic information.Based on depth convolution god
Recognition of face through network makes more efficient to identifying for face, identification accuracy greatly improve.
In some instances, the individual identification server 221 carries out nervus opticus network by multiple sample images
Training, obtains the nervus opticus network model analyzed for individual behavior;Wherein, individual identification is carried out to described image when described
Processing includes when carrying out individual behavior analysis to described image, and the individual identification server 221 is further defeated by described image
Enter the nervus opticus network model;According to the output of the nervus opticus network model, row individual in described image is determined
It is characterized;The behavioural characteristic of individual in determining described image is sent to the analysis alarm server 222.
Specifically, the individual behavior analytic unit in the individual identification module 202 of above-mentioned individual identification server 221
2022, nervus opticus network is trained by multiple sample images, obtains the nervus opticus net analyzed for individual behavior
Network model;Described image is inputted into the nervus opticus network model;According to the output of the nervus opticus network model, determine
Individual behavioural characteristic in described image;The behavioural characteristic of individual in determining described image is sent to the analysis alarm
Server 222.Here, above-mentioned first nerves network model and above-mentioned nervus opticus network model can be a neural network moulds
Type can also be two different neural network models, and the application is to this without limiting.In addition, as previously mentioned, above-mentioned multiple
Sample image can be preset or be acquired from actual scene, it is also possible to dynamic input.
For example, the image that the image capture device in a market acquires can be input to by individual behavior analytic unit 2022
Above-mentioned nervus opticus network model, to the individual in above-mentioned image carry out individual behavior analyze such as above-mentioned individual clothing, with
Body article determines behavioural characteristic individual in above-mentioned image into the time etc. in the region, it may also be said to determine above-mentioned individual
Attribute tags, such as " women ", " youth ", " women's dress ", " mother and baby's articles for use " and " children's garment " etc..
In some instances, the analysis alarm server 222 receives the institute that the individual identification server 221 is sent
State the behavioural characteristic of the individual in image;According to behavioural characteristic individual in described image, each individual in the region is determined
Behavioural characteristic;According to the behavioural characteristic of each individual in the region, the determining and associated people information in the region is described
People information includes:The corresponding density of stream of people distribution of behavioural characteristic, the corresponding stream of people's movement locus of behavioural characteristic, behavioural characteristic pair
At least one of number answered.
For example, the image of the image capture device acquisition in a market is a certain children's garment shop of the young woman in the market
Children's garment are bought, the individual behavior analytic unit 2022 in 221 individual identification modules 202 of individual identification server is defeated by the image
Enter above-mentioned nervus opticus network model, the behavior of the individual of the personnel is determined according to the output of above-mentioned nervus opticus network model
It is sent to analysis alarm server 222 characterized by " young woman ", " children's garment ", and by the behavioural characteristic of the individual of the personnel;Point
Analysis alarm server 222 can determine that the behavior of each individual in the market is special according to the behavioural characteristic of each individual in the image
Sign, and then determine the associated people information in the market, such as the corresponding density of stream of people distribution of behavioural characteristic " young woman " or people
Flow movement locus, i.e. the density of stream of people distribution of young woman's individual or stream of people's movement locus, the corresponding people of behavioural characteristic " children's garment "
Current density is distributed or stream of people's movement locus, that is, buys density of stream of people distribution or stream of people's movement locus of the individual of children's garment etc..
More further, the behavior of target individual can also be pre-configured in above-mentioned nervus opticus network model according to demand
Feature is had a preference for purchase such a behavior of sportswear such as young man, is pre-configured in above-mentioned nervus opticus network model
The behavioural characteristic of " young man " and " sportswear ".When individual identification server 221 is identified from the image collected with above-mentioned
The target individual of behavioural characteristic " young man " and " sportswear ", analysis alarm server 222 is according to individual identification server 221
Recognition result, analysis obtain the corresponding people information of behavior feature, for example, the corresponding density of stream of people distribution of behavior feature and
Movement locus and corresponding temporal information, and according to above-mentioned people information determination movement brand to formulate determining marketing strategy,
For example determine the movement brand that young man likes, carry out discounting activity etc. in the corresponding time.
In the other example of the application, according to the behavioural characteristic of each individual in the region, it can also determine
Target group corresponding with the behavioural characteristic in the region, so that administrative staff are special according to the target group and its behavior
Sign formulates corresponding operation and management strategy.For example, according to " cosmetics ", " women's dress ", " children's garment ", " weekend " three behavioural characteristics
It can determine the target group of " young woman ", and determine these " young woman " groups at weekend according to its behavioural characteristic
Stream of people's distribution density is higher, also would know that the region that the face movement locus of " young woman " group is mainly distributed, can also obtain
Know businessman's brand that " young woman " group is primarily upon, and then administrative staff can determine a kind of operation and management strategy:
Weekend in the advertising campaign of these regional arrangement women's dresses, children's garment and cosmetics, and can determine and invite participation advertising campaign businessman
Brand.
For example, it is assumed that administrative staff are the administrative staff in market, the administrative staff in above-mentioned market can be according to target group
Behavioural characteristic be adjusted operation way, such as it is big to flow of the people by being adjusted with the matched retail shop of the behavioural characteristic of target group
Region or plan the shop that the behavioural characteristic of some and target group matches and do advertising campaign etc..
In this way, the analysis based on individual behavior feature in each region, management node can easily and efficiently specify operation
And management strategy, considerable business earnings can not only be brought, moreover it is possible to help management node to local stream of people's distribution, trade company
Distribution etc. is efficiently regulated and controled.
As can be seen that the individual behavior analytic unit 2022 of individual identification server 221 can people in dynamic tracing area
Group's flowing, makes user understand stream of people's trend in the region, and precisely obtain the behavioural characteristic in the region, with
And people information corresponding with behavioural characteristic even target group, user can be helped to make significantly more efficient management and commercially determine
Plan.
Without loss of generality it is assumed that determine the region one or more target group and its behavioural characteristic can adopt
It is stored with following format:
Target group 1:Behavioural characteristic 1, behavioural characteristic 2, behavioural characteristic 3 ...
Target group 2:Behavioural characteristic 1, behavioural characteristic 2, behavioural characteristic 3 ...
Target group 3:Behavioural characteristic 1, behavioural characteristic 2, behavioural characteristic label 3 ...
In some instances, above-mentioned management information includes:In stream of people's statistical information, warning information, management strategy at least
One.
In some instances, when the management information includes at least stream of people's statistical information, the analysis alarm clothes
Be engaged in device 222, according in the people information density of stream of people distributed data and/or number distributed data draw people in the region
Current density distribution plane figure and/or number distribution plane figure;In response to the displaying request of the management node 23, the people is sent
Current density distribution plane figure and/or number distribution plane figure are to the management node.
In some instances, when the management information includes at least warning information, when determining the number in the region
When being more than second threshold more than the density of stream of people distributed data in first threshold and/or the region, the analysis alerting service
Device 222 generates crowded warning information, and the crowded warning information is sent to the management node 23.
Fig. 2 c show that a certain region density of stream of people distribution plane figure, the figure can be shown by management node, displaying
Form can be floating window formula, full screen etc..It is as shown in Figure 2 c, different in the density of stream of people of different zones in the 201c of the region,
The density of stream of people in region that white 202c is filled in the region be 0 people/square metre, in the region of vertical bar shade 203c filling
Density of stream of people be 2 people/square metre, the density of stream of people in the region of grid shade 204c filling be 4 people/square metre, dot shade
Density of stream of people in the region of 205c fillings be 6 people/square metre, the density of stream of people in the region of diagonal line hatches 206c filling is 8
People/square metre, the density of stream of people in the region of horizontal line shade 207c filling be the above people of 10 people/square metre.
For example, the displaying in response to administrative staff's client in market is asked, the density of stream of people in the market is distributed flat
Above-mentioned administrative staff's client occurs for face figure and/or number distribution plane figure so that it is shown to supply the administrator in the market
Member's real time monitoring.
In some instances, when the management information includes at least stream of people's statistical information, the analysis alarm clothes
Business device 222, stream of people's movement locus figure in the region is drawn according to stream of people's motion trace data in the people information;It rings
The displaying of management node 23 described in Ying Yu is asked, and sends stream of people's movement locus figure to the management node 23.
In some instances, when the management information includes at least warning information, when determining the number in the region
When being more than second threshold more than the density of stream of people distributed data in first threshold and/or the region, the analysis alerting service
Device 222 generates crowded warning information, and the crowded warning information is sent to the management node 23.Wherein, above-mentioned crowded
Warning information includes crowded position and congestion index;Above-mentioned first threshold is preset value with above-mentioned second threshold.It can be with
Find out the system can accurate stream of people's congestion levels in assessment area, and send crowded alarm in time, win processing accident
Time.
Still by taking above-mentioned zone is market as an example, when the number in the market is more than above-mentioned first threshold and/or as the quotient
Crowded warning information is generated when certain density of stream of people is more than above-mentioned second threshold in, and is sent to remote monitoring module 204
The crowded warning information, so that market administrative staff take corresponding measure, for example current limliting, the stream of people such as dredge at the measures.Specifically, false
If the region 201c in Fig. 2 c is above-mentioned zone, the above-mentioned second threshold of hypothesis without loss of generality is 10, therefore when in the region
Generate crowded warning information when density of stream of people is more than second threshold 10 in the region of horizontal line shade 207c filling, and to the pipe in market
Reason personnel's client sends crowded warning information so that it shows and plays above-mentioned crowded warning information, and then notifies the market
Administrative staff take corresponding measure.
In some instances, when the management information includes at least stream of people's statistical information, the analysis alarm server
The movement locus of the position of target individual in the people information and/or target individual is sent to the management node by 222
23。
For example, when in the image that analysis alarm server 222 is acquired according to an image capture device in a station of reception
Characteristics of human body's information determine that there are target individuals at the station, determine position and/or the movement locus of the target individual, and should
The position of target individual and/or movement locus are sent to the client of the administrative staff at the station, so that it takes corresponding measure.
In some instances, when the management information includes at least stream of people's statistical information, in response to the management node
23 displaying request, the corresponding density of stream of people of the behavioural characteristic in the region is distributed by the analysis alarm server 222, behavior
At least one of the corresponding stream of people's movement locus of feature, the corresponding number of behavioural characteristic are sent to the management node 23.
In some instances, it is asked in response to the displaying of the management node 23, the analysis alarm server 222 into one
The target group in the determining region and its behavioural characteristic are sent to management node 23 by step, for its with reference to formulate operation and
Management strategy.
For example, the administrative staff in market can according in the market behavioural characteristic and the stream of people corresponding with behavior feature
The people informations such as Density Distribution, stream of people's movement locus and number formulate corresponding operation and management strategy, such as according to " female
The behavioural characteristics such as dress ", " 7. -8 points at night " formulate 7. -8 points at night and carry out the activities such as Discount Promotion to women's dress.
In some instances, above-mentioned analysis alarm server 222 is receiving what the individual identification server 221 was sent
After characteristics of human body's information in described image, the analysis alarm server 222 is by the target individual in the people information
Position and/or the movement locus of target individual be sent to the management node 23.
In some instances, the analysis alarm server 222, further by described image face characteristic information with
The face characteristic information in each face information record in face database is matched, wherein a face information record
Mark including target individual and face characteristic information;If in the face characteristic information and the face database in described image
Face characteristic information in one face information record matches, then generates and indicate that there are this face information notes in the region
Record the identification information of the corresponding target individual.
Further, when face information record further comprises affiliated library classification, if in described image
The face characteristic information in face information record in face characteristic information and the face database matches, and the analysis is accused
Alert server, the affiliated library classification in further being recorded according to the face information determine the face characteristic information in described image
Affiliated library classification;Matching warning information is further generated according to the determining affiliated library classification, and the matching is alerted
Information is sent to the management node.Wherein, if face characteristic information in described image in face information record
Face characteristic information similarity be more than first threshold, the analysis alarm server 222, it is determined that the people in described image
Face characteristic information in face characteristic information and face information record matches.
Here, above-mentioned face database may include the public security systems list library such as runaway convict library, Missing Persons library, can also include
VIP client's list library includes that one or more face information records in the face database;Above-mentioned matching warning information is according to affiliated library
Classification it is different, content is also different.Such as when the face letter in the face characteristic information and face database of certain individual in image
Breath record matches, and the affiliated library classification that this face information record includes is lost children library, then analyzes alarm server
222 determine that the individual is target individual and its affiliated library classification is lost children library, and then generate and prompt for lost children's
Warning information is matched, and is sent to management node 23.In another example when in the face characteristic information and face database of certain individual in image
A face information record match, and this face information record include affiliated library classification be runaway convict library, then analyze announcement
Alert server 222 determines that the individual is target individual and its affiliated library classification is runaway convict library, and then generates and prompt for runaway convict's
Warning information is matched, and the matching warning information that this is prompted for runaway convict is sent to management node, such as the information of public security system
The management node in the region residing for management platform and the target individual, while being sent out to the monitoring node 21 residing for the target individual
Acquisition instructions are sent, are collected evidence so that it acquires the image of said target individual
In some instances, when determining the face in face characteristic information in described image and face information record
When characteristic information matches, the analysis alarm server 222 further sends acquisition instructions to the monitoring node, so that
It acquires the image of the corresponding target individual of face characteristic information in described image.
It can thus be seen that system provided by the present application supports the retrieval matching operation of magnanimity face database, mesh is quickly completed
The search matching of face is marked, and returns to suspicious human face data in time, realizes that auto-alarming notice, great power-assisted public security system are built
Vertical intelligence pursues and captures an escaped prisoner or finds the system of population of wandering away.
In some instances, management node 23 receive and show that the image capture device that monitoring node 21 is sent is adopted in real time
The image of the corresponding region of collection.
In some instances, management node 23 is asked in response to the displaying of user, is sent above-mentioned displaying to cloud platform 22 and is asked
It asks;Receive the density of stream of people distribution plane figure and/or number distribution plane figure in the region that cloud platform 22 is sent;And it shows
Density of stream of people distribution plane figure and/or number distribution plane figure in the region monitor in real time for administrative staff.
Further, cloud platform 22 can also at predetermined time intervals such as 5 minutes, the stream of people in the region of transmission
Density Distribution plan view and/or number distribution plane figure are to management node 23;Management node 23 receives and updates the display area
Density of stream of people distribution plane figure in domain and/or number distribution plane figure.
Fig. 2 d show that the example of a management node displaying density of stream of people distribution plane figure, the management node are a pipe
The client of reason personnel, the client is on the terminal device shown in Fig. 2 d.The terminal device is shown in real time on screen 201d
Image, above-mentioned administrative staff click density of stream of people profile button 202d, continue while screen shows above-mentioned realtime graphic,
A new interface is opened, shows density of stream of people distribution plane Figure 20 3d in the form of small window in the lower right corner of screen 201d, together
When be supplied to above-mentioned administrative staff's close button 204d, so that above-mentioned administrative staff close the interface at any time.Further, it rings
Number button 205d should be clicked in above-mentioned administrative staff, in the lower section of screen 201d displaying current persons count 206d.
In some instances, management node 23 further receive the above-mentioned of the analysis alarm server transmission of cloud platform 22
Crowded warning information and above-mentioned matching warning information, and be shown and play, to remind above-mentioned administrative staff.
In some instances, management node 23 further receive the described of the analysis alarm server transmission of cloud platform 22
The corresponding density of stream of people distribution of behavioural characteristic in region, the corresponding stream of people's movement locus of behavioural characteristic, the corresponding people of behavioural characteristic
At least one of number is simultaneously shown.
Can be seen that system provided by the present application by the information processing system above based on image analysis can be based on sea
The quick, intelligent people information for extracting each region of the image data of amount, and believed according to the generation management of people information intelligence
Breath, avoids artificial operation, improves the management operating efficiency for each region (such as market, tourist attraction etc.).
The corresponding above-mentioned information processing system based on image analysis, present invention also provides a kind of letters based on image analysis
Processing method is ceased, is applied to connect the cloud platform of multiple monitoring nodes.Fig. 3 shows provided by the embodiments of the present application based on image
The flow chart of the information processing method of analysis.As shown in figure 3, this approach includes the following steps:
Step 301:The image of the monitoring node corresponding region is received from monitoring node;Described image is by the monitoring node
In image capture device acquire in real time.
Here, the monitoring node includes one or more image capture device, to acquire above-mentioned monitoring node in real time
The image in the region at place, wherein the region can be public place, such as railway station, airport, scenic spot, market, subway
Deng described image includes picture and/or video etc..
Step 302:Individual identification processing is carried out to described image, described in the result determination handled according to the individual identification
The associated people information in region.
In some instances, carrying out individual identification processing to described image includes:Characteristics of human body's knowledge is carried out to described image
Not and/or individual behavior is analyzed.
In some instances, first nerves network is trained by multiple sample images, obtains being used for characteristics of human body
The first nerves network model of identification;Wherein, when it is described to described image carry out individual identification processing include to described image into
It is described to include to described image progress characteristics of human body's identification when pedestrian's body characteristics identify:Described image is inputted into first god
Through network model;According to the output of the first nerves network model, characteristics of human body's information in described image is determined.
In some instances, characteristics of human body's identification includes Head recognition, characteristics of human body's packet in described image
Include number of people information;Wherein, the result handled according to the individual identification determines the associated people information in the region, including:Root
According to the number of people information in described image, the number and/or density of stream of people data in the region are determined;According in the region
Number and/or density of stream of people data, determine and the associated people information in the region, the people information include:Density of stream of people
At least one of distribution, the distribution of stream of people's movement locus, number.
In some instances, characteristics of human body's identification includes recognition of face, characteristics of human body's packet in described image
Include face characteristic information;Wherein, the result handled according to the individual identification determines the associated people information in the region, wraps
It includes:According to the face characteristic information in described image, the identification information of target individual in the region, the target individual are determined
Identification information be used to indicate whether that there are the target individuals;According to the identification information of target individual in the region, determine
With the associated people information in the region, the people information includes:The position of target individual and/or the movement rail of target individual
Mark.
In some instances, nervus opticus network is trained by multiple sample images, obtains being used for individual behavior
The nervus opticus network model of analysis;Wherein, when it is described to described image carry out individual identification processing include to described image into
It is described to include to described image progress individual behavior analysis when row individual behavior is analyzed:Described image is inputted into second god
Through network model;According to the output of the nervus opticus network model, behavioural characteristic individual in described image is determined.
In some instances, the result handled according to the individual identification determines the associated people information in the region, wraps
It includes:According to behavioural characteristic individual in described image, the behavioural characteristic of each individual in the region is determined;According to the region
In each individual behavioural characteristic, determine with the associated people information in the region, the people information includes:Behavioural characteristic pair
At least one of the density of stream of people distribution answered, the corresponding stream of people's movement locus of behavioural characteristic, the corresponding number of behavioural characteristic.
Step 303:It is generated and the associated management information in the region according to the people information.
In some instances, the management information includes:In stream of people's statistical information, warning information, management strategy at least
One.
In some instances, when the management information includes at least stream of people's statistical information, the method is further
Including:According in the people information density of stream of people distributed data and/or number distributed data draw the stream of people in the region
Density Distribution plan view and/or number distribution plane figure;In response to the displaying request of the management node, it is close to send the stream of people
Distribution plane figure and/or number distribution plane figure are spent to the management node.
In some instances, when the management information includes at least stream of people's statistical information, the method is further
Including:Stream of people's movement locus figure in the region is drawn according to stream of people's motion trace data in the people information;Response
It is asked in the displaying of the management node, sends stream of people's movement locus figure to the management node.
In some instances, when the management information includes at least warning information, the method further includes:If
Number in the region is more than the density of stream of people distributed data in first threshold and/or the region and is more than second threshold, raw
At crowded warning information, the crowded warning information is sent to the management node.
In some instances, when the management information includes at least warning information, by the target in the people information
The position of individual and/or the movement locus of target individual are sent to the management node.
In some instances, when the management information includes at least stream of people's statistical information, in response to the management node
Displaying request, by the corresponding density of stream of people distribution of the behavioural characteristic in the region, the corresponding stream of people's movement locus of behavioural characteristic,
At least one of corresponding number of behavioural characteristic is sent to the management node.
In some instances, the face characteristic information according in described image, determines target individual in the region
Identification information, including:By the people in each face information record in the face characteristic information and face database in described image
Face characteristic information is matched, wherein the face information record includes name, the people of the corresponding people of the face characteristic information
Face characteristic information;If the people in face information record in the face characteristic information and the face database in described image
Face characteristic information matches, and determines the identification information of target individual in the region.
In some instances, the face information record further comprises affiliated library classification;When the face in described image
When the face characteristic information in face information record in characteristic information and the face database matches, according to the face
Affiliated library classification in information record determines the affiliated library classification of the face characteristic information in described image;According to determining
Affiliated library classification generates matching warning information, and the matching warning information is sent to the management node.
In some instances, each face information note in the face characteristic information and face database by described image
Face characteristic information in record is matched, including:If the face characteristic information in described image is remembered with the face information
The similarity of face characteristic information in record is more than first threshold, determines the face characteristic information in described image and the face
Face characteristic information in information record matches.
Further, when determining that the face in face characteristic information in described image and face information record is special
When levying information match, acquisition instructions are sent to the monitoring node, so that it acquires the face characteristic information in described image
The image of corresponding target individual.
Step 304:The associated management information in the region is sent to the corresponding management node in the region.
Here, the management node can be a management platform, such as the security administration platform of public security system, can also
It is a device end, such as the client of market administrator.
Can be seen that method provided by the present application by the information processing method above based on image analysis can be based on sea
The quick, intelligent people information for extracting each region of the image data of amount, and believed according to the generation management of people information intelligence
Breath, avoids artificial operation, improves the management operating efficiency for each region (such as market, tourist attraction etc.).
Below by taking the corresponding region of the monitoring node is a market as an example, at the above-mentioned information based on image analysis
Reason method is described in detail.This method is applied to implementation environment as shown in Figure 1.Fig. 4 shows above-mentioned based on image analysis
Information processing method flow, as shown in figure 4, this method mainly includes the following steps that:
Step 401:Monitoring node acquires the image of the monitoring node corresponding region, and the area that will be acquired in real time
The image in domain is sent to cloud platform and management node.
As previously mentioned, by taking the corresponding region of the monitoring node is a market as an example, therefore in the present embodiment, the area
Domain is market, and the client that the management node is market administrator.And the cloud platform is connected in the network in market
Each monitoring node of deployment, the monitoring node include one or more image capture devices and Internet of Things communication module 201,
Image can be acquired by image capture device, image is sent by Internet of Things communication module 201.
Step 402:The individual identification server of the cloud platform receives the image of the monitoring node corresponding region.
Step 403a:The individual identification server of the cloud platform carries out personal feature identification to described image, with determination
Characteristics of human body's information in described image.
In some instances, above-mentioned individual identification server instructs first nerves network by multiple sample images
Practice, obtains the first nerves network model identified for personal feature;Wherein, described image is carried out at individual identification when described
Reason includes when carrying out personal feature identification to described image, and described identified to described image progress personal feature includes:It will be described
Image inputs the first nerves network model;According to the output of the first nerves network model, determine in described image
Personal feature information, and the personal feature information in described image is sent to the analysis alarm server of the cloud platform.It is right
It includes Head recognition and recognition of face that described image, which carries out personal feature identification, and therefore, this step includes two sub-steps 4031
With sub-step 4032:
Step 4031:Above-mentioned individual identification server determines the number of people information in described image, and will be in described image
Number of people information is sent to the analysis alarm server.
Step 4032:Above-mentioned individual identification server can identify the face characteristic information in described image, and will be described
Face characteristic information in image is sent to the analysis alarm server.
Step 403b:The individual identification server of the cloud platform carries out individual behavior analysis to described image, determines institute
State behavioural characteristic individual in image.
In some instances, above-mentioned individual identification server instructs nervus opticus network by multiple sample images
Practice, obtains the nervus opticus network model analyzed for individual behavior;Described image is inputted into the nervus opticus network model;
According to the output of the nervus opticus network model, behavioural characteristic individual in described image is determined, and will be in described image
The behavioural characteristic of individual is sent to the analysis alarm server.
In some instances, the analysis alarm server is further according to behavioural characteristic individual in described image, really
The behavioural characteristic of each individual in the fixed region;According to the behavioural characteristic of each individual in the region, determine and the area
The associated people information in domain, the people information include:The corresponding density of stream of people distribution of behavioural characteristic, the corresponding people of behavioural characteristic
Flow at least one of movement locus, the corresponding number of behavioural characteristic.
It should be noted that step 403a~step 403b can synchronize execution, successively can not also sequentially distinguish
It executes.
Step 404a:The analysis alarm server of the cloud platform determines institute according to characteristics of human body's information in described image
State the associated people information in region.
In this step, since characteristics of human body's information includes number of people information and face characteristic information, this step includes two
Sub-steps 4041 and sub-step 4042:
Step 4041:The analysis alarm server determines the area further according to the number of people information in described image
Number in domain and/or density of stream of people data;And further according to the number and/or density of stream of people data in the region, really
The fixed and associated people information in the region.
Step 4042:The analysis alarm server determines institute further according to the face characteristic information in described image
The identification information of target individual in region is stated, the identification information of the target individual is used to indicate whether that there are the targets
Body;According to the identification information of target individual in the region, determine and the associated people information in the region, the people information
Including:The position of target individual and/or the movement locus of target individual.
Step 404b:The analysis alarm server of the cloud platform is determined according to the behavioural characteristic of the individual in described image
The associated people information in region.
Step 405a:The analysis alarm server of the cloud platform is according to the density of stream of people distribution number in the people information
According to and/or number distributed data draw density of stream of people distribution plane figure and/or number distribution plane figure in the region;In response to
The displaying of the management node is asked, and above-mentioned analysis alarm server sends the density of stream of people distribution plane figure and/or number
Distribution plane figure is to the management node.
In this step, the cloud platform sends density of stream of people distribution plane figure and/or number point in the region drawn
Cloth plan view to market administrative staff client.
Step 405b:The analysis alarm server of the cloud platform is according to stream of people's movement locus number in the people information
According to the stream of people's movement locus figure drawn in the region;In response to the displaying request of the management node, stream of people's fortune is sent
Trajectory diagram is moved to the management node.
Step 405c:When the number in the region is more than first threshold and/or when certain density of stream of people in the region
When more than second threshold, the analysis alarm server of the cloud platform generates crowded warning information, and the crowded alarm is believed
Breath is sent to the management node, so that related personnel takes corresponding measure in time.
In this step, the cloud platform sends the crowded warning information that generates to the client of market administrative staff,
So that it is shown
In other examples, when a certain face information record in the face recognition result and face database matches
When, acquisition instructions are sent to the monitoring node, so that it acquires the image of the corresponding target individual of the face recognition result
To collect evidence.
Step 405d:When the management information includes at least stream of people's statistical information, by the target in the people information
The position of individual and/or the movement locus of target individual are sent to the management node.
In this step, by the people in each face information record in the face characteristic information and face database in described image
Face characteristic information is matched, wherein the face information record includes name, the people of the corresponding people of the face characteristic information
Face characteristic information;If the people in face information record in the face characteristic information and the face database in described image
Face characteristic information matches, and determines the identification information of target individual in the region.
Further, when face information record further comprises affiliated library classification, if in described image
The face characteristic information in face information record in face characteristic information and the face database matches, according to the people
Affiliated library classification in face information record determines the affiliated library classification of the face characteristic information in described image;According to determining institute
Library classification belonging to stating generates matching warning information, and the matching warning information is sent to the management node.
In this step, the analysis alarm server of the cloud platform sends the matching warning information to the market generated and manages
The client of reason personnel.
In some instances, if a certain face information record in above-mentioned face characteristic information and face database matches
When, above-mentioned analysis alarm server sends acquisition instructions to monitoring node, so that its described face characteristic information of acquisition is corresponding
The image of target individual is to collect evidence.
Step 405e:The analysis alarm server of the cloud platform is asked in response to the displaying of the management node, by institute
It is corresponding to state the corresponding density of stream of people distribution of behavioural characteristic in region, the corresponding stream of people's movement locus of behavioural characteristic, behavioural characteristic
At least one of number is sent to the management node so that market administrative staff formulate corresponding operation and management strategy.
The analysis alarm server of the cloud platform checks request in response to management node, sends the determining region
Behavioural characteristic and people information corresponding with behavior feature, for example density of stream of people distribution, stream of people's movement locus and number are extremely
The client of market administrative staff.
In some instances, the analysis alarm server of the cloud platform is asked in response to the displaying of management node, will be true
Fixed target group and its behavioural characteristic are sent to the client of market administrative staff, for it with reference to formulate operation and management plan
Slightly.
It should be noted that step 405a~step 405e can synchronize execution, successively can not also sequentially distinguish
It executes.
Step 406a:The management node shows that density of stream of people distribution plane figure and/or number distribution are flat in the region
Face figure.
Step 406b:The management node shows stream of people's movement locus figure in the region.
Step 406c:The management node shows and plays the crowded warning information.
Step 406d:The management node shows and plays the matching warning information.
Step 406e:The management node shows behavioural characteristic and the stream of people corresponding with behavior feature in the region
Information.
It should be noted that step 406a~step 406e can synchronize execution, successively can not also sequentially distinguish
It executes.
In some instances, the monitoring node acquires the target individual in response to the acquisition instructions of the cloud platform
Image to collect evidence.
Present invention also provides a kind of information processing unit based on image analysis, it is applied to connect multiple monitoring nodes
The information processing method based on image analysis of the application proposition may be implemented in cloud platform.Fig. 5 shows that the embodiment of the present application carries
The structural schematic diagram of the information processing unit based on image analysis supplied.As shown in figure 5, the device includes receiving module 501, knows
Other module 502, analysis module 503 and sending module 504, wherein the function of each module is as follows:
Receiving module 501 receives the image of the monitoring node corresponding region from the monitoring node;Described image is by described
Image capture device in monitoring node acquires in real time;
Identification module 502 carries out individual identification processing to described image, is determined according to the result that the individual identification is handled
The associated people information in region;
Analysis module 503 generates and the associated management information in the region according to the people information;
Sending module 504 sends the associated management information in the region to the corresponding management node in the region.
Fig. 6 shows a kind of composite structural diagram of computing device 600.Individual identification can be installed on the computing device 600
Module 202, using the cloud platform as the information processing system based on image analysis, can also be equipped with analysis alarm module 203
Remote monitoring module 204 is also equipped with using the management node as the information processing system based on image analysis based on figure
As the function module of the information processing unit of analysis.As shown in fig. 6, the computing device includes one or more processor
(CPU) 602, communication module 604, memory 606, user interface 610, and the communication bus 608 for interconnecting these components.
Processor 602 can send and receive data to realize network communication and/or local communication by communication module 604.
User interface 610 includes one or more output equipments 612 comprising one or more speakers and/or one
Or multiple visual displays.User interface 610 also includes one or more input equipments 614 comprising such as, keyboard, mouse
Mark, voice command input unit or loudspeaker, touch screen displays, touch sensitive tablet, posture capture camera or other inputs are pressed
Button or control etc..
Memory 606 can be high-speed random access memory, such as DRAM, SRAM, DDR RAM or other deposit at random
Take solid storage device;Or nonvolatile memory, such as one or more disk storage equipments, optical disc memory apparatus, sudden strain of a muscle
Deposit equipment or other non-volatile solid-state memory devices.
Memory 606 stores the executable instruction set of processor 602, including:
Operating system 616 includes the program for handling various basic system services and for executing hardware dependent tasks;
Include the various application programs for the information processing based on image analysis, this application program can using 618
It realizes the process flow in above-mentioned each example, for example may include the information processing system based on image analysis shown in Fig. 2 a
All or part of module of unit or the shown information processing units based on image analysis of Fig. 5 some or all of in 200.Respectively
At least one of unit or module 202-203 or 204 module can be stored with machine-executable instruction, each module 501-504
At least one of module can be stored with machine-executable instruction.Processor 602 is by executing each module in memory 606
At least one mould machine-executable instruction in the block in 202-203 or 204 or 501-504, and then can realize above-mentioned each module
The function of 202-203 or at least one of 204 or 501-504 module.
It should be noted that step and module not all in above-mentioned each flow and each structure chart is all necessary, it can
To ignore certain steps or module according to the actual needs.Each step execution sequence be not it is fixed, can as needed into
Row adjustment.The division of each module is intended merely to facilitate the division functionally that description uses, and in actual implementation, a module can
It is realized by multiple modules with point, the function of multiple modules can also be realized by the same module, these modules can be located at same
In a equipment, it can also be located in different equipment.
Hardware module in each embodiment can in hardware or hardware platform adds the mode of software to realize.Above-mentioned software
Including machine readable instructions, it is stored in non-volatile memory medium.Therefore, each embodiment can also be presented as software product.
Therefore, some embodiments of the present application additionally provide a kind of computer readable storage medium, are stored thereon with calculating
Machine instructs, wherein the computer instruction realizes the step of any figure the method in above-mentioned Fig. 3-4 when being executed by processor.
In each example, hardware can be by special hardware or the hardware realization of execution machine readable instructions.For example, hardware can be with
Permanent circuit or logical device (such as application specific processor, such as FPGA or ASIC) specially to design are used to complete specifically to grasp
Make.Hardware can also include programmable logic device or circuit by software provisional configuration (as included general processor or other
Programmable processor) for executing specific operation.
In addition, each embodiment of the application can pass through the data processing journey by data processing equipment such as computer execution
Sequence is realized.Obviously, data processor constitutes the application.In addition, at the data being generally stored inside in a storage medium
Reason program by program by directly reading out storage medium or by installing or copying to depositing for data processing equipment by program
It stores up in equipment (such as hard disk and/or memory) and executes.Therefore, such storage medium also constitutes the application, present invention also provides
A kind of non-volatile memory medium, wherein being stored with data processor, this data processor can be used for executing the application
Any one of above method embodiment embodiment.
The corresponding machine readable instructions of module in Fig. 2 a, 2b and Fig. 5 can make the operating system operated on computer
Etc. completing some or all of operation described herein.Non-volatile computer readable storage medium storing program for executing can be inserted into computer
In set memory or the memory being arranged in the expanding element being connected with computer is write in interior expansion board.Peace
CPU in expansion board or expanding element etc. can be according to instruction execution part and whole practical operations.
In addition, the device and each module in each embodiment of the application can be integrated in a processing unit, also may be used
It, can also be during two or more devices or module be integrated in one unit to be that modules physically exist alone.It is above-mentioned
The form that hardware had both may be used in integrated unit is realized, can also be realized in the form of SFU software functional unit.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
With within principle, any modification, equivalent substitution, improvement and etc. done should be included within the scope of protection of the invention god.
Claims (19)
1. a kind of information processing method based on image analysis, which is characterized in that be applied to cloud platform, the cloud platform connects net
The each monitoring node disposed in network, the method includes:
The image of the monitoring node corresponding region is received from the monitoring node;Described image is by the image in the monitoring node
Collecting device acquires in real time;
Individual identification processing is carried out to described image, the result handled according to the individual identification determines the associated people in the region
Stream information;
It is generated and the associated management information in the region according to the people information;
The associated management information in the region is sent to the corresponding management node in the region.
2. according to the method described in claim 1, wherein, described handle described image progress individual identification includes:To described
Image carries out characteristics of human body's identification and/or individual behavior analysis.
3. described to include with the associated people information in the region according to the method described in claim 1, wherein:The region
The stream of people of any subregion in interior people information, the people information of any region set comprising the region, the region
At least one of information;
It is described to include with the associated management information in the region:Management information in the region, any for including the region
At least one of the management information of any subregion in the management information of regional ensemble, the region.
4. according to the method described in claim 2, further comprising:First nerves network is instructed by multiple sample images
Practice, obtains the first nerves network model identified for characteristics of human body;
Wherein, when it is described to described image carry out individual identification processing include to described image carry out characteristics of human body's identification when, institute
It states and includes to described image progress characteristics of human body's identification:
Described image is inputted into the first nerves network model;
According to the output of the first nerves network model, characteristics of human body's information in described image is determined.
5. according to the method described in claim 2, wherein, characteristics of human body's identification includes Head recognition, in described image
Characteristics of human body's information includes number of people information;
Wherein, the result handled according to the individual identification determines the associated people information in the region, including:
According to the number of people information in described image, the number and/or density of stream of people data in the region are determined;
According to the number and/or density of stream of people data in the region, determine and the associated people information in the region, the people
Stream information includes:At least one of density of stream of people distribution, the distribution of stream of people's movement locus, number.
6. according to the method described in claim 2, wherein, characteristics of human body's identification includes recognition of face, in described image
Characteristics of human body's information includes face characteristic information;
Wherein, the result handled according to the individual identification determines the associated people information in the region, including:
According to the face characteristic information in described image, the identification information of target individual in the region, the target are determined
The identification information of body is used to indicate whether that there are the target individuals;
According to the identification information of target individual in the region, determine and the associated people information in the region, stream of people's letter
Breath includes:The position of target individual and/or the movement locus of target individual.
7. according to the method described in claim 2, further comprising:Nervus opticus network is instructed by multiple sample images
Practice, obtains the nervus opticus network model analyzed for individual behavior;
Wherein, when it is described to described image carry out individual identification processing include to described image carry out individual behavior analysis when, institute
It states and includes to described image progress individual behavior analysis:
Described image is inputted into the nervus opticus network model;
According to the output of the nervus opticus network model, behavioural characteristic individual in described image is determined.
8. according to the method described in claim 2, wherein, the result handled according to the individual identification determines that the region is associated with
People information, including:
According to behavioural characteristic individual in described image, the behavioural characteristic of each individual in the region is determined;
According to the behavioural characteristic of each individual in the region, determine and the associated people information in the region, stream of people's letter
Breath includes:The corresponding density of stream of people distribution of behavioural characteristic, the corresponding stream of people's movement locus of behavioural characteristic, the corresponding people of behavioural characteristic
At least one of number.
9. according to the method described in claim 1, wherein, the management information includes:Stream of people's statistical information, warning information, pipe
Manage at least one of strategy.
10. according to the method described in claim 9, wherein, when the management information includes at least stream of people's statistical information,
The method further includes:
According in the people information density of stream of people distributed data and/or number distributed data to draw the stream of people in the region close
Spend distribution plane figure and/or number distribution plane figure;
In response to the displaying request of the management node, the density of stream of people distribution plane figure and/or number distribution plane are sent
Figure is to the management node.
11. according to the method described in claim 9, wherein, when the management information includes at least stream of people's statistical information,
The method further includes:
Stream of people's movement locus figure in the region is drawn according to stream of people's motion trace data in the people information;
In response to the displaying request of the management node, stream of people's movement locus figure is sent to the management node.
12. according to the method described in claim 9, wherein, when the management information includes at least warning information, the method
Further comprise:
If the number in the region is more than the density of stream of people distributed data in first threshold and/or the region and is more than second
Threshold value generates crowded warning information, and the crowded warning information is sent to the management node.
13. according to the method described in claim 6, wherein, the face characteristic information according in described image, determine described in
The identification information of target individual in region, including:
By in the face characteristic information and face database in described image each face information record in face characteristic information into
Row matching, wherein a face information record includes the mark and face characteristic information of target individual;
If the face characteristic in face information record in the face characteristic information and the face database in described image
Information match then generates and indicates that there are the mark letters that this face information records the corresponding target individual in the region
Breath.
14. according to the method for claim 13, wherein the face information record further comprises affiliated library classification;
Wherein, if in face information record in face characteristic information and the face database in the described image
Face characteristic information matches, then generates and indicate that there are this face informations to record the corresponding target individual in the region
Identification information, including:
Face characteristic letter in the face information record in the face characteristic information and the face database in described image
When manner of breathing matches, the affiliated library classification in being recorded according to the face information determines the institute of the face characteristic information in described image
Belong to library classification;
Matching warning information is generated according to the determining affiliated library classification, and the matching warning information is sent to the pipe
Manage node.
15. according to the method described in claim 9, wherein, when the management information includes at least warning information, by the people
The position of target individual in stream information and/or the movement locus of target individual are sent to the management node.
16. according to the method described in claim 9, wherein, when the management information includes at least stream of people's statistical information, responding
It is asked in the displaying of the management node, the corresponding density of stream of people distribution of the behavioural characteristic in the region, behavioural characteristic is corresponded to
Stream of people's movement locus, at least one of the corresponding number of behavioural characteristic be sent to the management node.
17. according to the method for claim 13, wherein the method further includes:
When determining that the face characteristic information in the face characteristic information and face information record in described image matches,
Acquisition instructions are sent to the monitoring node, so that it acquires the corresponding target individual of face characteristic information in described image
Image.
18. a kind of information processing unit based on image analysis, which is characterized in that be applied to cloud platform, the cloud platform connection
The each monitoring node disposed in network, described device include:
Receiving module receives the image of the monitoring node corresponding region from the monitoring node;Described image is saved by the monitoring
Image capture device in point acquires in real time;
Identification module carries out individual identification processing to described image, and the result handled according to the individual identification determines the area
The associated people information in domain;
Analysis module generates and the associated management information in the region according to the people information;
Sending module sends the associated management information in the region to the corresponding management node in the region.
19. a kind of storage medium, which is characterized in that be stored with machine readable instructions, at least one processor can be made to execute such as
Claim 1-17 any one of them methods.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810128166.XA CN108345854A (en) | 2018-02-08 | 2018-02-08 | Information processing method, device, system based on image analysis and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810128166.XA CN108345854A (en) | 2018-02-08 | 2018-02-08 | Information processing method, device, system based on image analysis and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108345854A true CN108345854A (en) | 2018-07-31 |
Family
ID=62958651
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810128166.XA Pending CN108345854A (en) | 2018-02-08 | 2018-02-08 | Information processing method, device, system based on image analysis and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108345854A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109117788A (en) * | 2018-08-10 | 2019-01-01 | 重庆大学 | A kind of public transport compartment crowding detection method merging ResNet and LSTM |
CN110188644A (en) * | 2019-05-22 | 2019-08-30 | 广东寰球智能科技有限公司 | A kind of staircase passenger's hazardous act monitoring system and method for view-based access control model analysis |
CN110837829A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Control method and system of sweeping robot |
CN110968020A (en) * | 2019-12-13 | 2020-04-07 | 微创(上海)网络技术股份有限公司 | System operation monitoring visual platform |
CN111292502A (en) * | 2020-02-17 | 2020-06-16 | 浙江大华技术股份有限公司 | Method and device for preventing early warning of lost and storage medium |
CN111310564A (en) * | 2020-01-13 | 2020-06-19 | 陈丙涛 | Signal triggering system and method for protecting pregnant woman |
CN111338275A (en) * | 2020-02-21 | 2020-06-26 | 江苏大量度电气科技有限公司 | Method and system for monitoring running state of electrical equipment |
CN111523624A (en) * | 2020-04-26 | 2020-08-11 | 京东方科技集团股份有限公司 | Crowd shunting guide system |
CN112418063A (en) * | 2020-11-19 | 2021-02-26 | 北京软通智慧城市科技有限公司 | Face recognition method and device, electronic equipment and storage medium |
CN113051975A (en) * | 2019-12-27 | 2021-06-29 | 深圳云天励飞技术有限公司 | People flow statistical method and related product |
CN113869779A (en) * | 2021-10-13 | 2021-12-31 | 深圳联和智慧科技有限公司 | People flow monitoring method and system based on city management and cloud platform |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101335864A (en) * | 2007-06-28 | 2008-12-31 | 当代天启技术(北京)有限公司 | Method and system for number of outdoor video receiving people statistic |
CN101445122A (en) * | 2008-12-25 | 2009-06-03 | 北京交通大学 | System of urban track traffic for passenger flow monitor and emergency evacuation |
CN101741658A (en) * | 2008-11-25 | 2010-06-16 | 北京中星微电子有限公司 | Video monitoring system and interaction method in same |
CN101753985A (en) * | 2008-11-27 | 2010-06-23 | 北京中星微电子有限公司 | Video monitoring system, and medium flow transmitting control device and method in the video monitoring system |
EP2362636A1 (en) * | 2010-02-26 | 2011-08-31 | Research In Motion Limited | Object detection and selection using gesture recognition |
CN105404860A (en) * | 2015-11-13 | 2016-03-16 | 北京旷视科技有限公司 | Method and device for managing information of lost person |
CN106650581A (en) * | 2016-09-27 | 2017-05-10 | 腾讯科技(深圳)有限公司 | Visitor flow rate statistics method and device |
CN106778637A (en) * | 2016-12-19 | 2017-05-31 | 江苏慧眼数据科技股份有限公司 | A kind of statistical method to men and women's passenger flow |
CN107067159A (en) * | 2017-03-09 | 2017-08-18 | 深圳华博高科光电技术有限公司 | Smart city management and dispatching plateform system |
US20180115751A1 (en) * | 2015-03-31 | 2018-04-26 | Westire Technology Limited | Smart city closed camera photocell and street lamp device |
-
2018
- 2018-02-08 CN CN201810128166.XA patent/CN108345854A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101335864A (en) * | 2007-06-28 | 2008-12-31 | 当代天启技术(北京)有限公司 | Method and system for number of outdoor video receiving people statistic |
CN101741658A (en) * | 2008-11-25 | 2010-06-16 | 北京中星微电子有限公司 | Video monitoring system and interaction method in same |
CN101753985A (en) * | 2008-11-27 | 2010-06-23 | 北京中星微电子有限公司 | Video monitoring system, and medium flow transmitting control device and method in the video monitoring system |
CN101445122A (en) * | 2008-12-25 | 2009-06-03 | 北京交通大学 | System of urban track traffic for passenger flow monitor and emergency evacuation |
EP2362636A1 (en) * | 2010-02-26 | 2011-08-31 | Research In Motion Limited | Object detection and selection using gesture recognition |
US20180115751A1 (en) * | 2015-03-31 | 2018-04-26 | Westire Technology Limited | Smart city closed camera photocell and street lamp device |
CN105404860A (en) * | 2015-11-13 | 2016-03-16 | 北京旷视科技有限公司 | Method and device for managing information of lost person |
CN106650581A (en) * | 2016-09-27 | 2017-05-10 | 腾讯科技(深圳)有限公司 | Visitor flow rate statistics method and device |
CN106778637A (en) * | 2016-12-19 | 2017-05-31 | 江苏慧眼数据科技股份有限公司 | A kind of statistical method to men and women's passenger flow |
CN107067159A (en) * | 2017-03-09 | 2017-08-18 | 深圳华博高科光电技术有限公司 | Smart city management and dispatching plateform system |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109117788A (en) * | 2018-08-10 | 2019-01-01 | 重庆大学 | A kind of public transport compartment crowding detection method merging ResNet and LSTM |
CN110837829A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Control method and system of sweeping robot |
CN110188644A (en) * | 2019-05-22 | 2019-08-30 | 广东寰球智能科技有限公司 | A kind of staircase passenger's hazardous act monitoring system and method for view-based access control model analysis |
CN110968020A (en) * | 2019-12-13 | 2020-04-07 | 微创(上海)网络技术股份有限公司 | System operation monitoring visual platform |
CN113051975B (en) * | 2019-12-27 | 2024-04-02 | 深圳云天励飞技术有限公司 | People flow statistics method and related products |
CN113051975A (en) * | 2019-12-27 | 2021-06-29 | 深圳云天励飞技术有限公司 | People flow statistical method and related product |
CN111310564B (en) * | 2020-01-13 | 2021-03-30 | 青岛大学附属医院 | Signal triggering system and method for protecting pregnant woman |
CN111310564A (en) * | 2020-01-13 | 2020-06-19 | 陈丙涛 | Signal triggering system and method for protecting pregnant woman |
CN111292502A (en) * | 2020-02-17 | 2020-06-16 | 浙江大华技术股份有限公司 | Method and device for preventing early warning of lost and storage medium |
CN111338275A (en) * | 2020-02-21 | 2020-06-26 | 江苏大量度电气科技有限公司 | Method and system for monitoring running state of electrical equipment |
CN111523624A (en) * | 2020-04-26 | 2020-08-11 | 京东方科技集团股份有限公司 | Crowd shunting guide system |
CN112418063A (en) * | 2020-11-19 | 2021-02-26 | 北京软通智慧城市科技有限公司 | Face recognition method and device, electronic equipment and storage medium |
CN113869779A (en) * | 2021-10-13 | 2021-12-31 | 深圳联和智慧科技有限公司 | People flow monitoring method and system based on city management and cloud platform |
CN113869779B (en) * | 2021-10-13 | 2023-06-13 | 深圳联和智慧科技有限公司 | Urban management-based people stream monitoring method, system and cloud platform |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108345854A (en) | Information processing method, device, system based on image analysis and storage medium | |
Gao et al. | Human action monitoring for healthcare based on deep learning | |
Kalra et al. | Dronesurf: Benchmark dataset for drone-based face recognition | |
CN107370983A (en) | Acquisition methods and device for the whereabouts track of video monitoring system | |
JP6267861B2 (en) | Usage measurement techniques and systems for interactive advertising | |
CN110363131B (en) | Abnormal behavior detection method, system and medium based on human skeleton | |
CN109299683B (en) | Security protection evaluation system based on face recognition and behavior big data | |
CN109819208A (en) | A kind of dense population security monitoring management method based on artificial intelligence dynamic monitoring | |
CN110263613A (en) | Monitor video processing method and processing device | |
WO2018035667A1 (en) | Display method and apparatus, electronic device, computer program product, and non-transient computer readable storage medium | |
CN112991656A (en) | Human body abnormal behavior recognition alarm system and method under panoramic monitoring based on attitude estimation | |
CN111312367A (en) | Campus personnel abnormal psychological prediction method based on self-adaptive cloud management platform | |
Gupta et al. | CrowdVAS-Net: A deep-CNN based framework to detect abnormal crowd-motion behavior in videos for predicting crowd disaster | |
US20100153597A1 (en) | Generating Furtive Glance Cohorts from Video Data | |
CN106296017A (en) | Management method that a kind of guest room is moved in and device | |
JP6764214B1 (en) | Congestion information notification system | |
Atrey et al. | Effective multimedia surveillance using a human-centric approach | |
CN110163154A (en) | Video monitoring system based on artificial intelligence | |
CN109684494A (en) | A kind of scenic spot looking-for-person method, system and cloud server | |
Navarathna et al. | Estimating audience engagement to predict movie ratings | |
CN106529500A (en) | Information processing method and system | |
CN109905423A (en) | A kind of intelligent management system | |
JP6744652B1 (en) | Congestion information notification system | |
Khan et al. | Classification of human's activities from gesture recognition in live videos using deep learning | |
CN109977856B (en) | Method for identifying complex behaviors in multi-source video |
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 |