CN110210527A - Maritime Law Enforcement reconnaissance system based on machine vision joint perception - Google Patents
Maritime Law Enforcement reconnaissance system based on machine vision joint perception Download PDFInfo
- Publication number
- CN110210527A CN110210527A CN201910399849.3A CN201910399849A CN110210527A CN 110210527 A CN110210527 A CN 110210527A CN 201910399849 A CN201910399849 A CN 201910399849A CN 110210527 A CN110210527 A CN 110210527A
- Authority
- CN
- China
- Prior art keywords
- target
- law enforcement
- perception
- photoelectricity
- deep learning
- 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
- 230000008447 perception Effects 0.000 title claims abstract description 50
- 230000005622 photoelectricity Effects 0.000 claims abstract description 39
- 238000013135 deep learning Methods 0.000 claims abstract description 30
- 238000013527 convolutional neural network Methods 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 9
- 238000003384 imaging method Methods 0.000 claims description 9
- 230000004927 fusion Effects 0.000 claims description 8
- 238000003331 infrared imaging Methods 0.000 claims description 6
- 230000010354 integration Effects 0.000 claims description 4
- 238000012790 confirmation Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 6
- 230000019771 cognition Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 abstract description 2
- 238000000034 method Methods 0.000 description 15
- 230000008569 process Effects 0.000 description 7
- 238000012545 processing Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000005764 inhibitory process Effects 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 239000003595 mist Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004587 chromatography analysis Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention devises a kind of Maritime Law Enforcement reconnaissance system based on machine vision joint perception, including preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus, Situation Awareness equipment, deep learning work station, preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus are located at law enforcement forward quarter, Background Region, naval target for front and rear scouts perception, deep learning work station is the calculating center of backstage target identification, for calculate preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus, Situation Awareness equipment data find naval target information.By multimode scouting and deep learning cognition technology means, reach accurate, rapid identification naval target characteristic information effect；It realizes that real time enhancing perceives under complex background environment, promotes the marine accurate reconnaissance capability in law enforcement naval vessel；Combine with multi-source naval target and scout perception, fulfillment capability, raising sea police's ship joint scouts evidence obtaining and complexity situation controls ability.
Description
Technical field
The invention belongs to ship electronic weapon technologies field more particularly to a kind of seas based on machine vision joint perception
Law enforcement reconnaissance system.
Background technique
According to the mission task that law enforcement naval vessels assign, in enforcing law, naval vessel faces Various Complex situation, especially prominent
Fast reaction situation in enforcing law is sent out, law enforcement naval vessel needs to make quick, reliable and precision strike finger for these situation
Wave control.
The current main problems faced of law enforcement ship naval reconnaissance task includes the following:
1) marine single mode identifies that target is unfavorable
Since maritime environment usually changes multiplicity, not seeing often occurs in law enforcement naval vessel perceptual image, it is not far to see, it is inaccurate to see
Situation.It at sea executes in task scene, is influenced by factors, is often difficult to only by single visible optical sensor
It was found that or target easy to be lost, discovery target can not effectively judge which be unfriendly target naval vessel, which be our naval vessel, which
It is civilian boat, and is identified often by modes such as artificial inquiry, micro-judgments, system intelligent working level is not high.
2) marine multiple target handles difficulty in real time
Sea climate environment is changeable in real time, environmental background is complicated, generally requires during execution task to machine vision sense
The multiple targets known are handled in time, along with the target identification for carrying out deep learning needs a large amount of concurrent collaboratives to calculate, are held
Method process needs to generate output recognition result in real time, how quickly to accurately identify the difficult point that perception is scouted as law enforcement ship.
3) marine complicated situation information integration capability is insufficient
Although the equipment such as pathfinder, search radar, photodetection in current law enforcement ship it can be found that surrounding naval vessel,
It is that there are inconsistent types for target information, it can not effective integration objective information.This law enforcement ship is accurately understood ambient enviroment,
Grasp live information situation, efficient execution task is totally unfavorable.
Summary of the invention
The technical problem to be solved by the invention is to provide a kind of Maritime Law Enforcement scoutings based on machine vision joint perception
There is system multi-source naval target joint to scout perception, fulfillment capability, improves sea police's ship joint and scouts evidence obtaining and the complicated situation palm
Control ability is commanded for army, police, public affair, the naval reconnaissance under complicated situation of law enforcement ship, can effectively be promoted and be held using the system
Method naval vessel situation information processing capacity.
The technical solution adopted by the present invention to solve the technical problems is: perceiving including preceding photoelectricity awareness apparatus, rear photoelectricity
Equipment, Situation Awareness equipment, deep learning work station, preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus are located at law enforcement ship
Front, Background Region, the naval target for front and rear scout perception, and Situation Awareness equipment is scouted for Maritime Law Enforcement
Joint perception and multi-source fusion calculate；Deep learning work station is used for preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus, situation sense
Know the identification learning and comparing of equipment output naval target information.
According to the above technical scheme, preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus are by visible light sensor, infrared imaging
Sensor, laser imaging sensor, laser range sensor are integrated among servo turntable.
According to the above technical scheme, the target image obtained in real time is passed through image preprocessing and image by deep learning work station
Enhancing, and target signature and the identification of image preliminary classification are extracted, the analysis of compare feature naval vessel library, and it is based on model ship library
Carry out deep learning, real-time perfoming target classification label and target identification.
According to the above technical scheme, deep learning work station strengthens target information confirmation herein in connection with radar, AIS, navigation system.
According to the above technical scheme, deep learning work station is based on convolutional neural networks (Convolutional Neural
Network, CNN) and depth convolutional neural networks (DCNN), construct multi-level neural network model.
According to the above technical scheme, parallel based on GPUs (Graphics Processing Units, more image processing units)
Collaboration carries out the image enhancement of multiple target.
According to the above technical scheme, on-board sensor will perceive maritime environment background image, tracked, identified, extract it is more
A target image target 1, target 2 ... target n are respectively created more in the deep learning work station computation processor course of work
A worker thread respectively executes the collaboration in each thread in each image processor GPUs (GPU1, GPU2 ..., GPUn)
It calculates, carries out the parallel computation of scan picture, it is multiple as a result, and carrying out sea that output is generated on deep learning work station
Study on Trend and subject fusion.
Particular technique principle of the present invention is as follows:
One, the naval target intelligent recognition of multimode joint perception and deep learning
The visible light signal of traditional naval target perception images in photodetector target through electro-optical system by propagation in atmosphere
Face is the complex process of an electrical image that extraterrestrial target two dimensional image is changed into time series.Due to sea climate environment
Changeable in real time, in single visible light sensor perception, target photoelectric image is interfered by marine atmosphere, light environment
Greatly, noise is more, easy to be lost, and the multiple photoelectric sensors acquisition visual image qualities of different weather conditions are irregular how
Accurate and excellent perception naval target true picture is the difficult point that law enforcement ship scouts perception.
This system combines electronic reconnaissance, electricity by a variety of photoelectricity perceptive modes such as visible light, infrared light, laser imagings
The various ways such as magnetic feature detection joint perception naval target, this system also innovative usage Laser Tomographic sensor and other biographies
Sensor carries out joint perception, solve the problems, such as Maritime Law Enforcement scouting see it is not far, do not see, realize law enforcement ship it is remote, in, it is close
Distance scouts, identification, evidence obtaining, promotes my marine right-safeguarding enforcement effort advantage and ability.
It is often not only that can identify Maritime target type in marine right-safeguarding enforcing law, it is known that be big ship, canoe, floating
Object or person person, but naval target precise information is understood with greater need for depth, including is grasped its ships and light boats specifications and models, carried force
The more information such as device, important equipment, operating status.The method of this system is then building naval target characteristic feature database, real
When perception analysis target image information be compared with characteristic target library, carry out system depth study and analyzing and training, and constantly
Strengthen target characteristic database, final precision target information that is rapid, accurately generating marine right-safeguarding law enforcement needs.
Two, the multi-Target Image enhancing based on GPUs concurrent collaborative
Since sea climate environment is changeable in real time, the complicated ring perceived to machine vision is generally required during execution task
Border image, multiple targets are handled in time, along with the target identification for carrying out deep learning needs a large amount of concurrent collaboratives to calculate,
Enforcing law needs to generate output recognition result in real time, and common process calculating is difficult to undertake these requirements.
Multi-Target Image enhancing based on GPUs concurrent collaborative, major way are first to construct target identification computation model, will
Multi-Target Image identification is mapped on more GPUs hardware, respectively executes multithreading cooperative mode parallel in every GPU, quickly
It handles Real-time image enhancement to calculate, realizes that multiple targets are handled in time, calculate, analyzed, obtain preferable scouting perception real-time
Energy is obviously improved.
This system is based on the collaboration of GPUs hardware and accelerates, quickly excellent by identification technologies such as parallel computation, Digital Image Processing
Change image defogging, denoising process, improve algorithm for image enhancement, is enhanced in image restoration, background inhibition, Objective extraction process
Optimization, can be only achieved better quality, faster the time obtain target image identification as a result, to realize at sea multiple target it is quick
Statistical error ability.
Three, evidence obtaining is scouted in the comprehensive perception of multi-source information
Electro-optical tracking device has the function of that general motion target tracking, electro-optical tracking device can be with law enforcements on law enforcement naval vessel
Evidence taking equipment carries out joint evidence obtaining.Due to electro-optical tracking device and law enforcement evidence taking equipment installation site difference, general photoelectricity with
In rear portion, law enforcement evidence taking equipment, in front, different viewing blind zones is individually present, amphitypy equipment all works in joint in track equipment
The mode of scheduling and law enforcement evidence obtaining, and the interface protocol of the information interaction of planning apparatus just can be achieved mutually that complementation is blind, combines and takes
Card ability.
The optimization that this system is identified using various information sources (such as radar, AIS, navigation system) existing on ship, needle
To the same target of multiple information sources perception, first progress time synchronization registration, multiple source sensor data acquisition is optimized accurate
Data acquire and measure the realistic objective physics measuring value of a certain synchronization time, then carry out spatial registration and target data again
It is comprehensive, efficient, accurate aid decision and commander are carried out for commander.Comprehensive multiple information sources carry out perception target and compare fusion,
Accurate target information is obtained in time, is usually faced marine complicated situation to cope with law enforcement ship, will effectively be promoted naval vessel Situation Awareness
With law enforcement reconnaissance capability.
The beneficial effect comprise that: scouted by multimode and deep learning cognition technology means, reach it is accurate,
Identification naval target characteristic information effect rapidly；It realizes that real time enhancing perceives under complex background environment, it is marine to promote law enforcement naval vessel
Accurate reconnaissance capability；Combine with multi-source naval target and scout perception, fulfillment capability, improves sea police's ship joint and scout evidence obtaining and multiple
Miscellaneous situation controls ability.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples, in attached drawing:
Fig. 1 is Maritime Law Enforcement reconnaissance system equipment composite structural diagram in the embodiment of the present invention；
Fig. 2 is the naval target intelligent identification Method of multimode joint perception and deep learning in the embodiment of the present invention；
Fig. 3 is the multi-Target Image Enhancement Method based on GPUs concurrent collaborative in the embodiment of the present invention；
Fig. 4 is that evidence collecting method is scouted in the comprehensive perception of multi-source information in the embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, not
For limiting the present invention.
A kind of Maritime Law Enforcement reconnaissance system based on machine vision joint perception, including preceding light are provided in the embodiment of the present invention
Electric awareness apparatus, rear photoelectricity awareness apparatus, Situation Awareness equipment, deep learning work station.Preceding photoelectricity awareness apparatus, rear light inductance
Know that equipment, Situation Awareness equipment, deep learning work station are connected with each other by network.As shown in Figure 1, preceding photoelectricity awareness apparatus,
Photoelectricity awareness apparatus is located at law enforcement forward quarter, Background Region, the scouting perception of the naval target for front and rear afterwards.
Preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus by visible light sensor, infrared imaging sensor, laser imaging sensor,
The various ways detecting sensor part such as laser range sensor is integrated among servo turntable.Deep learning work station is backstage mesh
The calculating center for marking intelligent recognition, for finding naval target information.Situation Awareness equipment is used for the joint that Maritime Law Enforcement is scouted
Perception and multi-source fusion calculate；Deep learning work station is set for preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus, Situation Awareness
The identification learning and comparing of standby output naval target information.
The visible light signal of conventional target perception images in photodetector target surface through electro-optical system by propagation in atmosphere, is
The complex process of one electrical image that extraterrestrial target two dimensional image is changed into time series.And single visible light sensor sense
During knowing, target photoelectric image by marine atmosphere, light environment interference is big, noise is more, easy to be lost.Conventional target perception away from
From it is close, do not see, distinguish unknown, how farther, more preferable, more quasi- perception naval target true picture is the difficulty that law enforcement ship scouts perception
Point.
This system integrates a variety of photoelectricity perceptive modes such as visible light, infrared light, laser imaging, sets in the photoelectricity perception of Fig. 4
In standby each perception device, it is seen that optical sensor perceives marine medium-sized ship under fogless meteorological condition on daytime；Infrared imaging sensing
Device is used in night, the visible lights unfavorable conditions perception imaging such as have mist or closely perceives；It is imaged and is sensed using laser (chromatography)
Device perceived distance naval target and can then be imaged farther out in the case where there is the unfavorable conditions such as the visible lights such as mist, infrared；Distance measuring sensor is then
For measuring my ship and target line distance.System can also combine the various ways perception such as electronic reconnaissance, electromagnetic signature detection
Naval target, realize law enforcement ship it is remote, in, close reconnaissance perception, promote my marine right-safeguarding enforcement effort advantage and ability.
As shown in Fig. 2, law enforcement ship carries out during right-safeguarding enforcement effort, visible light, infrared imaging, laser imaging will be passed through
Etc. multimodes imaging perception is carried out to naval target, can also the technological means such as integrated electronics scouting, electromagnetic signature assist in identifying sea
Upper target, the target image obtained in real time pass through image preprocessing and image enhancement, and at the beginning of extracting target signature and image
Classification and Identification, then the analysis of compare feature naval vessel library are walked, and deep learning is carried out based on model ship library, in conjunction with radar, AIS, is led
The multiple source interfaces such as boat system strengthen target information confirmation, final real-time perfoming target classification label and precision target identification, complete
The scouting of naval target perceives.
It is often not only that can identify Maritime target type in marine right-safeguarding enforcing law, it is known that be big ship, canoe, floating
Object or person person, but naval target precise information is understood with greater need for depth, including is grasped its ships and light boats specifications and models, carried force
The more information such as device, important equipment, operating status.Deep learning is based on the typical convolutional neural networks of artificial intelligence
(Convolutional Neural Network, CNN) and depth convolutional neural networks (DCNN), construct multi-level neural network
Model is made of multiple layers such as convolutional layer, pond layer and Quan Lian stratum.Pair typical CNN/DCNN technology is used for picture classification, i.e.,
Input picture can only mostly determine the classification of picture.This system needs more accurate and clearer master goal information, and system is first
Naval target property data base is constructed, extraction and analysis target image information is compared with characteristic target library, carries out system depth
Study and analyzing and training, continually strengthen target characteristic database, it is final rapidly, accurately generate marine right-safeguarding law enforcement need it is accurate
Target information.
Detection and identification in video streaming are realized to naval target based on current depth study, marked in video streaming in real time
Infuse target information, including target category and target position.Enhance learning art by depth, it is big using computer real-time analysis
Data are measured, the automation of target detection analysis is realized, provides foundation for Maritime Law Enforcement action；Meanwhile for emphasis observed object
It is lasting to obtain, Various types of data is accumulated, entirety observation, analysis and interpretation to target is converted into, data is transformed into relevant valence
The target intelligence of value.
Since sea climate environment is changeable in real time, the complicated ring perceived to machine vision is generally required during execution task
Border image, multiple targets are handled in time, along with the target identification for carrying out deep learning needs a large amount of concurrent collaboratives to calculate,
Enforcing law needs to generate output recognition result in real time, and common process calculating is difficult to undertake these requirements.
Multi-Target Image enhancing based on GPUs concurrent collaborative, major way are first to construct target identification computation model, i.e.,
Perception-processing-calculating-analysis model.On-board sensor will perceive marine complex environment background image, be tracked, identified, be mentioned
Multiple target image targets 1 are taken, multiple worker threads are respectively created in target 2 ... target n in the computation processor course of work,
Respectively execute the cooperated computing in each thread in each image processor GPUs (GPU1, GPU2 ..., GPUn), quickly into
It is multiple as a result, simultaneously further progress sea Study on Trend to generate output on a workstation for the parallel computation of row scan picture
And subject fusion, the scouting sense of marine multiple target is realized during intelligent recognition, enhancing processing, parallel computation, Study on Trend
Know, as shown in Figure 3.
This system is based on the collaboration of GPUs hardware and accelerates, quickly excellent by identification technologies such as parallel computation, Digital Image Processing
Change image defogging, denoising process, improve algorithm for image enhancement, is enhanced in image restoration, background inhibition, Objective extraction process
Optimization, can be only achieved better quality, faster the time obtain target image identification as a result, to realize at sea multiple target it is quick
Statistical error ability.
General electro-optical tracking device has the function of photoelectric tracking on law enforcement naval vessel, and electro-optical tracking device can collect evidence with law enforcement
Equipment carries out joint evidence obtaining.Due to the difference of electro-optical tracking device and law enforcement evidence taking equipment installation site, general photoelectric tracking is set
, in front, different viewing blind zones is individually present, amphitypy equipment all works in combined dispatching in rear portion, law enforcement evidence taking equipment
With the mode of law enforcement evidence obtaining, and the interface protocol of the information interaction of planning apparatus just can be achieved mutually complementary blind, joint and collect evidence energy
Power.
Law enforcement ship photoelectricity perceive integration of equipments visual TV sensor, infrared imaging sensor, laser imaging sensor,
The devices such as distance measuring sensor supply power supply to each device by device power source, Laser Power Devices.Photoelectricity evidence taking equipment passes through angle measurement system
System, servo-system, photoelectricity awareness apparatus, photoelectricity evidence taking equipment form joint and scout law enforcement, and deep learning work station is by analysis
Calculating provides photoelectricity awareness apparatus, photoelectricity evidence taking equipment control instruction is moved for equipment servo.It is connect simultaneously by interface equipment
The naval target information of multi-source equipment (radar, AIS, navigation system) perception outside receiving, and letter is carried out through Situation Awareness equipment
Breath compares and data fusion, outputs results on display screen.
It should be understood that for those of ordinary skills, it can be modified or changed according to the above description,
And all these modifications and variations should all belong to the protection domain of appended claims of the present invention.
Claims (7)
1. a kind of Maritime Law Enforcement reconnaissance system based on machine vision joint perception, which is characterized in that set including the perception of preceding photoelectricity
Standby, rear photoelectricity awareness apparatus, Situation Awareness equipment, deep learning work station, preceding photoelectricity awareness apparatus, rear photoelectricity awareness apparatus point
Wei Yu not enforce the law forward quarter, Background Region, and the naval target for front and rear scouts perception, and Situation Awareness equipment is for sea
The joint perception and multi-source fusion that upper law enforcement is scouted calculate；Deep learning work station is used for preceding photoelectricity awareness apparatus, rear light inductance
Know the identification learning and comparing of equipment, Situation Awareness equipment output naval target information.
2. the Maritime Law Enforcement reconnaissance system according to claim 1 based on machine vision joint perception, which is characterized in that preceding
Photoelectricity awareness apparatus, rear photoelectricity awareness apparatus by visible light sensor, infrared imaging sensor, laser imaging sensor, swash
Ligh-ranging sensor integration is among servo turntable.
3. the Maritime Law Enforcement reconnaissance system according to claim 1 or 2 based on machine vision joint perception, feature exist
In the target image obtained in real time is passed through image preprocessing and image enhancement by deep learning work station, and extracts target
Feature and image preliminary classification identification, compare feature naval vessel library analysis, and based on model ship library carry out deep learning, in real time into
Row target classification label and target identification.
4. the Maritime Law Enforcement reconnaissance system according to claim 3 based on machine vision joint perception, which is characterized in that deep
It spends study and work station and strengthens target information confirmation herein in connection with radar, AIS, navigation system.
5. the Maritime Law Enforcement reconnaissance system according to claim 3 based on machine vision joint perception, which is characterized in that deep
It spends study and work station and is based on convolutional neural networks and depth convolutional neural networks, construct multi-level neural network model.
6. the Maritime Law Enforcement reconnaissance system according to claim 3 based on machine vision joint perception, which is characterized in that base
The image enhancement of multiple target is carried out in GPUs concurrent collaborative.
7. the Maritime Law Enforcement reconnaissance system according to claim 6 based on machine vision joint perception, which is characterized in that ship
Set sensor will perceive maritime environment background image, be tracked, identified, extract multiple target image targets 1, target 2 ... mesh
N is marked, multiple worker threads are respectively created in the deep learning work station computation processor course of work, in each image processor
GPUs (GPU1, GPU2 ..., GPUn) respectively executes the cooperated computing in each thread, carries out the parallel of scan picture
It calculates, it is multiple as a result, and carrying out marine Study on Trend and subject fusion that output is generated on deep learning work station.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910399849.3A CN110210527A (en) | 2019-05-14 | 2019-05-14 | Maritime Law Enforcement reconnaissance system based on machine vision joint perception |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910399849.3A CN110210527A (en) | 2019-05-14 | 2019-05-14 | Maritime Law Enforcement reconnaissance system based on machine vision joint perception |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110210527A true CN110210527A (en) | 2019-09-06 |
Family
ID=67785883
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910399849.3A Pending CN110210527A (en) | 2019-05-14 | 2019-05-14 | Maritime Law Enforcement reconnaissance system based on machine vision joint perception |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110210527A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111338382A (en) * | 2020-04-15 | 2020-06-26 | 北京航空航天大学 | Unmanned aerial vehicle path planning method guided by safety situation |
| CN111624589A (en) * | 2020-04-27 | 2020-09-04 | 中国人民解放军军事科学院国防科技创新研究院 | Marine target data fusion system and method based on space-based radio monitoring |
| CN112308883A (en) * | 2020-11-26 | 2021-02-02 | 哈尔滨工程大学 | Multi-ship fusion tracking method based on visible light and infrared images |
| CN113484864A (en) * | 2021-07-05 | 2021-10-08 | 中国人民解放军国防科技大学 | Unmanned ship-oriented navigation radar and photoelectric pod collaborative environment sensing method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105630882A (en) * | 2015-12-18 | 2016-06-01 | 哈尔滨工业大学深圳研究生院 | Remote sensing data deep learning based offshore pollutant identifying and tracking method |
| CN107659614A (en) * | 2017-08-28 | 2018-02-02 | 安徽四创电子股份有限公司 | A kind of more base station type waters surveillance control systems and its monitoring control method |
| US20180060808A1 (en) * | 2008-06-27 | 2018-03-01 | Cargometrics Technologies, Llc | System and method for generating commodity flow information |
| CN108416361A (en) * | 2018-01-18 | 2018-08-17 | 上海鹰觉科技有限公司 | A kind of information fusion system and method based on sea survaillance |
| CN109709589A (en) * | 2019-01-09 | 2019-05-03 | 深圳市芯鹏智能信息有限公司 | A kind of air-sea region solid perceives prevention and control system |
-
2019
- 2019-05-14 CN CN201910399849.3A patent/CN110210527A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180060808A1 (en) * | 2008-06-27 | 2018-03-01 | Cargometrics Technologies, Llc | System and method for generating commodity flow information |
| CN105630882A (en) * | 2015-12-18 | 2016-06-01 | 哈尔滨工业大学深圳研究生院 | Remote sensing data deep learning based offshore pollutant identifying and tracking method |
| CN107659614A (en) * | 2017-08-28 | 2018-02-02 | 安徽四创电子股份有限公司 | A kind of more base station type waters surveillance control systems and its monitoring control method |
| CN108416361A (en) * | 2018-01-18 | 2018-08-17 | 上海鹰觉科技有限公司 | A kind of information fusion system and method based on sea survaillance |
| CN109709589A (en) * | 2019-01-09 | 2019-05-03 | 深圳市芯鹏智能信息有限公司 | A kind of air-sea region solid perceives prevention and control system |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111338382A (en) * | 2020-04-15 | 2020-06-26 | 北京航空航天大学 | Unmanned aerial vehicle path planning method guided by safety situation |
| CN111338382B (en) * | 2020-04-15 | 2021-04-06 | 北京航空航天大学 | Unmanned aerial vehicle path planning method guided by safety situation |
| CN111624589A (en) * | 2020-04-27 | 2020-09-04 | 中国人民解放军军事科学院国防科技创新研究院 | Marine target data fusion system and method based on space-based radio monitoring |
| CN111624589B (en) * | 2020-04-27 | 2021-11-02 | 中国人民解放军军事科学院国防科技创新研究院 | Marine target data fusion system and method based on space-based radio monitoring |
| CN112308883A (en) * | 2020-11-26 | 2021-02-02 | 哈尔滨工程大学 | Multi-ship fusion tracking method based on visible light and infrared images |
| CN113484864A (en) * | 2021-07-05 | 2021-10-08 | 中国人民解放军国防科技大学 | Unmanned ship-oriented navigation radar and photoelectric pod collaborative environment sensing method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110210527A (en) | Maritime Law Enforcement reconnaissance system based on machine vision joint perception | |
| CN105373135B (en) | A kind of method and system of aircraft docking guidance and plane type recognition based on machine vision | |
| CN106845502B (en) | Wearable auxiliary device for equipment maintenance and visual equipment maintenance guiding method | |
| CN108731587A (en) | A kind of the unmanned plane dynamic target tracking and localization method of view-based access control model | |
| CN108303096B (en) | Vision-assisted laser positioning system and method | |
| CN105550670A (en) | Target object dynamic tracking and measurement positioning method | |
| CN109934108B (en) | Multi-target and multi-type vehicle detection and distance measurement system and implementation method | |
| CN105306892B (en) | A kind of generation of ship video of chain of evidence form and display methods | |
| Lipschutz et al. | New methods for horizon line detection in infrared and visible sea images | |
| CN103714321A (en) | Driver face locating system based on distance image and strength image | |
| CN104166137A (en) | Target comprehensive parameter tracking measurement method based on display of radar warning situation map | |
| CN104517095A (en) | Head division method based on depth image | |
| CN102789578A (en) | Infrared remote sensing image change detection method based on multi-source target characteristic support | |
| CN108469817A (en) | The unmanned boat obstruction-avoiding control system merged based on FPGA and information | |
| CN109697428B (en) | Unmanned aerial vehicle identification and positioning system based on RGB _ D and depth convolution network | |
| CN112597877A (en) | Factory personnel abnormal behavior detection method based on deep learning | |
| CN108182704A (en) | Localization method based on Shape context feature | |
| CN105447431B (en) | A kind of docking aircraft method for tracking and positioning and system based on machine vision | |
| CN116862832A (en) | Three-dimensional live-action model-based operator positioning method | |
| CN112560922A (en) | Vision-based foggy-day airplane autonomous landing method and system | |
| Lu et al. | Detecting small target of ship at sea by infrared image | |
| Jian et al. | Maritime target detection and tracking | |
| CN109543553A (en) | The photoelectricity recognition and tracking method of low small slow target based on machine learning | |
| CN114092522A (en) | Intelligent capture tracking method for take-off and landing of airport airplane | |
| Wang et al. | Target detection for construction machinery based on deep learning and multisource data fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |