CN110443161A - Monitoring method based on artificial intelligence under a kind of scene towards bank - Google Patents
Monitoring method based on artificial intelligence under a kind of scene towards bank Download PDFInfo
- Publication number
- CN110443161A CN110443161A CN201910652598.5A CN201910652598A CN110443161A CN 110443161 A CN110443161 A CN 110443161A CN 201910652598 A CN201910652598 A CN 201910652598A CN 110443161 A CN110443161 A CN 110443161A
- Authority
- CN
- China
- Prior art keywords
- voice
- early warning
- video
- suspected target
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 13
- 238000012544 monitoring process Methods 0.000 title claims abstract description 11
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 7
- 238000005457 optimization Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 206010000117 Abnormal behaviour Diseases 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 230000003542 behavioural effect Effects 0.000 claims description 2
- 238000009434 installation Methods 0.000 claims 1
- 238000000926 separation method Methods 0.000 abstract description 9
- 230000006870 function Effects 0.000 description 10
- 230000006399 behavior Effects 0.000 description 4
- 238000013480 data collection Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- 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
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/24—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being the cepstrum
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Acoustics & Sound (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Closed-Circuit Television Systems (AREA)
- Alarm Systems (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides the monitoring methods based on artificial intelligence under a kind of scene towards bank, in order to improve early warning accuracy, by the way of video image and voice two-stage early warning, video image carries out early warning to suspected target first, then it is directed to suspected target, using speech Separation technology, further confirm that whether suspected target needs early warning.During speech Separation, the optimization object function weighted using space encoding is conducive to the neighbouring relations on seeker's object space, is weighted by space encoding, the reliability of voice flow separation can be improved when coding by the way of Gray code.The present invention realizes simply, meets the needs of practical application.
Description
Technical field
The present invention relates to the monitoring methods based on artificial intelligence under a kind of scene towards bank.
Background technique
Bank be manage deposit, make loans, exchange, the business such as savings, undertake the financial institution of credit intermediary, be state key
Safety precaution unit, with scale diversification, financial services equipment is numerous, it is complicated to enter and leave personnel, it is wide etc. to manage coverage
Feature.But the miscellaneous criminal activity in recent years, within the scope of financial industry is commonplace.
The Activity recognition of people is always an important research direction of computer vision field, different in detection and identification video
Chang Hangwei has become a challenging hot research problem at present.It is traditional that rely primarily on security work person artificial
The random emergency event of monitor full time and suspicious event, need a large amount of manpower.Due to the template in video monitoring system
Classifier has no idea to construct all people's body posture, so, only by video image detect potential threat have compared with
It is big difficult.
Summary of the invention
The shortcomings that in view of the prior art, the present invention propose the monitoring side based on artificial intelligence under a kind of scene towards bank
Method, this method are applied under bank's scene, are equipped with video monitoring camera and sound pick-up, according to video and voice signal into
Row two-stage early warning, it is characterised in that:
Level-one video early warning: according to video image, suspected target is extracted, the specific steps are as follows:
1. detecting the human body in video based on the Background difference of gauss hybrid models, image background is removed；
2. extracting the target interbehavior feature of video using convolutional neural networks ALexNet for target, behavior is obtained
Characteristic probability value；
3. two hidden-layer network area partial objectives for normal behaviour and abnormal behaviour are utilized, to suspected target early warning；
Second level phonetic warning: being directed to suspected target, extracts speaker's voice, the specific steps are as follows:
1. mixing voice is resolved into time frequency unit；
2. marking by pitch tracking and time frequency unit, the pitch contour and corresponding voice flow of input signal are obtained；
3. extracting frequency cepstral coefficient (GFCC) eigenmatrix of mixing voice；
4. dividing region according to video image, encoded using Gray code, the optimization aim of design space coding weighting
Function, objective function form are as follows:
Wherein, L is voice flow quantity to be extracted, GrIt is gray encoding, g is class vector, and F is GFCC feature square
Battle array, WkIt (g) is that kth ties up component in class vector g, F value range is Wk(g), NUMk(g) and VkIt (g) is that kth is tieed up in g respectively
The element number and mean value of component GFCC eigenmatrix, C are the mean value of GFCC eigenmatrix, (*)TRepresenting matrix transposition；
5. the voice flow of cluster seeking majorized function maximum value combines；
6. according to voice flow, early warning.
In conclusion the present invention forecasts accuracy to improve alert under bank's scene, using video image and voice two
The mode of grade early warning, during voice flow separation, the optimization object function weighted using space encoding uses lattice when coding
The mode of thunder code is conducive to the neighbouring relations on seeker's object space, is weighted by space encoding, and voice flow separation can be improved
Reliability.Moreover, the present invention does not need to be trained voice data collection acquisition priori knowledge during speech Separation, it is real
It is now simple, high reliablity.
Detailed description of the invention
Fig. 1 is the flow chart of the embodiment of the present invention.
Specific embodiment
Illustrate embodiments of the present invention below by way of specific specific example, those skilled in the art can be by this specification
Disclosed content is implemented easily.
The present invention proposes that the monitoring method based on artificial intelligence under a kind of scene towards bank, this method are applied in bank
Under scene, video monitoring camera and sound pick-up are installed, two-stage early warning is carried out according to video and voice signal, the method is as follows:
Level-one video early warning: according to video image, extracting suspected target, using Ubuntu platform, by OpenCV and
The library TensorFlow, the specific steps are as follows:
1. detecting the human body in video based on the Background difference of gauss hybrid models, image background is removed, is utilized
The background subtraction of BackgroundSubtrctorMOG2 function realization video image；
2. extracting the target interbehavior feature of video using convolutional neural networks ALexNet for target, behavior is obtained
Characteristic probability value.ALexNet includes that 5 convolutional layers and 3 full articulamentums, the output of the last one full articulamentum are sent to
In softmax layers, behavioural characteristic probability value is the value of float type.Neural network convolution operation utilizes function conv_2d ()
It realizes.
3. two hidden-layer network area partial objectives for normal behaviour and abnormal behaviour are utilized, to suspected target early warning.It utilizes
TensorFlow deep learning platform realizes two hidden-layer network, completes the realization of two hidden-layer network and training pattern.
Second level phonetic warning: being directed to suspected target, extracts speaker's voice, the specific steps are as follows:
1. mixing voice is resolved into time frequency unit.By 64 Gammatone Superimposed Filters at bandpass filter group,
The centre frequency equidistantly distributed of each filter, the frequency coverage of entire filter group are 50Hz~5000Hz.Then, with
40ms is frame length, 20ms is that frame moves, and accordingly does time domain sub-frame processing to the filtering of each frequency channel.
2. marking by pitch tracking and time frequency unit, the pitch contour and corresponding voice flow of input signal are obtained.Base
Sound tracking uses viterbi algorithm, and fundamental tone observation probability is calculated by the significance of every frame candidate fundamental frequency, fundamental tone transition probability
Pitch variation rate by counting voice data set obtains, and probability is the observation probability of first frame in each voiced segments.Base
Sound tracking carries out in each voiced segments, finds out an optimal fundamental tone sequence.It is marked, is obtained by pitch tracking and time frequency unit
The pitch contour of input signal and voice flow while correspondence.Wherein, while voice flow is indicated with two-value mask, and 1 represents correspondence
Time frequency unit is labeled, and 0 indicates unmarked.
3. extracting frequency cepstral coefficient (GFCC) eigenmatrix of mixing voice.Language while by two-value mask and correspondence
Sound flows through filter feature unit, obtains by the unit of 1 label, the unit not being labeled is removed.For each frame, by acquisition
It is converted by the unit of 1 label by discrete cosine transform operation, ultimately forms the GFCC eigenmatrix of voice signal.
4. dividing region according to video image, encoded using Gray code, the optimization aim of design space coding weighting
Function, objective function form are as follows:
Wherein, L is voice flow quantity to be extracted, GrIt is gray encoding, g is class vector, and F is GFCC feature square
Battle array, WkIt (g) is that kth ties up component in class vector g, F value range is Wk(g), NUMk(g) and VkIt (g) is that kth is tieed up in g respectively
The element number and mean value of component GFCC eigenmatrix, C are the mean value of GFCC eigenmatrix, (*)TRepresenting matrix transposition.
Voice flow quantity L to be extracted is determined according to Gray code adjacency on geometry number, former during gray encoding
Then upper each personage corresponds to an individual Gray code, and it is reasonable that this requires image-region to divide.
5. by the method for exhaustion, the voice flow combination of cluster seeking majorized function maximum value.System starts first to choose at random
L unit in choosing while voice flow, is assigned in L classification, is then ranked up to the voice flow unit not being selected.
6. according to voice flow, early warning.
In conclusion the present invention forecasts accuracy to improve alert under bank's scene, using video image and voice two
The mode of grade early warning, during voice flow separation, the optimization object function weighted using space encoding uses lattice when coding
The mode of thunder code is conducive to the neighbouring relations on seeker's object space, is weighted by space encoding, and voice flow separation can be improved
Reliability.Moreover, the present invention does not need to be trained voice data collection acquisition priori knowledge during speech Separation, it is real
It is now simple, high reliablity.The present invention effectively overcomes various shortcoming in the prior art and has height application value.
Claims (1)
1. the monitoring method based on artificial intelligence under a kind of scene towards bank, this method is applied under bank's scene, installation
There are video monitoring camera and sound pick-up, two-stage early warning carried out according to video and voice signal, it is characterised in that:
Level-one video early warning: according to video image, suspected target is extracted, the specific steps are as follows:
1. detecting the human body in video based on the Background difference of gauss hybrid models, image background is removed；
2. extracting the target interbehavior feature of video using convolutional neural networks ALexNet for target, behavioural characteristic is obtained
Probability value；
3. two hidden-layer network area partial objectives for normal behaviour and abnormal behaviour are utilized, to suspected target early warning；
Second level phonetic warning: being directed to suspected target, extracts speaker's voice, the specific steps are as follows:
1. mixing voice is resolved into time frequency unit；
2. marking by pitch tracking and time frequency unit, the pitch contour and corresponding voice flow of input signal are obtained；
3. extracting the frequency cepstral coefficient eigenmatrix of mixing voice；
4. dividing region according to video image, encoded using Gray code, the optimization object function of design space coding weighting,
Objective function form are as follows:
Wherein, L is voice flow quantity to be extracted, GrIt is gray encoding, g is class vector, and F is frequency cepstral coefficient feature
Matrix, WkIt (g) is that kth ties up component in class vector g, F value range is Wk(g), NUMk(g) and VkIt (g) is that kth is tieed up in g respectively
The element number and mean value of component frequencies cepstrum coefficient eigenmatrix, C are the mean value of frequency cepstral coefficient eigenmatrix, (*)T
Representing matrix transposition；
5. the voice flow of cluster seeking majorized function maximum value combines；
6. according to voice flow, early warning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910652598.5A CN110443161B (en) | 2019-07-19 | 2019-07-19 | Monitoring method based on artificial intelligence in banking scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910652598.5A CN110443161B (en) | 2019-07-19 | 2019-07-19 | Monitoring method based on artificial intelligence in banking scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110443161A true CN110443161A (en) | 2019-11-12 |
| CN110443161B CN110443161B (en) | 2023-08-29 |
Family
ID=68429779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910652598.5A Active CN110443161B (en) | 2019-07-19 | 2019-07-19 | Monitoring method based on artificial intelligence in banking scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110443161B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116475905A (en) * | 2023-05-05 | 2023-07-25 | 浙江闽立电动工具有限公司 | Control system and method for angle grinder |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902972A (en) * | 2012-09-14 | 2013-01-30 | 成都国科海博计算机系统有限公司 | Human behavior characteristic extraction method and system and abnormal behavior detection method and system |
| US8704668B1 (en) * | 2005-04-20 | 2014-04-22 | Trevor Darrell | System for monitoring and alerting based on animal behavior in designated environments |
| CN105426820A (en) * | 2015-11-03 | 2016-03-23 | 中原智慧城市设计研究院有限公司 | Multi-person abnormal behavior detection method based on security monitoring video data |
| CN106295470A (en) * | 2015-05-21 | 2017-01-04 | 北京文安智能技术股份有限公司 | A kind of bank self-help service area early-warning monitoring method, Apparatus and system |
| CN106373430A (en) * | 2016-08-26 | 2017-02-01 | 华南理工大学 | Intersection pass early warning method based on computer vision |
| CN108052859A (en) * | 2017-10-31 | 2018-05-18 | 深圳大学 | A kind of anomaly detection method, system and device based on cluster Optical-flow Feature |
| US20180357414A1 (en) * | 2017-06-07 | 2018-12-13 | International Business Machines Corporation | Cognitive learning to counter security threats for kinematic actions in robots |
| CN109447048A (en) * | 2018-12-25 | 2019-03-08 | 苏州闪驰数控系统集成有限公司 | A kind of artificial intelligence early warning system |
-
2019
- 2019-07-19 CN CN201910652598.5A patent/CN110443161B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8704668B1 (en) * | 2005-04-20 | 2014-04-22 | Trevor Darrell | System for monitoring and alerting based on animal behavior in designated environments |
| CN102902972A (en) * | 2012-09-14 | 2013-01-30 | 成都国科海博计算机系统有限公司 | Human behavior characteristic extraction method and system and abnormal behavior detection method and system |
| CN106295470A (en) * | 2015-05-21 | 2017-01-04 | 北京文安智能技术股份有限公司 | A kind of bank self-help service area early-warning monitoring method, Apparatus and system |
| CN105426820A (en) * | 2015-11-03 | 2016-03-23 | 中原智慧城市设计研究院有限公司 | Multi-person abnormal behavior detection method based on security monitoring video data |
| CN106373430A (en) * | 2016-08-26 | 2017-02-01 | 华南理工大学 | Intersection pass early warning method based on computer vision |
| US20180357414A1 (en) * | 2017-06-07 | 2018-12-13 | International Business Machines Corporation | Cognitive learning to counter security threats for kinematic actions in robots |
| CN108052859A (en) * | 2017-10-31 | 2018-05-18 | 深圳大学 | A kind of anomaly detection method, system and device based on cluster Optical-flow Feature |
| CN109447048A (en) * | 2018-12-25 | 2019-03-08 | 苏州闪驰数控系统集成有限公司 | A kind of artificial intelligence early warning system |
Non-Patent Citations (2)
| Title |
|---|
| AMIRA BEN MABROUK.ET.: "Abnormal behavior recognition for intelligent video surveillance systems: A review", vol. 91, pages 480 - 491, XP085229786, DOI: 10.1016/j.eswa.2017.09.029 * |
| 冯亚闯: "视频中的异常事件检测算法研究", no. 4, pages 136 - 10 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116475905A (en) * | 2023-05-05 | 2023-07-25 | 浙江闽立电动工具有限公司 | Control system and method for angle grinder |
| CN116475905B (en) * | 2023-05-05 | 2024-01-09 | 浙江闽立电动工具有限公司 | Control system and method for angle grinder |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110443161B (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108709633B (en) | Distributed optical fiber vibration sensing intelligent safety monitoring method based on deep learning | |
| CN105022835B (en) | A kind of intelligent perception big data public safety recognition methods and system | |
| CN106778595B (en) | Method for detecting abnormal behaviors in crowd based on Gaussian mixture model | |
| CN105095856B (en) | Face identification method is blocked based on mask | |
| CN110070530A (en) | A kind of powerline ice-covering detection method based on deep neural network | |
| CN111339883A (en) | Method for identifying and detecting abnormal behaviors in transformer substation based on artificial intelligence in complex scene | |
| CN107491749B (en) | Method for detecting global and local abnormal behaviors in crowd scene | |
| CN113707175B (en) | Acoustic event detection system based on feature decomposition classifier and adaptive post-processing | |
| CN105574489A (en) | Layered stack based violent group behavior detection method | |
| CN110728252A (en) | Face detection method applied to regional personnel motion trail monitoring | |
| CN117671887B (en) | Intelligent security early warning management method and system based on big data | |
| CN117079351B (en) | Method and system for analyzing personnel behaviors in key areas | |
| CN110443161A (en) | Monitoring method based on artificial intelligence under a kind of scene towards bank | |
| CN108256551A (en) | A kind of vehicle checking method based on region convolutional neural networks | |
| CN117079197A (en) | Intelligent building site management method and system | |
| CN115240142B (en) | Outdoor key place crowd abnormal behavior early warning system and method based on cross media | |
| CN117037264A (en) | Prison personnel abnormal behavior identification method based on target and key point detection | |
| CN115171006B (en) | Detection method for automatically identifying person entering electric power dangerous area based on deep learning | |
| CN114358667B (en) | Scenic spot risk prediction model construction method based on RBF (radial basis function) network learning | |
| CN115393802A (en) | Railway scene unusual invasion target identification method based on small sample learning | |
| Nyajowi et al. | CNN real-time detection of vandalism using a hybrid-LSTM deep learning neural networks | |
| CN112329743B (en) | Abnormal body temperature monitoring method, device and medium in epidemic situation environment | |
| CN116401290B (en) | Personnel security inspection method based on metal carrying capacity data | |
| CN113177513B (en) | Method, device, equipment and storage medium for detecting wearing of safety helmet | |
| CN113743406B (en) | Deep learning-based personnel detection method for production safety |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |